The Economist Asia Edition – July 27, 2019

→InAmerica,recessionstypicallyfollowaftertheyieldcurvehasinverted

→Atfirstglance,therelationshipappearsmurkierinothercountries

*NationalBureauofEconomicResearchdefinition Sources:DatastreamfromRefinitiv;HaverAnalytics;IMF;OECD;TheEconomist

0

2

-2

Quarterlyaverage 4

1960 1970 1980 1990 2000 2010 2019

-10123

Theyieldcurveshowstherateofinterestatwhich

governmentsborrowmoneyoverdifferenttime

periods.Typicallyit costsmoretoborrowforlonger.

Wheninterestratesarehigherintheshortterm

thanthelongterm,thecurveisinverted.

Recessions*

Differenceininterestratesbetweenten-year

bondsandthree-monthbills,percentagepoints

Canada1974-1993

Recessions

Sweden1980-99 Britain1994-2013

UnitedStates1960-2019

Q1 1994 Q2 2019

Germany1996-2015

↑Interestrate

Invertedfor

14 daysin 1989

Time→

3-mth 10-yr 3-mth 10-yr 3-mth 10-yr

2.35%

-0.02%↗

3.33%

2.33%

6.09%

Inverted

Normal

GDP,%changeonpreviousquarter

↓Invertedyieldcurves

Q2 2019 wasthefirst

quartersincethefinancial

crisisof 2008 inwhich

America’syieldcurve

wasinvertedonaverage.

-4

0

4

2

-2

-4

0

4

2

-2

1974 80 85 90 93 1980 85 90 95 99 1994 2000 05 10 13 1996 2000 05 10 15

↗

TheEconomistJuly 27th 2019 69

M

any economistssee the link between

gdpgrowth and yield curves as a curi-

ous case of American exceptionalism. In

general, interest rates rise as borrowing pe-

riods get longer, because the risks of de-

fault and rising inflation grow over time.

But occasionally this pattern reverses, and

short-term rates exceed long-term ones.

In America, such “inversions” have

foreshadowed economic turmoil. For all

eight recessions since 1960, three-month

interest rates exceeded ten-year ones on at

least one day during the previous year. The

signal has sounded just one false alarm.

There are good reasons why yield-curve

inversions tend to precede recessions. At

the short end, when central banks raise

rates, the curve flattens and the economy

slows. On the long side, when a recession

looms, investors expect that central banks

will cut rates to soften the blow. That low-

ers long-term yields, flattening the curve.

This logic should apply everywhere. Yet

only in America has the curve been a sooth-

sayer. In a dataset of 16 other rich countries,

reaching as far back as 1960, 51 of the 95 re-

cessions were not preceded by an inversion

during the previous two years. Moreover,

the curve seems prone to crying wolf. On 63

occasions, these non-American economies

kept growing despite inverted yield curves.

The yield curve’s failure to foresee re-

cessions outside the United States has led

some scholars to dismiss its predictive

power as a fluke. With so few recessions in

America, there is insufficient evidence to

determine the strength of the relationship.

However, squashing yield curves and

growth figures into a pair of binaries—in-

verted or not, and recession or not—leaves

precious data on the cutting-room floor. A

better test would check whether flattening

curves foreshadow slowdowns, and steep-

ening ones presage economic acceleration.

Seen through this lens, America is not

an outlier. In 15 of 17 countries, changes in

spreads correlated with changes in growth

the next year. Overall, a one-percentage-

point move in spreads predicted a 0.55-

point change in growth in the same direc-

tion. The effect was strongest in Switzer-

land, at 1.1 points; America ranked third.

Economists do not appear to make full

use of this well-known indicator. If they

did, blending their predictions with yield-

curve data would be no more accurate than

using consensus projections alone. How-

ever, we found that consensus forecasts

made a year in advance accounted for 57%

of variance in gdp. In contrast, the blend

explained 64%—a large improvement.

Changes in monetary-policy tools

mean that the curve may lose some of its

predictive power in future. Because central

banks have bought long-dated bonds in

quantitative-easing schemes, they now af-

fect both sides of the yield curve directly.

That makes long-term interest rates a less

reliable proxy for market expectations.

But if history is any guide, America

shouldexpect a deceleration. Its curve has

flattened by 1.1 points in the past year, im-

plyinggrowth will slow from 3% to 2%. 7

Yield curves help predict gdpgrowth

across the rich world

Curveball

Graphic detailEconomic forecasting

Inversions are bad omens everywhere

GDP, %changeona yearearlier,averageof

17 OECDcountries,1960-2019,weightedbyGDP

042618102

Quarters

0

1

2

3

Following normal

yield curve

Following yield-

curve inversion