You are on page 1of 38

International Transmission of Japanese Monetary Shocks

The views expressed in this presentation are the views of the author and do not necessarily reflect the views or policies of the Asian Development Bank Institute (ADBI), the Asian Development Bank (ADB), its Board
of Directors, or the governments they represent. ADBI does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use. Terminology used may not
necessarily be consistent with ADB official terms.

International Transmission of Japanese Monetary


Shocks Under Low and Negative interest Rates:
A FAVAR Approach1

Mark M. Spiegel Andrew Tai

FRB San Francisco

ADBI Conference
Implications of Negative Interest Rates for Emerging Asia,
Tokyo Japan, December 1-2, 2016
1
The views expressed herein are those of the authors and do not necessarily
reflect the views of the Federal Reserve Bank of San Francisco or the Federal
Reserve System.
1 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Examine implications of recent shocks to Japanese


monetary policies

I Assessing impact in negative interest rate environment


challenging
I Recent BOJ policies both conventional and unconventional
I 0 percent target for its 10-year bond yield
I Committed to overshooting two-percent inflation target
I Implications of policies opaque around the zero lower bound
I Movements of short rates into negative nominal rate territory
may be limited, and not fully reflect impact of commitments
on agents expectations

2 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Follow literature in evaluating movements in longer-term


assets
I Swanson and Williams (2014): examine movements in 2-year
U.S. Treasury yields
I We look at shocks to 2-year JGBs
I Nominal movements in Japanese 2-year rate are more muted
than US, but larger than policy rate
I 2-year JGB yield recently negative, providing increased stimulus
I Expansionary monetary policy at or near the zero lower
nominal bound achieved through unconventional policies:
I Long-term asset purchases or policies that change expectations
I Assess impact of shocks on Japan economy, and three trading
partners: Korea, China and the United States

3 / 38
International Transmission of Japanese Monetary Shocks
Introduction

2 year JGBs move less than US, but more than policy rates
8
6
4
2
0

1998 2000 2003 2006 2009 2012 2015

US 2y Japan 2y BoJ policy rate

4 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Large VAR literature finds on global U.S. policy spillovers


[e.g. CEE (1996)]
I Kim (2001): spillovers of U.S. monetary policy on G6
I Canova (2005): spillovers from U.S. policy to Latin America
I Strong evidence of U.S. spillovers reflects unique role of U.S.
economy
I In contrast, Mackowiak (2006) looks for spillovers from
Japanese monetary policy shocks
I Small Asian countries: Hong Kong, Korea, Malaysia,
Philippines, Singapore and Thailand
I Finds only modest impact of Japanese policy on Asian
neighbors
I Suggests beggar-thy-neighbor effect of expansionary
Japanese monetary policy modest
5 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Reexamine Japanese monetary shocks for more recent


1998-2015 period
I Mackowiak (2006) sample ends in 2002 conventional policy
rates as shocks
I Use 2-year JGB innovations instead
I Use Global FAVAR of Liu, Tai, Spiegel (2016)
I Extends Bernanke and Boivin [BB (2003)] FAVAR to allow for
separate estimation of external conditions as a latent variable
I FAVAR designed to assess movements in concepts, such as
economic activity that are not directly observable
I FAVAR approach mixes alternative measures of concepts
through the use of factor analysis
I BB (2003): Estimated latent activity factors likely to be more
accurate measures than any individual series
I Also eliminate need for ad hoc specification decisions
6 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Global FAVAR of LST (2017) adds latent index of


foreign activity

I Foreign activity index weighted by trade volumes for a


countrys prominent trading partners
I Yields a country-specific global conditions index
I Useful for conditioning for changes in relevant foreign activity
in other trading partner countries
I By including global conditions index, the Global FAVAR
isolates impacts of the Japanese shocks

7 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Base specification
I Base specification includes the two latent activity factors
I Local conditions index (LCI): reflects domestic economic
activity proxied by the first principal component of nine
indicators of domestic activity
I Global conditions index (GCI): Controls for effects of external
conditions faced by each country
I Each countrys GCI estimated from 11 monthly times series
for the top 9 individual trading partners of each country
I Weighted by the trade shares of each country
I Yields a trade-weighted series for each measured indicator
I Estimate first principal component of these 11 trade-weighted
indicators along with 4 common series capturing global
financial conditions to construct GCI
I 2 latent factors included in 6 variable VAR
8 / 38
International Transmission of Japanese Monetary Shocks
Introduction

Results

I Positive shock to 2-year Japanese JGB rate associated with


reduced economic activity and inflation in Japan
I Also deterioration in external conditions faced by Japanese
economy
I Results for 3 trading partners suggest only modest and
insignificant impacts
I Overall, appears that primary implications of Japanese
monetary policy shocks are internal
I Baseline FAVAR results confirmed through forecast error
variance decompositions and other robustness tests

9 / 38
International Transmission of Japanese Monetary Shocks
Empirics

Local conditions index (LCI)


I Estimate Local conditions index as in LST (2016)
I Generated as first principal component of domestic economic
indicators
I Denoting indicators by Xt and factor by Ft , observables are
related to the factors through

Xt = Ft + ut , (1)
where ut denotes idiosyncratic noise and is matrix of factor
loadings
I Domestic variables include industrial production,
unemployment, housing starts or permits, stock prices, 1-year
government bond yields, 2-year government bond yields, M1,
M2, and PPI
10 / 38
International Transmission of Japanese Monetary Shocks
Empirics

Global conditions index (GCI)

I GCI also estimated as a latent variable as in LST (2016)


I 11 time series for the top 9 trading partners
I Indicator variables include IP, unemployment, CPI, housing
starts or permits, equity prices, 1-year govt bonds, 2-year govt
bonds, 3-month govt bonds, M1, M2, and PPI
I Compute principal components of trade-weighted averages of
11 time series plus 4 common series: oil prices, EMBI index,
euro zone sentiment, and EU sentiment
I GCI is measured as the first principal component

11 / 38
International Transmission of Japanese Monetary Shocks
Empirics

Japan Local and Global Conditions Indices 1998-2016

8
BoJ adopts

6
inflation targeting

4
2
0
-2
-4
-6
LCI GCI

-8
1997 2000 2003 2006 2009 2012 2015

Figure : Japan Local and Global Conditions Indices 1998-2016. 12 / 38


International Transmission of Japanese Monetary Shocks
Empirics

Japanese LCI and GCI both appear to be plausible

I LCI and GCI closely correlated


I Steep drop in the Japanese GCI during the great recession,
while LCI falloff surprisingly tranquil relative
I Both LCI and GCI pick up around the middle of 2013
I Election of Shinzo Abe as Prime Minister and launch of
Abenomics reforms
I Substantive pickup after the formal adoption of inflation
targeting by the Bank of Japan in 2013
I Pickups almost completely reversed by the end of our sample

13 / 38
International Transmission of Japanese Monetary Shocks
Japan FAVAR model

Japan FAVAR model

I Specification includes 2-yr US T-bill rate, LCI, CPI inflation,


GCI, the broad trade-weighted Japanese yen, and the Japan
2-yr JGB rate, in that order
I Ordering reflects Choleski identification restriction that latent
factors (LCI and GCI) and observables (inflation and the
interest rate) do not respond to shock in 2-yr JGB rate in
impact period, while 2-yr JGB rate allowed to respond to the
other shocks
I Estimate the impulse responses using Bayesian methods with
Sims-Zha (1998) priors

14 / 38
International Transmission of Japanese Monetary Shocks
Japan FAVAR model

Japan IRFs
.68 Error Bands
0.05 0.05

Local conditions
0 0
US 2y

-0.05 -0.05

-0.1 -0.1

-0.15 -0.15

-0.2

0.05 0.1

Global conditions
0 0
Inflation

-0.05 -0.1

-0.1 -0.2

-0.15

0.04
0.02 0.03
0.02
Broad JPY

0.015
JPN 2y

0.01
0.01
0
0.005 -0.01

0 -0.02

6 12 18 24 36
Months

15 / 38
International Transmission of Japanese Monetary Shocks
Japan FAVAR model

Japan impact of shock to 2-year JGB rate

I Positive shock to Japanese 2-year rate leads to expected set of


responses
I The rise in Japanese rates is followed by a decline over time to
both Japanese domestic activity as measured by the LCI and
inflation
I Also see significant decline in Japans global conditions index
I Persistent increase in the value of the yen
I VAR results suggest that shock to Japanese 2-year rate moves
aggregate Japanese variables in expected manner and has
significant effects, even in environment of low and negative
prevailing short-term rates

16 / 38
International Transmission of Japanese Monetary Shocks
Japan FAVAR model

Forecast Error Variance: Japanese Activity from Japan 2-yr

Table : Forecast Error Variance: Japanese Activity from Japan 2-yr


Months LCI Inflation GCI Broad Japan yen Japan 2 yr
1 0.00 0.00 0.00 0.00 42.16
12 0.34 0.93 2.02 86.54 17.71
24 8.04 6.08 7.74 77.89 12.35
36 13.84 15.57 9.57 70.67 13.18

17 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

Korea FAVAR model

I Korea model includes US 2-year rate, Koreas LCI, CPI


inflation rate, GCI, nominal broad won index, and the
Japanese 2-year rate, in that order
I Ordering again reflects Choleski restriction: LCI, GCI, and
observables do not respond to 2-year JGB rate in impact
period, while 2-year JGB rate allowed to respond
I Only change is substitution of Korean real exchange rate

18 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

Korea Impulse Response Functions


.68 Error Bands

0.05

Local conditions
0.05
0
US 2y

0
-0.05

-0.1 -0.05
-0.15

0.1

Global conditions
0.1
Inflation

0.05
0

0 -0.1

-0.05 -0.2

0
0.04
Real exchange rate

-0.005
0.03
JPN 2y

0.02
-0.01
0.01
-0.015 0
-0.01
-0.02

6 12 18 24 36
Months

19 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

Korea impact of shock to 2-year JGBs

I Movements in Korean variables very muted


I Only modest movements in economic activity and inflation
I Neither statistically significant at even a 68% confidence level
I Also modest response in Korea GCI
I Significantly positive at peak around 6 months after shock, but
point estimates small
I Do see depreciation in Korean won, as expected
I Probably behind the modest increase in Korean GCI

20 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

China FAVAR model

I FAVAR specification particularly useful for China, where


quality of data is in question
I China FAVAR model includes US 2-year rate, China LCI, CPI
inflation rate, GCI, nominal broad renminbi index, and the
Japanese 2-year rate, in that order
I Ordering again reflects Choleski restriction: LCI, GCI, and
observables do not respond to 2-year JGB rate in impact
period, while 2-year JGB rate allowed to respond
I Only change is substitution of Chinese real exchange rate

21 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

China Impulse Response Functions


.68 Error Bands
0.05 0.05

Local conditions
0
0
US 2y

-0.05

-0.1 -0.05

-0.15
-0.1

0.1
0.15

Global conditions
0.1 0
Inflation

0.05
-0.1
0

-0.05 -0.2

-0.1

#10 -3
0.04
8
Real exchange rate

0.03
6 0.02
JPN 2y

4 0.01
2 0

0 -0.01
-0.02
-2
6 12 18 24 36
Months

22 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

China impact of shock to 2-year JGBs

I Qualitative results for China similar to Korea


I Modest decline in Chinese activity and a modest increase in
inflation
I Usually insignificant at 68% confidence level
I Even smaller impact on Chinese GCI than observed for Korea
I Probably attributable to Chinese management of exchange rate
I Broad renminbi rate largely unresponsive to Japanese yen
I General patterns similar to those for Korea, but less persistent

23 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

US FAVAR model

I US FAVAR model includes US 2-year rate, US LCI, CPI


inflation rate, GCI, nominal broad renminbi index, and the
Japanese 2-year rate, in that order
I Ordering again reflects Choleski restriction: LCI, GCI, and
observables do not respond to 2-year JGB rate in impact
period, while 2-year JGB rate allowed to respond
I Only change is substitution of US dollar real exchange rate

24 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

US Impulse Response Functions


.68 Error Bands
0.05

Local conditions
0
0
US 2y

-0.05 -0.05

-0.1
-0.1
-0.15

0.1
0.05

Global conditions
0.05
0
Inflation

0
-0.05
-0.05
-0.1 -0.1
-0.15 -0.15

#10 -3
0.05
Real exchange rate

4
0.04
2
JPN 2y

0.03
0
0.02
-2
0.01
-4 0
-6 -0.01
6 12 18 24 36
Months

25 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

US impact of shock to 2-year JGBs

I Estimated responses for the US local conditions index even


more modest than Korea and China
I Almost no movement in LCI
I Do observe a modest decline in US inflation rate
I Would follow from the exchange rate appreciation in Japan
feeding into the price of US imports from that country
I However, US exchange rate, surprisingly appreciates
I Inflation results puzzling

26 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

Forecast Error Variance: Effects of US and Japan 2-yr

Table : Forecast Error Variance: Effects of US and Japan 2-yr

Global LCIs
Impact of JPN 2y Impact of US 2y
Months Japan Korea China United States Japan Korea China United Sta
1 0.00 0.00 0.00 0.00 5.53 4.37 0.25 9.98
12 0.34 1.52 2.24 0.68 22.78 4.16 6.18 14.15
24 8.04 1.58 3.94 0.61 33.11 8.65 4.42 23.37
36 13.84 1.64 5.53 3.23 33.83 7.22 4.42 21.12
Inflation
Impact of JPN 2y Impact of US 2y
Months Japan Korea China United States Japan Korea China United Sta
1 0.00 0.00 0.00 0.00 0.05 0.02 1.95 0.13
12 0.93 0.48 3.23 6.22 6.79 7.10 2.83 1.66
24 6.08 4.83 5.26 4.78 6.40 8.63 8.87 4.51
36 15.57 7.43 4.79 4.88 6.00 11.83 8.07 6.65

27 / 38
International Transmission of Japanese Monetary Shocks
FAVAR models for other countries

FEVs: 2-yr JGB shocks impact Japanese conditions, but


modest impact on trading partners
I US shocks dominate for US and other trading partners in
terms of variability of LCIs
I Surprisingly, US 2-year shocks even dominant for Japanese
economy itself!
I With exception of US, obtain same results for inflation
I US shocks dominant for Japan, China, and Korea inflation
relative to shocks to 2-year Japanese JGBs
I However, JGB shocks dominant for US inflation
I This result puzzling
I However, overall results confirm impact of JGB 2-year shocks
modest relative to impact of US 2-year shocks

28 / 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Alternative Japanese LCI specification

I Base LCI specification includes one and two-year Japanese


government bond yields
I As introduced only in principal component, no prima facie
reason to expect that inclusion raises spurious correlation
between our estimated Japanese LCI and 2-year rate shocks
I To make sure, reestimate our LCI with these yields excluded

29 / 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Japan IRFs with LCIs excluding interest rates


.68 Error Bands

0 0.05

Local conditions
-0.05
US 2y

-0.1
-0.05
-0.15
-0.1
-0.2

0.06 0.05

Global conditions
0.04 0
0.02
Inflation

-0.05
0 -0.1
-0.02
-0.15
-0.04
-0.2
-0.06
-0.25

0.025
0.05
0.02 0.04
Broad JPY

JPN 2y

0.03
0.015
0.02
0.01
0.01
0.005 0
-0.01
0
6 12 18 24 36
Months

30 / 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Results are similar, but wider confidence intervals

I Continue to find that positive shock to Japanese 2-year rate


followed by decline in both Japanese LCI and Japanese
inflation
I Also continue to see a decline in Japans global conditions
index, as well as persistent increase in the value of the yen
I However, alternative specification displays much wider
confidence bands, resulting in statistical insignificance at a
68% confidence level for most variables.

31 / 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Forecast Error Variance: Effects US and Japan 2-year with


alternative LCI

Table : Forecast Error Variance: Effects US and Japan 2-year with


alternative LCI
Impact of US 2y Impact of JPN 2y
Months Local conditions Inflation Local conditions Inflation
1 4.45 0.15 0.00 0.00
12 18.94 5.90 0.55 0.46
24 19.85 5.83 0.49 1.64
36 18.56 5.12 0.52 2.65

32 / 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Alternative VAR specification: Exchange rate last

I Follow Mackowiak (2006), consider alternative VAR ordering


with exchange rate last and 2-year JGB rate next-to-last
I Meant to reflect CEE (1999) claim that monetary authorities
do not possess complete information, and therefore respond to
some variables only with a lag
I Implies that the Japanese 2-year rate can only respond to
shocks to the Japanese exchange rate after a one-period lag.

33 / 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Impulse Response Functions, Exchange Rate Ordered Last


Local conditions Inflation
0.05 0.05
United States

0
0
-0.05
-0.05
-0.1
-0.1
-0.15
-0.15

0.05
0.05
0
0
Japan

-0.05
-0.05
-0.1
-0.1
-0.15

-0.15

0.1

0.05
0.05
Korea

0
-0.05

-0.05

0.05
0.15

0.1
0
China

0.05

0
-0.05
-0.05

-0.1 -0.1
6 12 18 24 36
34 / 38
Months
International Transmission of Japanese Monetary Shocks
Robustness checks

Forecast Error Variance: Effects US and Japan 2-year with


alternative LCI
Table : Forecast Error Variance: Effects of US and Japan 2-yr with
exchange rates last

Global LCIs
Impact of JPN 2y Impact of US 2y
Months Japan Korea China United States Japan Korea China United Sta
1 0.00 0.00 0.00 0.00 5.53 4.37 0.25 9.98
12 0.36 3.05 2.28 0.30 22.78 4.16 6.18 14.15
24 7.58 3.07 4.04 0.58 33.11 8.65 4.42 23.37
36 12.95 2.87 5.80 4.07 33.83 7.22 4.42 21.12
Inflation
Impact of JPN 2y Impact of US 2y
Months Japan Korea China United States Japan Korea China United Sta
1 0.00 0.00 0.00 0.00 0.05 0.02 1.95 0.13
12 0.75 2.82 3.16 4.12 6.79 7.10 2.83 1.66
24 6.16 2.65 5.33 3.18 6.40 8.63 8.87 4.51
36 15.61 3.92 4.86 3.42 6.00 11.83 8.07 356.65
/ 38
International Transmission of Japanese Monetary Shocks
Robustness checks

Overall results are similar

I Continue to find a lagged negative and statistically significant


response in both Japanese LCI and inflation
I Responses of the other countries modest and largely
insignificant
I For forecast error variance analyses, U.S. 2-year shocks
continue to dominate activity variables, but inflation results
remain more mixed

36 / 38
International Transmission of Japanese Monetary Shocks
Conclusion

Conclusion
I Evaluate implications of Japanese monetary policy shock at
low or negative short-term rates as proxied by a shock to the
Japanese 2-year rate
I Examine dynamics of responses to 2-year JGB rate shock
using a FAVAR framework in which both local and global
conditions are proxied by latent factors
I Results indicate that shocks to the Japanese 2-year
government bond rate affect the Japanese economy in
predicted manner, but only modest impact on other countries
I Casts doubt on suggestion that efforts by Japan to stimulate
its economy through easy monetary policy will have beggar
thy neighbor implications for trading partners
I Confirms earlier studies, conducted over much more normal
periods with more standard VAR methodologies
37 / 38
International Transmission of Japanese Monetary Shocks
Conclusion

Caveats
I 2-year Japanese government bond yield not directly-controlled
policy instrument of the Bank of Japan
I Do not directly speak to policies available at low or negative
short-term rates
I Can only draw conclusions about impact of movements in
2-year rates from policies, which are still unclear
I Forecast error variance decompositions based on the
conditions prevailing over our recent sample period
I Observed shocks to the 2-year rate over this period modest,
contributing to inference drawn that movements in the 2-year
JGB rate contributed relatively little to observed variability
I Larger 2-year shocks to the 2-year rate may have had more
substantial effects

38 / 38

You might also like