Martina Jašová, Princeton University Richhild Moessner, Bank for International Settlements Előd Takáts, Bank for International Settlements Eesti Pank, Tallinn, Estonia 21 July 2017

1. Restricted
Exchange rate pass-through:
What has changed since the crisis?
Martina Jašová, Princeton University
Richhild Moessner, Bank for International Settlements
Előd Takáts, Bank for International Settlements
Eesti Pank, Tallinn, Estonia
21 July 2017
The views expressed in this presentation are those of the authors and not necessarily those of the BIS.

2. Restricted 2
Motivation
 Exchange rate pass-through at the centre of economic policy and
central bank thinking: (Jackson Hole)
 Bank of England: Kristin Forbes (2014, 2015)
 BIS: Global Economy Meeting, EM Deputy Governors’ Meeting
and EM Governors’ Meeting notes
 SNB: Bonadio, Fischer & Saure (2016): Speed of exchange rate
pass-through in light of a large exchange rate shock
 Question: How has pass-through to consumer prices evolved, for
both advanced and emerging economies?
 What could have driven the changes in pass-through?

3. Restricted 3
Our contribution
 New, cross-country evidence for pass-through trends:
 low and stable in advanced economies
- Forbes (2014, 2015), …
 higher but declining in emerging markets
- Mihaljek and Klau (2008), Aleem and Lahiani (2014) and Lopez-
Villavicencio and Mignon (2016)
 Show that lower inflation leads to lower pass-through in EMEs
 Calvo and Reinhart (2002) and Choudhri and Hakura (2006),
Takhtamanova (2010)
 Show that it is useful to control for non-linearities when
estimating pass-through (particularly for EMEs)
 Bussière (2013), Cheikh and Rault (2015) and Alvarez et al. (2016)

4. Restricted 4
Theory: some mechanisms driving pass-through
 Menu costs: costly to change prices
 Non-linearity: smaller exchange rate movements ignored,
larger ones are incorporated
 Lower inflation – less frequent price changes – less pass-
through
 Temporary exchange rate movements are less passed
through than permanent ones
 Invoicing currency (mostly USD) matters
 Pricing-to-market: less pass-through in larger economies

5. Restricted 5
Data
 Quarterly data
 1994 Q1 -2015 Q4
 11 advanced economies
 Australia, Canada, Denmark, the euro area, Japan, New Zealand,
Norway, Sweden, Switzerland, the United Kingdom and the United
States
 22 emerging economies
 Argentina, Brazil, Chile, China, Colombia, the Czech Republic, Hong
Kong SAR, Hungary, India, Indonesia, Israel, Korea, Mexico,
Malaysia, Peru, the Philippines, Poland, Russia, Singapore, South
Africa, Thailand and Turkey

6. Restricted 6
Methodology
 Dynamic panel regression with non-linearities:
 Technique: GMM: to address endogeneity problem
Arellano and Bover (1995) and Blundell and Bond (1998)
 For cross-checking we estimate all results also with WG
(within group) estimator: better small sample properties
 Separate estimates for advanced and emerging economies
𝜋𝑖𝑡 = 𝛼𝑖 + 𝛽𝑡 + 𝛿𝜋𝑖𝑡−1 −
𝑗=0
3
𝛾𝑗∆𝑁𝐸𝐸𝑅𝑖𝑡−𝑗 −
𝑘=0
3
𝜇 𝑘∆𝑁𝐸𝐸𝑅𝑖𝑡−𝑘
2
−
𝑙=0
3
𝜆𝑙∆𝑁𝐸𝐸𝑅𝑖𝑡−𝑙
3
+ 𝜙𝑦𝑖𝑡 + 𝜀𝑖𝑡

7. Restricted 7
Measuring linear pass-through over three horizons
 Contemporaneous:
 Yearly:
 Long-run:
 And we can measure non-linear pass-through similarly
− (𝛾0∆𝑁𝐸𝐸𝑅𝑖𝑡 + 𝛾1∆𝑁𝐸𝐸𝑅𝑖𝑡−1 +𝛾2 ∆𝑁𝐸𝐸𝑅𝑖𝑡−2 +𝛾3 ∆𝑁𝐸𝐸𝑅𝑖𝑡−3)
𝛾0 + 𝛾1 +𝛾2 +𝛾3
𝛾0 + 𝛾1 +𝛾2 +𝛾3
1 − 𝛿
𝛾0

8. Restricted 8
Exchange rate pass-through
Six-year rolling windows, based on equation (1) Graph 1
Emerging market economies Advanced economies
Emerging economies: pass-through declines after the crisis
Advanced economies: stable and low pass-through

9. Restricted 9
Regression results to quantify post-crisis change
 Use full sample: 1994-2015
 2008 Q3 – 2009 Q2, crisis years omitted
 Add dummy for post crisis years (2009 Q3-2015 Q4)
πit = αi + βt + δπit−1 − γj∆NEERit−j
3
j=0
− μk∆NEERit−k
2
3
k=0
− νl∆NEERit−l
3
3
l=0
+ ϕyit
+ δDDtπit−1 − γjD Dt∆NEERit−j
3
j=0
− μkD Dt∆NEERit−k
2
3
k=0
− νlD Dt∆NEERit−l
3
3
l=0
+ ϕDDtyit + εit

10. Restricted 10
Pass-through after the crisis
Explanatory variables EMEs AEs
Exchange rate pass-through (pre-crisis):
Contemporaneous ERPT 0.109*** 0.00469
(0.0265) (0.00616)
Yearly ERPT 0.200*** 0.00573
(0.0539) (0.00707)
Long-run ERPT 0.670*** 0.00855
(0.135) (0.0111)
Post-crisis interaction dummy:
Dt*Contemporaneous ERPT -0.0896*** 0.0218*
(0.0261) (0.0108)
Dt*Yearly ERPT -0.118*** 0.0460**
(0.0410) (0.0206)
Dt*Long-run ERPT -0.107*** 0.0533*
(0.0378) (0.0241)
Exchange rate pass-through (post-crisis):
Contemporaneous ERPT +
Dt*Contemporaneous ERPT
0.0190* 0.0265**
(0.0110) (0.00889)
Yearly ERPT +
Dt*Yearly ERPT
0.0820*** 0.0518***
(0.0188) (0.0153)
Long-run ERPT +
Dt*Long-run ERPT
0.209*** 0.0970***
(0.0521) (0.0279)

11. Restricted 11
Inflation: a potential explanation (menu costs)
Historical development (1994-2015) Post-crisis development (2009-2015)

12. Restricted 12
Can lower inflation explain lower pass-through in EMEs?
 Add interaction term for inflation and exchange rate movement
πit = αi + βt + δπit−1 − γj∆NEERit−j
3
j=0
− μk∆NEERit−k
2
− νl∆NEERit−l
3
3
l=0
3
k=0
+ ϕyit
− γjππit−4∆NEERit−j
3
j=0
− μkπ πit−4∆NEERit−k
2
3
k=0
− νlππit−4∆NEERit−l
3
3
l=0
+ εit

13. Restricted 13
Lower inflation – lower pass-through
Explanatory variables EMEs AEs
Exchange rate pass-through:
Contemporaneous ERPT 0.0656*** 0.00391
(0.0161) (0.00475)
Yearly ERPT 0.119*** 0.0167
(0.0268) (0.00930)
Long-run ERPT 0.276*** 0.0259
(0.0708) (0.0155)
Inflation interaction:
Inflationt-4*
Contemporaneous ERPT
0.677 1.557
(0.504) (1.226)
Inflationt-4*
Yearly ERPT
1.546* 0.936
(0.787) (1.314)
Inflationt-4*
Long-run ERPT
3.578* 1.454
(1.283) (1.895)

14. Restricted 14
Lower ERPT in EMEs linked to lower inflation
 Exchange rate pass-through in emerging economies on average
decreased after the financial crisis, and this decline is linked to
declining inflation.
 By contrast, in advanced economies, where inflation has tended to
be consistently low, exchange rate pass-through has also
remained low.
 Despite recent decline in emerging economies, pass-through
estimates are still lower in advanced than in emerging economies.
 These results are consistent with the menu cost theory of price
setting: when inflation is higher, exchange rate changes are
passed through more quickly and to a larger extent because firms
have to adjust prices frequently anyway (see Taylor (2000) for a
model of ERPT with sticky prices)

15. Restricted 15
Robustness checks I
 Pattern of declining pass-through in EMEs and low pass-through
in advanced economies is robust
 It holds similarly for contemporaneous (quarterly), yearly and
long-run pass-through estimates.
 It also does not depend on the length of rolling window
estimates: 3, 4, 5, 6 and 8-year rolling windows all show the
same pattern.
 Results also robust to the choice of econometric methodology:
 While our main methodology uses an Arellano and Bover
(1995) and Blundell and Bond (1998) type of system GMM
panel estimates,
 the pattern remains under difference GMM and within group
estimators (different small sample properties)

16. Restricted 16
Robustness checks II
 Results robust for control choice:
 While our main methodology uses time fixed effects to ensure
that common global shocks do not affect the estimates,
 the results also hold when dropping these fixed effects and
explicitly controlling for the global business cycle, oil prices,
and inflation expectation (Philips curve) – details in the paper
 Result robust to using bilateral exchange rates against the USD
instead of NEERs as in the main specification
 Gopinath (2015), invoicing currency

17. Restricted 17
Nonlinearity
 Relevant to control for non-linear effects of exchange rate
movements
 A the menu cost theory suggests, larger exchange rate
movements have a stronger chance to overcome the menu cost
of price changes and thereby are more likely to be passed-
through to consumer prices.
 Hence, naive linear estimates of pass-through would show an
increase in emerging markets after the taper tantrum when
exchange rate volatility increased sharply.
 Yet, we show that this increase disappears when one properly
controls for non-linearities
 One example of an extreme exchange rate movement in
Switzerland as analyzed in Bonadio, Fischer & Saure (2016)

18. Restricted 18
Pass-through omitting non-linear terms
Six-year rolling windows, based on equation (1) without non-linear terms Graph 3
Emerging market economies Advanced economies
Omitting non-linear terms

19. Restricted 19
Caveats
 Results apply for groups of countries, not individual countries
 We only looked at macroeconomic factors,
 While microeconomic factors (such as market structure…)
matter too
 Monetary policy might be special at zero lower bound
 Less room to manoeuvre
 No distinction between exogenous or endogenous ER shocks
 Forbes et al. (2015) and Shambaugh (2008)

20. Restricted 20
Conclusions
 Monetary policy needs to understand exchange rate pass-
through to consumer prices
 Jackson Hole, Bank of England, SNB
 BIS meetings: Global Economy Meeting, EM Deputy
Governors’ Meeting and EM Governors’ Meeting notes
 Three main takeaways
 Exchange rate pass-through (short and long-run) to
consumer prices has been
- low and stable in advanced economies
- higher but declining after the crisis in emerging markets
 Declining pass-through in EMEs linked to declining inflation
 Non-linearities matter when estimating ERPT