Effects of U.S. Quantitative Easing on Emerging Market Economies

Transcription

1 Effects of U.S. Quantitative Easing on Emerging Market Economies Saroj Bhattarai Arpita Chatterjee Woong Yong Park 3 University of Texas at Austin University of New South Wales 3 University of Illinois at Urbana-Champaign 6 th Joint BOC/ECB Conference June 8-9, 5

2 Motivation After 8, with the short-term interest rate at the ZLB, the Federal Reserve engaged in QE policy Active empirical literature on the effects (if any) of QE Literature largely focusses on domestic implications of QE Much popular discussion on spillovers to emerging markets Fragile Five countries (Brazil, India, Indonesia, South Africa, and Turkey) thought to be particularly vulnerable Were Fragile Five countries affected differently from the rest?

3 Motivation Literature largely focusses on announcement effects of QE Effects around narrow /-day windows following policy changes Advantages: can establish causality/exogeneity Disadvantages: high-frequency financial variables only; dynamic effects? Develop a framework suitable for Inferring both real and financial implications of QE Analyzing dynamic effects Studying both domestic effects and emerging market spillovers

4 3 Securities held outright (Bil $) Nominal effective exchange rates year Treasury yields (%) 5 5 S&P5 index Motivation US variables mm m m Date 8m mm m Date 8mm m m Date Notes: [] Sep 8 Lehman Brothers [] [] Nov, Dec 8 and Mar 9, QE [5] Nov, QE [6] Sep, MEP [7] [8] Sep, Dec, QE3 [9] May 3, Taper scare 8mm m m Date

5 Brazilian Real South African Rand Indian Rupee Turkish Lira 8 9 Indonesian Rupiah (in thousands) Motivation Exchange rates against USD m m m m Date 8m m m m Date 8m m m m Date 8m m m m Date 8m m m m Date

6 Our Approach Identified monthly BVAR with US data Balance sheet variable as a policy instrument from 8 to mid- Macro variables: output and consumer prices Financial variables: govt bond and equity prices Zero non-recursive restrictions to identify a US QE shock Given the identified US QE shock, assess effects on emerging markets Focus first on the Fragile Five countries and then extend to others Financial variables: exchange rates, bond and equity prices, capital flows Macro variables: output, consumer prices, trade flows

7 Related Literature Announcement effects Gagnon et al (); Krishnamurthy and Vissing-Jorgensen () VAR based identification Gambacorta et al (); Baumeister and Benati (); Wright () International effects of US QE policies Neely (); Chen et al (); Glick and Leduc (); Bauer and Neely (3) Effects on emerging markets/fragile Five Eichengreen and Gupta (3); Dahlhaus and Vasishtha (); Aizenman et al ()

8 VAR Framework Asset side component of the Fed s balance sheet as policy instrument Securities held outright by the Fed Measure of size and not composition of assets Approach similar to Gambacorta et al () A reaction function similar to conventional monetary policy The Fed responds systematically to the state of the economy Isolate the non-systematic component (shock) Fed observes current long-term Treasury yields while setting policy

9 VAR Framework Consider a VAR model A y t = A + (L)y t + ε t Use (non-recursive) short-run restrictions for identification Sims and Zha (6a,b) and Leeper, Sims, and Zha (996) Identify structural shock ε QE,t with restrictions on A Bayesian inference with a Minnesota-type prior

10 US QE Shock Identification A matrix (similar to Sims and Zha (6b)) Industrial PCE Securities -year S&P5 production deflator held-outright Treasury yields index Prod X Prod X X I X X X X X F X X a a MS a 3 a X : the corresponding coeffi cient of A is not restricted at all Blanks: the corresponding coeffi cient of A is restricted to zero Liquidity Priors: Corr (a, a ) =.8 and Corr (a 3, a ) =.8

11 Spillover Effects of QE Shock Extract the US QE shock and assess dynamic effects on emerging market economies with country specific BVARs Bayesian inference with a Minnesota-type prior Effectively assume a block exclusion structure z t = B z t + + B p z t p + D ε QE,t + + D q ε QE,t q + u t Specification Baseline: variable (IP, CPI, 3 month interest rate and USD exchange rate) VAR with the US QE shock as an exogenous variable VAR controls for domestic dynamics and shocks After baseline estimation, one additional variable at a time

12 (annualized) US QE Shock IRFs of US variables IP. PCE deflator 8 Securities year yields S&P Median responses 68% error bands.

13 US QE Shock Shock series and changes in securities held outright m m m m Date QE shocks (rescaled, left axis) Reduced form shocks to log(securities) (left axis) Growth rates in securities held outright (right axis)

14 US QE Shock Variance decomposition of US variables What is the contribution of the US QE shock? Mean and [6%, 8%] quantile Industrial PCE Securities -year S&P5 production deflator held-outright Treasury yields index Impact [.,.] [.,.] [.33,.78] [.,.5] [.,.6] 3 month [.,.] [.,.5] [.9,.7] [.,.33] [.,.] 6 month [.,.8] [.,.3] [.8,.7] [.,.33] [.,.] month [.,.6] [.5,.6] [.9,.57] [.,.36] [.,.33]

15 Spillover Effects of QE Shock USD exchange rate: Fragile five Brazil 5 India Indonesia South Africa Turkey

16 points points points points points Spillover Effects of QE Shock Long-term interest rate: Fragile five.5 Brazil. India. Indonesia South Africa. Turkey

17 Spillover Effects of QE Shock Stock price: Fragile five Brazil South Africa India Turkey Indonesia

18 Spillover Effects of QE Shock Equity flows: Fragile five Brazil India Indonesia South Africa Turkey 5 5

19 points points points points points Spillover Effects of QE Shock Net exports (US): Fragile five Brazil India Indonesia South Africa. Turkey

20 Spillover Effects of QE Shock Output: Fragile five Brazil.5 India Indonesia South Africa 3 Turkey 5 5

21 Spillover Effects of QE Shock CPI: Fragile five Brazil India Indonesia South Africa Turkey

22 Spillover Effects of QE Shock Now consider other emerging market economies Were the Fragile Five different? Qualitative or quantitative differences? Extended sample: Chile, Colombia, Malaysia, Mexico, Peru, South Korea, Taiwan, and Thailand Same specification for the country specific BVARs

23 Spillover Effects of QE Shock USD exchange rate: Other countries Chile 3 5 Colombia 6 5 Malaysia Mexico Peru 3 South Korea.5 Taiwan.5 Thailand

24 points points points points points points points points Spillover Effects of QE Shock Long-term interest rate: Other countries Chile Colombia Malaysia Mexico Peru South Korea. Taiwan. Thailand

25 Spillover Effects of QE Shock Stock price: Other countries Chile 6 5 Colombia Malaysia 6 5 Mexico Peru South Korea 5 Taiwan 6 Thailand

26 Spillover Effects of QE Shock Equity flows: Other countries Chile Colombia 5 Malaysia 5 Mexico Peru South Korea Taiwan 5 Thailand

27 points Comparison of Spillover Effects Medians of the two groups Fragile five countries respond more Exchange rates against USD Stock market indices Fragile five Other countries Long term yields 6 Cumulative equity flows

28 Pooled Spillover Effects Panel VAR Estimate the average effect of the US QE shock with a panel BVAR Allow for dynamic heterogeneity Random coeffi cient approach that partially pools the cross-section Bayesian inference with a Minnesota-type prior Consider for country i, z i,t = B i, z i,t + + B i,p z i,t p + D i, ε QE,t + + D i,q ε QE,t q + u i,t with u i,t N (, Σ i ), where B i,j = B j + v Bi,j D i,k = D k + v Di,k with v Bi,j N (, Ω Bi,j ) and vdi,k N (, Ω Di,k )

29 Pooled Spillover Effects Panel VAR

30 Extensions/Robustness Recursive short-run restrictions in US VAR? Extended 7 variable US VAR Additional corporate yields and asset prices Alternate measures of output, prices, and long-term Treasury yields in baseline US VAR

31 Robustness Recursive identification- Inference on long-term yields different.8 IP. PCE 8 Securities year yields S&P5 3

32 Robustness Recursive identification- Inference on long-term yields different.8 IP. PCE. year yields Securities 6 5 S&P5 3

33 Robustness Extended US QE Shock Identification Extended 7-variable VAR A matrix Ind PCE Securities -year Private S&P5 Additional prod deflator held Treas yields yields index asset price Prod X Prod X X I X X X X X X I X X X X X X X F X X a a F X X X X X MS a 3 a Private sector yields (BofA Merrill Lynch US corporate -5 year index; 3 year conventional mortgage rate) Additional asset prices (Effective exchange rate; Core Logic house price index)

34 Robustness Extended VAR.6 IP.5 PCE deflator 8 Securities year yields Mortgage 3 year Yield S&P5 NEER

35 Robustness Extended VAR IP. Corporate bond yields PCE deflator S&P5.5 Securities 8 6 NEER....6 year yields.5..5

36 Robustness Extended VAR.6 IP. PCE deflator 6 Securities year yields Mortgage3 House Price S&P

37 Summary of Domestic Effects of U.S. QE Shock Strong and consistent effect on both financial and real variables QE shock is estimated to Increase IP and PCE Deflator Lower long-term yields Increase stock price Depreciate the USD

38 Summary of Spillover Effects of U.S. QE Shock Relatively strong and mostly consistent effects on financial variables Appreciation against USD Reduction in long term yield Stock market boom Positive effect on equity flows Weak effects on macro variables Some evidence on reduction of net exports to the US (Fragile Five) No significant effect on IP or CPI Fragile Five countries respond more strongly than others

39 Theoretical Channels Our results might be consistent with reaching for yield or risk-taking channel of monetary policy transmission Borio and Zhu (), Bruno and Shin () Extend open economy models to account for results here Some unconventional monetary policy channels in the literature Central bank expands credit intermediation: Gertler and Karadi () Increases (otherwise scarce) collateral: Williamson () Signalling under discretion: Bhattarai, Eggertsson, and Gafarov (5)

40 Future Work Systematic policy effect evaluation Control for anticipation of QE policy Spillovers to small-open developed countries (e.g. Canada, Australia, New Zealand,...)?