US Monetary Spillovers to Latin America The Role of Long-Term Interest Rates by Elias Albagli, Danilo Leiva-Leon, and Diego Saravia Discussion by: Constantino Hevia Universidad Torcuato di Tella Nineteenth Annual Conference of the Central Bank of Chile Santiago, Chile November, 2015
1 / 17
Outline Goal of the paper:
• To evaluate the effect of shocks to US interest rates on Latin American countries • Before and after US interest rates hit the zero bound Results:
• Shocks to long rates increase unemployment, inflation, exchange rate, and decreases stock returns • Shocks to shorter rate (1 year) have a small impact • Zero-lower bound period: a “term premium” shock seems to affect Latin American variables. Relevant question:
• U.S. is expected to increase interest rates soon • In the past, US interest rate hikes were associated with crises in emerging countries 2 / 17
Comments
• On the methodology • On the intepretation of the shocks • Some final comments and suggestions
3 / 17
Methodology • Monthly data from 5 largest LATAM countries: Brazil, Chile, Colombia, Mexico, and Peru over 2003-2015 ◦ ◦ ◦ ◦
Unemployment Inflation Exchange rate Stock returns
• “Exogenous block: US interest rate. • “Endogenous block” of LATAM variables ◦ Extract 1 factor common to each set of variables ◦ Alternative: pool the 20 variables and extract a few common factors • Factor Augmented VAR: ◦ Run VAR on the factors and US interest rate imposing exogeneity of the latter •
(Ordering of the other variables doesn’t really matter.)
◦ Run regression of LATAM variables on factors and US interest rates ◦ Compute impulse responses to a US interest rate shock 4 / 17
Empirical model
Ft rtUS
Xt rtUS
=
=
f
Λ 0
r
Λ I
Ψf (L) Ψf,US (L) 0 ΨUS (L)
Σ=
Ft rtUS
Σf 0
+
ǫt 0
+
0 ΣUS
Ft−1 US rt−1
ǫt ∼ N (0, Ω)
ef,t eUS,t
et ∼ N (0, Σ)
• Zero in upper right corner rules out contemporaneous effect of US shocks on factors!
5 / 17
Extracted factors
Unemployment
4
Brazil Chile Colombia Mexico Peru
2 0
Inflation
5
0
-2 -4 2002
2004
2006
2008
2010
2012
2014
2016
Exchange rate
5
-5 2002
2004
2006
2008
2010
2012
2014
2016
2012
2014
2016
Stock return
40
20 0 0
-5 2002
2004
2006
2008
2010
2012
2014
2016
-20 2002
2004
2006
2008
2010
6 / 17
Is 1 factor per variable enough?
Unemployment
0.9 0.8 0.7
0.8 0.6 0.4
1
2
3
4
5
1
Factors Exchange rates
2
0.9 0.8 0.7
3
4
5
4
5
Factors Stock return
1
% of variance
1
% of variance
Inflation
1
% of variance
% of variance
1
0.9 0.8 0.7 0.6
1
2
3
Factors
4
5
1
2
3
Factors
7 / 17
Example: unemployment
unemployment
Fit: 1 factor
2
2
1
1
0
0
-1
-1
-2
-2
2002
2004
2006
2008
2010
2012
2014
2002
2004
Fit: 2 factors 2
1
1
0
0
-1
-1
-2
-2 2004
2006
2008
2010
2008
2010
2012
2014
2012
2014
Fit: 3 factors
2
2002
2006
2012
2014
2002
2004
2006
2008
2010
8 / 17
What if we extract factors pooling all the data? Fraction of overall variance explained by factors
1
Fraction of variance
0.9
0.8
0.7
0.6
0.5
0.4 0
2
4
6
8
10
12
14
16
18
20
Factors 9 / 17
Rather minor differences in specification matter a lot Baseline model (shock to 10 year interest rate)
Response of Unemployment
0.1
Response of Inflation
0.2
Brazil Chile Colombia Mexico Peru
0
0.1
0
-0.1 -0.1 -0.2
-0.2
-0.3 0
5
10
15
20
25
30
35
40
Response of Exchange rate
0.1
0
5
10
15
20
25
30
35
40
30
35
40
Response of Stock return
0.2 0.15 0.1
0
0.05 0 -0.05
-0.1
-0.1 -0.15 -0.2
-0.2 0
5
10
15
20
25
30
35
40
0
5
10
15
20
25
10 / 17
Rather minor differences in specification matter a lot Extracting 4 factors pooling all the data
Response of Unemployment
0.1
Response of Inflation
0.2
0.1 0 0
Brazil Chile Colombia Mexico Peru
-0.1
-0.2
-0.1
-0.2
-0.3 0
5
10
15
20
25
30
35
40
Response of Exchange rate
0.1
0
5
10
15
20
25
30
35
40
30
35
40
Response of Stock return
0.2 0.15 0.1
0
0.05 0 -0.05
-0.1
-0.1 -0.15 -0.2
-0.2 0
5
10
15
20
25
30
35
40
0
5
10
15
20
25
11 / 17
Rather minor differences in specification matter a lot Extracting 8 factors pooling all the data
Response of Unemployment
0.1
Response of Inflation
0.2
0.1 0 0
Brazil Chile Colombia Mexico Peru
-0.1
-0.1
-0.2
-0.2
-0.3 0
5
10
15
20
25
30
35
40
Response of Exchange rate
0.1
0
5
10
15
20
25
30
35
40
30
35
40
Response of Stock return
0.2 0.15 0.1
0
0.05 0 -0.05
-0.1
-0.1 -0.15 -0.2
-0.2 0
5
10
15
20
25
30
35
40
0
5
10
15
20
25
12 / 17
Why just 1 interest rate at a time? • Somewhat odd to have an exogenous block with only 1 variable at a time ◦ ◦ ◦ ◦
1 year interest rate 10 years interest rate risk-neutral rate (≈expectation hypothesis rate) term premium (long rate - risk neutral rate)
• It may be interesting to add all rates and/or risk premia together in an exogenous block and identify the different components
• Identification may be challenging: Cholesky not the best choice
13 / 17
US interest rates
6 FF rate 1 year rate 10 years rate
5 4 3 2 1 0 2004
2006
2008
2010
2012
2014
14 / 17
US interest rates • Conclusion of the paper: ◦ 1 year rate does not affect LATAM macro and financial variables ◦ Long rate does • But in about half of the sample short interest rate was virtually zero. • Also, the process for the 1 year rate has a structural break. • Difficult to identify meaningful shocks using Cholesky decomposition from the 1 year interest rates.
15 / 17
What is a US interest rate shock? • What is an identified “US interest rate shock” ◦ Is it a monetary policy shock? ◦ Is it a demand shock? ◦ Is it a productivity shock? • Also, long-rate may be high because ◦ monetary policy changes the long rate through the expectations hypothesis ◦ or long-rate may reflect higher expected inflation • What is a term-permium shock? ◦ Term premium tends to forecast economic activity • Implications may be very different depending on the intepretation of the shock. • Model may be misspecified to answer the proposed question: ◦ May need more variables to identify more “primitive” shocks like demand, supply, monetary, etc.
16 / 17
Concluding comments and suggestion • Very relevant topic given expected increase in US interest rates: ◦ Will a tighter US monetary policy disruptive, as in the past? • At the current stage, it is difficult to interpret the results. • It may be better to focus on a clearly defined monetary policy shock
• High frequency identification using information around FOMC meetings may be useful approach
17 / 17