2392 Views
643 Downloads |
Effects of fiscal credibility on inflation expectations: evidence from an emerging economy
Juan Camilo Anzoátegui-Zapata*
Juan Camilo Galvis-Ciro*
Juan Camilo Galvis-Ciro
Affiliation: Universidad Pontificia Bolivariana, Department of Economics, Medellin, Colombia
0000-0001-6680-275X
Article | Year: 2021 | Pages: 125 - 148 | Volume: 45 | Issue: 1 Received: July 16, 2020 | Accepted: November 10, 2020 | Published online: March 9, 2021
|
FULL ARTICLE
FIGURES & DATA
REFERENCES
CROSSMARK POLICY
METRICS
LICENCING
PDF
Source: Author´s elaboration. Data from the Central Bank of Colombia. Trend calculated with the Hodrick-Prescott filter.
Source: Author´s elaboration. Data from the Central Bank of Colombia.
Source: Author´s elaboration. Data from the Central Bank of Colombia.
Source: Author's elaboration. Data from the Central Bank of Colombia.
|
OLS
|
GMM-HAC
|
Dependent variable: E(π)
|
Model (1)
|
Model (2)
|
Model (3)
|
Model (4)
|
Model (5)
|
Model (1)
|
Model (2)
|
Model (3)
|
Model (4)
|
Model (5)
|
Constant
|
0.8989***
(0.2031)
[4.4248]
|
0.0362
(0.2488)
[0.1455]
|
0.6253**
(0.2608)
[2.3970]
|
0.9408***
(0.2767)
[3.3991]
|
0.6288**
(0.3236)
[1.9629]
|
0.9959***
(0.2499)
[3.9840]
|
0.3482**
(0.1554)
[2.2405]
|
0.6309***
(0.1903)
[3.3140]
|
0.9878***
(0.1745)
[5.6591]
|
1.0719***
(0.3441)
[3.1124]
|
E(πt-1)
|
0.6092***
(0.0851)
[7.1545]
|
0.8396***
(0.0593)
[14.1422]
|
0.3724***
(0.1189)
[3.1319]
|
0.8238***
(0.0504)
[16.3314]
|
0.4751***
(0.1174)
[4.0448]
|
0.5825***
(0.1077)
[5.4078]
|
0.7714***
(0.0450)
[17.1204]
|
0.3668***
(0.0840)
[4.3202]
|
0.7946***
(0.0366)
[21.7099]
|
0.4685***
(0.0597)
[7.8414]
|
CRED LOSSt
|
0.1054**
(0.0412)
[2.5584]
|
0.1133**
(0.0478)
[2.4474]
|
0.0947**
(0.0402)
[2.3528]
|
0.1269***
(0.0436)
[2.9071]
|
0.0915**
(0.0411)
[2.2243]
|
0.1052***
(0.0371)
[2.8319]
|
0.0959***
(0.0317)
[3.0221]
|
0.0972**
(0.0458)
[2.1210]
|
0.1213***
(0.0312)
[3.8856]
|
0.0994***
(0.0249)
[3.9793]
|
VIXt-1
|
0.0001
(0.0038)
[0.0315]
|
0.0033
(0.0044)
[0.7421]
|
0.0040
(0.0036)
[1.1140]
|
0.0047
(0.0038)
[1.2400]
|
0.0017
(0.0037)
[0.4600]
|
0.0003
(0.0021)
[0.1541]
|
0.0044
(0.0042)
[1.0539]
|
0.0022
(0.0029)
[0.7648]
|
0.0057**
(0.0026)
[2.1536]
|
0.0004
(0.0024)
[0.1691]
|
πt-1
|
0.0979***
(0.0277)
[3.5290]
|
|
0.1677***
(0.0397)
[4.2245]
|
|
0.1334***
(0.0408)
[3.2657]
|
0.1053***
(0.0312)
[3.3666]
|
|
0.1600***
(0.0252)
[6.3330]
|
|
0.1405***
(0.0178)
[7.8799]
|
et-1
|
|
0.0001**
(6.60E-05)
[2.0913]
|
|
0.0003**
(0.0001)
[1.9674]
|
0.0002*
(0.0001)
[1.9534]
|
|
9.84E-05*
(5.66E-05)
[1.7388]
|
|
0.0005**
(0.0002)
[2.2540]
|
0.0001*
(6.63E-05)
[1.9117]
|
yt-1
|
4.2515*
(2.3130)
[1.8380]
|
5.2183**
(2.6785)
[1.9681]
|
|
|
5.2481**
(2.3384)
[2.2442]
|
3.3457***
(0.9125)
[3.6661]
|
5.6147***
(1.4500)
[3.8720]
|
|
|
3.6543***
(0.9729)
[3.7558]
|
Ut-1
|
|
|
0.0969**
(0.0427)
[2.2690]
|
0.0503*
(0.0289)
[1.7391]
|
0.1257**
(0.0608)
[2.1095]
|
|
|
0.0962**
(0.0440)
[2.3806]
|
0.0472***
(0.0185)
[2.5415]
|
0.0989***
(0.0347)
[2.8453]
|
Dt
|
0.8585***
(0.1993)
[4.3064]
|
-0.2435
(0.2559)
[-0.9517]
|
0.8723***
(0.1858)
[4.6937]
|
0.9731***
(0.2023)
[4.8087]
|
0.9283***
(0.1967)
[4.7174]
|
0.9354**
(0.3533)
[2.6475]
|
0.2069
(0.3345)
[0.6163]
|
0.7943***
(0.2019)
[3.5744]
|
1.0797***
(0.3648)
[3.3984]
|
1.1167***
(0.3112)
[3.5882]
|
R2adj
|
0.91
|
0.87
|
0.92
|
0.91
|
0.92
|
0.91
|
0.85
|
0.89
|
0.89
|
0.91
|
F-statistic
|
87.05
|
65.83
|
107.59
|
97.42
|
35.57
|
|
|
|
|
|
Prob(F-Stat)
|
0.00
|
0.00
|
0.00
|
0.00
|
0.00
|
|
|
|
|
|
LM Test (1)
|
0.93
|
1.39
|
2.04
|
1.37
|
21.51
|
|
|
|
|
|
p-value (LM Test)
|
0.40
|
0.25
|
0.14
|
0.26
|
0.00
|
|
|
|
|
|
Breusch-Pagan-Godfrey test
|
0.60
|
0.96
|
0.67
|
1.39
|
1.79
|
|
|
|
|
|
p-value (B-P-G test)
|
0.72
|
0.46
|
0.66
|
0.23
|
0.14
|
|
|
|
|
|
J-statistic
|
|
|
|
|
|
4.87
|
5.21
|
2.39
|
8.50
|
3.79
|
Prob(J-stat)
|
|
|
|
|
|
0.56
|
0.39
|
0.66
|
0.20
|
0.70
|
Instruments
|
|
|
|
|
|
13
|
12
|
11
|
13
|
15
|
N. Instr./N. Obs.
|
|
|
|
|
|
0.20
|
0.19
|
0.17
|
0.20
|
0.23
|
Note: Marginal significance levels: (***) denotes 0.01, (**) denotes 0.05, and (*) denotes 0.10. Standard errors are in parentheses and t-statistics in brackets. P(F-statistic) report the respective p-value of the F-test. P(J-statistic) report the respective p-valued of the J-test. The list of GMM instruments are presented in table A6 (appendix).
Source: Author´s elaboration. Data from the Central Bank of Colombia.
Variable
|
Variable description
|
Data source
|
E(π)
|
Inflation expectations
computed by the Central Bank of Colombia. Inflation expectations are 1 year
forward. The survey is called Macroeconomic Projections of Local and Foreign
Analysts Data in %.
|
Central Bank of
Colombia
|
CRED LOSS
|
Credibility. The
indicator was constructed with the difference between the observed deficit
and the expected deficit.
|
Devised by
authors.
|
|
DEF
|
Fiscal deficit observed
by the Central National Government (% GDP) - Accumulated in the last 4
quarters. Data in %.
|
Central Bank of
Colombia
|
|
E(DEF)
|
Fiscal deficit
expectations (% GDP). Expectations come from central bank surveys. The
forecasting deficit is the overall fiscal deficit. The survey is called
Macroeconomic Projections of Local and Foreign Analysts. Data in %
|
Central Bank of
Colombia
|
|
π
|
Inflation
accumulated in 12 months measured by the variation of the consumer price
index.
Data in %.
|
Central Bank of
Colombia
|
|
e
|
Exchange rate
(month average) of the Colombian peso / United States dollar.
|
Central Bank of
Colombia
|
|
y
|
Gross domestic
product. Series was built on real Colombian currency with constant prices
from 2005. The variable is seasonally adjusted by the central bank. In the
models, the series was used in natural logarithm.
|
Central Bank of
Colombia
|
|
U
|
Unemployment
rate with seasonal adjustment X12. Data in %.
|
Central Bank of
Colombia
|
|
VIX
|
Volatility of
the stock market index S&P 500.
|
Federal Reserve
of St. Louis
|
|
D
|
A dummy
variable. The dummy takes a value of 1 for year 2008 and 0 for the rest.
|
Devised by
authors.
|
|
Variable
|
Mean
|
Min.
|
Max.
|
SD.
|
E(π)
|
3.9455
|
2.8766
|
5.6900
|
0.7804
|
DEF
|
3.0272
|
1.2166
|
4.6816
|
0.8691
|
E(DEF)
|
2.4996
|
0.6500
|
4.0800
|
0.9183
|
CRED LOSS
|
0.9798
|
0.0200
|
3.1400
|
0.7920
|
π
|
4.3057
|
1.8464
|
8.2436
|
1.6576
|
e
|
2,360.02
|
1,762.14
|
3,462.01
|
495.15
|
U
|
10.6105
|
8.4389
|
14.2076
|
1.3231
|
y
|
12.0552
|
11.7126
|
12.3205
|
0.1811
|
VIX
|
18.1577
|
9.5100
|
44.1400
|
7.7106
|
Note: E(π), DEF, E(DEF), CRED LOSS, π and U were used in %.
Series
|
ADF
|
PP
|
KPSS
|
Lags
|
Esp.
|
Test
|
C.V (1%)
|
Banda
|
Esp.
|
Test
|
C.V (1%)
|
Banda
|
Esp.
|
Test
|
C.V (5%)
|
E(π)
|
1
|
C
|
-2.44
|
-3.54
|
3
|
N
|
-1.28
|
-2.60
|
4
|
C, T
|
0.16
|
0.14
|
CRED LOSS
|
2
|
N
|
-1.16
|
-2.60
|
1
|
N
|
-1.61
|
-2.60
|
5
|
C, T
|
0.15
|
0.14
|
π
|
1
|
C
|
-3.15
|
-3.56
|
3
|
N
|
-1.06
|
-2.60
|
3
|
C, T
|
0.16
|
0.14
|
e
|
0
|
N
|
-0.65
|
-2.60
|
1
|
N
|
-0.53
|
-2.60
|
6
|
C, T
|
0.23
|
0.14
|
U
|
0
|
C
|
-2.86
|
-3.55
|
3
|
C
|
-2.86
|
-3.55
|
5
|
C
|
0.84
|
0.46
|
y
|
0
|
C, T
|
-1.75
|
-4.11
|
4
|
C, T
|
-1.59
|
-4.13
|
6
|
C
|
1.00
|
0.46
|
VIX
|
0
|
C
|
-3.61
|
-3.54
|
1
|
C
|
-3.49
|
-3.54
|
5
|
C, T
|
0.15
|
0.14
|
Note: C.V., critical value. Trend (T) and intercept (I) are included based on Schwarz criterion. ADF – the final choice of lag was made based on Schwarz criterion. PP and KPSS – spectral estimation method is Bartlett kernel and the Newey-West Bandwidth is used.
Lag
|
Schwarz
|
Lag
|
Schwarz
|
Equation
(4)
|
|
Equation (5)
|
|
0
|
8.81
|
0
|
14.31
|
1
|
1.50*
|
1
|
6.74*
|
2
|
2.08
|
2
|
7.56
|
Equation
(6)
|
|
Equation (7)
|
|
0
|
2.95
|
0
|
19.57
|
1
|
-3.75*
|
1
|
13.45*
|
2
|
-3.07
|
2
|
14.70
|
Note: * denotes the lag order selection.
Data Trend:
|
None
|
None
|
Linear
|
Linear
|
Quadratic
|
Test type
|
No
intercept
No trend
|
Intercept
No Trend
|
Intercept
No trend
|
Intercept
trend
|
Intercept
trend
|
Equation (4) trace
|
2
|
3
|
2
|
2
|
3
|
Equation (5) trace
|
2
|
2
|
2
|
2
|
3
|
Equation (6) trace
|
1
|
1
|
2
|
2
|
3
|
Equation (7) trace
|
1
|
1
|
1
|
1
|
2
|
Johansen’s cointegration test
|
Hyp. N. of CE (s)
|
Eigen
value
|
Trace
statistic
|
Critical
value (0.05)
|
p-value.
|
Equation (4)
|
|
|
|
|
R=0
|
0.50
|
96.73
|
69.81
|
0.00
|
R≤1**
|
0.46
|
60.18
|
47.85
|
0.00
|
R=2
|
0.31
|
27.34
|
29.79
|
0.09
|
Equation (5)
|
|
|
|
|
R=0
|
0.62
|
116.19
|
76.97
|
0.00
|
R≤1**
|
0.50
|
65.87
|
54.07
|
0.00
|
R=2
|
0.27
|
29.77
|
35.19
|
0.17
|
Equation (6)
|
|
|
|
|
R=0**
|
0.56
|
95.52
|
76.97
|
0.00
|
R≤1
|
0.35
|
52.48
|
54.07
|
0.06
|
Equation (7)
|
|
|
|
|
R=0**
|
0.62
|
109.60
|
88.80
|
0.00
|
R≤1
|
0.42
|
59.65
|
63.87
|
0.10
|
Note: *Model selected by the Schwarz criterion. Values based on Mackinnon. **Denotes the null hypothesis rejection at 5%.
Model 1
|
E(π)t-2,
E(π)t-3, CRED LOSSt₋₂, CRED LOSSt₋3,
πt-2, πt-3, yt₋4, yt₋5,
VIXt₋2, VIXt₋3, VIXt₋4,
Dt-1
|
Model 2
|
E(π)t-2,
E(π)t-3, CRED LOSSt₋₂, et-3, et-4,
yt₋3, yt₋4, VIXt₋₂,
VIXt₋3 VIXt₋4, Dt-1
|
Model 3
|
E(π)t-2,
E(π)t-3, E(π)t-4, CRED LOSSt₋₂, CRED LOSSt₋3,
πt-2, πt-3, Ut₋₂, VIXt₋₂, Dt-1
|
Model 4
|
E(π)t-2,
E(π)t-3, E(π)t-4, CRED LOSSt₋₂, CRED LOSSt₋3,
et-2, et-3, Ut₋2, Ut₋3,
Ut₋4, VIXt₋₂, Dt-1
|
Model 5
|
E(π)t-2,
E(π)t-3, CRED LOSSt₋₂, CRED LOSSt₋3,
et-2, et-3, yt₋4, yt₋5,
Ut₋2, Ut₋3, VIXt₋₂, πt-2,
πt-3, Dt-1
|
Figure 1Inflation expectations in the Colombian economy (in %) DISPLAY Figure
Figure 2Fiscal deficit expectations and overall fiscal deficit observed in Colombia (in % of GDP) DISPLAY Figure
Figure 3Fiscal credibility loss in Colombia case (in %) DISPLAY Figure
Figure 4Inflation expectations and fiscal credibility loss in Colombia DISPLAY Figure
Table 1Inflation expectations determinants in Colombia DISPLAY Table
Figure 5Fiscal credibility coefficient: recursive estimates, model 1, 2, 3, 4, 5 (in %) DISPLAY Figure
Table A1Sources of data and description of the variables DISPLAY Table
Table A2Descriptive statistics DISPLAY Table
Table A3Unit root tests (ADF, PP, and KPSS) DISPLAY Table
Table A4VAR lag order selection criteria (with constant) DISPLAY Table
Table A5Number of cointegrating relations by model DISPLAY Table
Table A6List of GMM instrument DISPLAY Table
* The authors would like to thank two anonymous referees for helpful comments on earlier versions of this paper. Any remaining errors are the sole responsibility of the authors.
1 Beside the presented models, we also estimated models with output gap replacing GDP as well as by including more lags of inflation. The obtained results were not statistically significant and are available upon request.
2 CRED LOSS is calculated in the units in which the fiscal deficit is measured as percentage of GDP. Therefore, an increase of one unit in CRED LOSS means that deficit expectations are 1% (% of GDP) above the observed deficit.
|
|
March, 2021 I/2021 |