Description
The purpose of this paper is to study the short-term macroeconomic effects of the fiscal
policy in Colombia for the 1980-2007 period using a structural vector autoregression (SVAR) model.
Journal of Financial Economic Policy
Assessing the macroeconomic effects of fiscal policy in Colombia
Ignacio Lozano Karen Rodríguez
Article information:
To cite this document:
Ignacio Lozano Karen Rodríguez, (2011),"Assessing the macroeconomic effects of fiscal policy in
Colombia", J ournal of Financial Economic Policy, Vol. 3 Iss 3 pp. 206 - 228
Permanent link to this document:http://dx.doi.org/10.1108/17576381111152209
Downloaded on: 24 January 2016, At: 21:42 (PT)
References: this document contains references to 19 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 601 times since 2011*
Users who downloaded this article also downloaded:
Matthew Kofi Ocran, (2011),"Fiscal policy and economic growth in South Africa", J ournal of Economic
Studies, Vol. 38 Iss 5 pp. 604-618http://dx.doi.org/10.1108/01443581111161841
Richard C.K. Burdekin, King Banaian, Mark Hallerberg, Pierre L. Siklos, (2011),"Fiscal and monetary
institutions and policies: onward and upward?", J ournal of Financial Economic Policy, Vol. 3 Iss 4 pp.
340-354http://dx.doi.org/10.1108/17576381111182918
Vera Ogeh Soli, Simon Kwadzogah Harvey, Edmond Hagan, (2008),"Fiscal policy, private investment and
economic growth: the case of Ghana", Studies in Economics and Finance, Vol. 25 Iss 2 pp. 112-130 http://
dx.doi.org/10.1108/10867370810879438
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Assessing the macroeconomic
effects of ?scal policy in Colombia
Ignacio Lozano and Karen Rodr? ´guez
Economics Research Department, Banco de la Repu´blica, Bogota´, Colombia
Abstract
Purpose – The purpose of this paper is to study the short-term macroeconomic effects of the ?scal
policy in Colombia for the 1980-2007 period using a structural vector autoregression (SVAR) model.
Design/methodology/approach – The authors’ benchmark is a ?ve-variable SVAR model which
includes government spending, output, tax revenues, in?ation and short-term interest rates.
In addition, the authors speci?ed six-variable VAR models, adding in turn private consumption,
private investment, the unemployment rate and the real minimum wage to the last set of variables.
Two alternative identi?cation techniques are used in the VARs to check the robustness of the results.
Findings – The following effects of a positive government spending shock are found. First, the GDP
responds positively and signi?cantly during the ?rst six quarters. The cumulative output multiplier
?uctuates between 1.12 and 1.19. Second, both in?ation and nominal interest rates respond positively
and signi?cantly. Third, the authors ?nd a signi?cant positive response by both private consumption
and private investment. Finally, the unemployment rate reacts negatively and signi?cantly.
Research limitations/implications – The most surprising result comes from the response of
output to a positive shock in taxes. Nonetheless, the positive respond of the GDP is short lived and has
little signi?cance.
Practical implications – The authors’ results support the smoothing role of ?scal policy on output
?uctuations, which implies its capacity to restore real activity effectively in critical times like the ones
currently being forecast.
Originality/value – The negligible results found previously for Colombia could be related to the
?scal data used, which are not keep coherence with national accounting. To solve these obstacles, a
quarterly ?scal database is assembled on an approximately accrual basis for the general government.
Keywords Colombia, Fiscal policy, Macroeconomics, Public ?nance, Monetary policy,
Keynesian economics
Paper type Case study
1. Introduction
The macroeconomic effects of ?scal policy in Colombia have not received enough
attention from analysts. Little is known about the effects on consumption and
investment resulting from the repeated tax reforms that have been implemented since
the middle of the 1980s. Neither is there conclusive evidence about the effects on output,
employment, prices and interest rates produced by the swift expansion in government
expenditure that occurred during the last few decades. The lack of knowledge in these
aspects does not contribute to the successful execution of ?scal expansion programs as
have been recently implemented by industrial countries and some emerging economies
to compensate the global ?nancial crisis of 2008-2009. Despite the dif?culty in
evaluating these issues, this paper attempts to provide evidence on the subject for
Colombia, using a structural vector autoregression (SVAR) model.
The SVAR technique has been traditionally used to assess the effects of the
monetary policy shocks. However, since the end of the 1990s, this type of model has been
employed to estimate the effects of ?scal policy shocks, but especially for the USA
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JFEP
3,3
206
Journal of Financial Economic Policy
Vol. 3 No. 3, 2011
pp. 206-228
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/17576381111152209
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and for European countries. For emerging economies like Colombia, the use of SVAR
model is scarce, largely owing to the lack of good quality data. This model relies heavily
on the existence of consistent quarterly data over a suf?ciently long period of time.
Using information contained in the Banco de la Repu´blica’s (BR’s) (The Colombia’s
Central Bank) monthly review, data from the National Planning Department (DNP) and,
especially, data from the National Department of Statistics (DANE), a quarterly
database for selected ?scal variables for the 1980:1-2007:4 period was constructed. The
?scal database is assembled on an approximately accrual basis for the general
government, which is coherent with the rest of the macroeconomic variables.
Revenue and expenditures of the Colombian Government are split by components to
differentiate their particular macroeconomic effects. Fromthe revenue side, a distinction
is made between income and consumption taxes since not all taxes give rise to similar
distortions on real activities. As a result of several tax reforms, the tax burden (at the
national level) rose from 7.8 to 16.7 percent of the GDP between 1990 and 2008. Among
the government expenditures, consumption is differentiated from investment spending
since ?scal authorities do not have the same discretion to make decisions about themand
macroeconomic variables could respond differently to innovations in these items. The
size of central government spending has increased from9.6 to 21.8 percent of the GDP in
the last two decades and almost 90 percent corresponds to consumption expenditures.
The consequences of ?scal shocks are evaluated particularly on output, private
consumption, private investment, unemployment, prices, real minimum wages and the
short-term nominal interest rates.
The main results found in this study are consistent with the real business cycle
theory and Keynesian models of both traditional partial equilibrium and new general
equilibrium types. They typically predict that an increase in government expenditures
will increase output, private consumption, employment and real interest rates. The
effects of a positive shock on net taxes are less conclusive as a whole although some
remarkable results were found. The remainder of this paper is organized as follows:
the preview papers on this subject are reviewed in Section 2; the data and the
methodological issues are described in Section 3; results are discussed in Section 4;
?nally, conclusions are drawn in Section 5.
2. Review of the literature
Some empirical research has been done but mainly for industrial countries.
For example, Blanchard and Perotti (2002) used a three-variable baseline VAR
which included government spending, net taxes and private real GDP for US data. The
identi?cation of the variables is obtained by imposing contemporaneous restrictions on
them based on the institutional features of the US tax and expenditure systems. Their
results are consistent with standard Keynesian analysis in that positive public
expenditure shocks and negative tax shocks have signi?cant and positive effects on
GDP and consumption. However, the response of private investment to increased
expenditures is negative (and positive to tax reduction), which is more consistent with
the standard neoclassical model.
A similar identi?cation method has been employed by Perotti (2004) and Gal? ´ et al.
(2006) but for different VAR speci?cations. In the case of Perotti, a ?ve-variable
baseline VAR was used for ?ve OECD and the monetary policy for short-term interest
rates and prices was included. His main results show that the effects of ?scal policy
Effects
of ?scal policy
in Colombia
207
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on GDP tend to be small and substantially weaker over time and that only in the
post-1980 (structural break) period is there some evidence of (small) positive effects of
government spending on interest rates. In the case of Gal? ´ et al., a four-variable baseline
VAR which included employment and real interest rates was used for the USA. Their
results show that output and consumption rise in response to a positive expenditure
shock. The labor variables react similarly.
Another identi?cation technique is employed by Fata´s and Mihov (2001), who used a
?ve-variable baseline VAR for US data and focused on public expenditure effects.
Anincrease inspendingleads to a persistent rise inprivate output withconsumption and
residential investment being the driving factors. The expansionary ?scal policy is also
associated with raising manufacturing wages and increasing total private employment.
In addition, the response of the real interest rate is always positive and signi?cant.
For the USA, Caldara and Kamps (2008) provide new evidence over the 1955-2006
period. Controlling for differences in the way the reduced form models are speci?ed,
they showed that all identi?cation approaches yielded similar results as regards
government spending shocks. GDP, consumption and the real wage all signi?cantly
increased while following a hump-shaped pattern. In contrast, there are strongly
diverging results as regards the effects of tax shocks.
In the context of EU countries, Marcellino (2002) imposes contemporaneous
restrictions to identify a VAR for France, Germany, Italy and Spain. He found
non-homogeneous responses from those countries along with some unexpected effects.
Giordano et al. (2005) for Italy, Biau and Girard (2005) for France, De Castro (2006) for
Spain and Perotti (2005) and Heppke-Falk et al. (2006) for Germany are authors of other
empirical studies that adopted a methodology that is relatively homogeneous to the one
used in our study on Colombia.
The results for UE economies are mixed but, in general:
.
The short-term impact of expenditure shocks is expansionary and, for the
majority of cases, the output multipliers are larger than one.
.
Positive effects on private consumption are passed through by the previous
result.
.
Shocks to net tax revenue have negligible effects on the majority of
macroeconomic variables and, in some cases, the signs are contrary to what is
expected in a Keynesian framework.
.
It is usual for the reaction of in?ation and interest rates to expansionary ?scal
policies to be positive.
.
There is no consensus on the effects on private investment.
There have been few attempts to apply SVAR models to the assessment of the ?scal
shocks in emerging countries. Cerda et al. (2005) used the standard three-variable VAR
for Chile. Their results were not conclusive since they found that a positive expenditure
(tax) shock had a negative (and marginal) effect on output during the ?rst quarter and
afterwards, the effect died out. For the Czech Republic, S
?
tikova´ (2006) initially applied
the usual three-variable SVAR and had signi?cant results. Afterward, they utilized the
?ve-variable SVAR, which included in?ation and short-term interest rate. Surprisingly,
the expansionary effects of spending (and tax cutting) became insigni?cant in the latter
speci?cation and the effects on prices and interest rates were not conclusive.
JFEP
3,3
208
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
In a comparative study of Chile and Colombia, Restrepo and Rinco´n (2006)
employed the standard three-variable SVAR for the 1990-2005 period. Their results
show that, for the Chilean case, an increase in tax revenues (expenditures) had a
transitory negative (positive) effect on GDP. Nevertheless, for Colombia, the results are
at odds with intuition since a tax revenue shock had no impact on GDP and the effect of
an expenditure shock was almost negligible.
We presume that the negligible results for Colombia could be related to the ?scal
data used. In particular, tax revenue and total expenditures were used on cash basis,
corresponded to the non-?nancial public sector – including state-owned companies.
Neither of these de?nitions (basis of recording and the public sector coverage) is
coherent with national accounting. To solve these obstacles, a quarterly ?scal database
is assembled on an approximately accrual basis for the general government. Thus, we
hope that this paper will contribute to the limited amount of evidence that exists for
Colombia by providing a more detailed analysis of ?scal policy effects.
3. Data and empirical methodology
a. The baseline SVAR
Our baseline model is a ?ve-variable VAR model, which included quarterly data on
real GDP ( y
t
), real government spending on goods and services (g
t
), real net tax
revenues (t
t
), in?ation (p
t
) measured by the consumer price index (CPI) and nominal
short-term interest rates (r
t
). In addition, six-variable VAR models were speci?ed, while
adding, in turn, real private consumption (c
t
), real private investment (i
t
),
unemployment rate (n
t
) and the real wage index (w
t
) to the set of variables.
By collecting the ?ve endogenous variables in the k-dimensional vector Y
t
, the
reduced form of the VAR model can be written as:
Y
t
¼ BðLÞY
t21
þ U
t
ð1Þ
where B(L) is lag polynomial, and U
t
is the vector of reduced form innovations with
EðU
t
Þ ¼ 0, EðU
t
U
0
t
Þ ¼
P
U
and EðU
t
U
0
s
Þ ¼ 0 for s – t. To transform the reduced
form model into a structural model, an AB model is usually employed. The AB model
describes the relationship between the reduced form disturbances, U
t
, and the
structural disturbances V
t
:
AU
t
¼ BV
t
ð2Þ
where, it is assumedthat the structural disturbances are not correlatedwitheachother, i.e.
the variance-covariance matrix of structural disturbances S
V
is diagonal. The structural
form of the VAR can be obtained by pre-multiplying the ?rst equation with matrix A:
AY
t
¼ ABðLÞY
t21
þ AU
t
¼ ABðLÞY
t21
þ BV
t
¼ DðLÞY
t21
þ BV
t
ð3Þ
In equation (3), matrix A describes the contemporaneous relationship between the
variables collected in vector Y
t
. Solving this equation for Y
t
, the structural
moving-average representation, whose coef?cients are the structural impulse-response
functions, is obtained.
Identi?cation under the Blanchard-Perotti’s approach is procured by imposing
contemporaneous restrictions based on the institutional features of the tax and
expenditure system, as well as of the timing of tax and expenditure responses to
economic activity. The A and B matrices take the following form:
Effects
of ?scal policy
in Colombia
209
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1 0 0; 5 0 0
2a
g
y
1 0 2a
t
y
0
2a
g
p
2a
y
p
1 2a
t
p
0
0 21:47 21:29 1 0
2a
g
r
2a
y
r
2a
p
r
2a
t
r
1
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
u
g
t
u
y
t
u
p
t
u
t
t
u
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
¼
b
g
0 0 0 0
0 b
y
0 0 0
0 0 b
p
0 0
b
t
g
0 0 b
t
0
0 0 0 0 b
r
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
v
g
t
v
y
t
v
p
t
v
t
t
v
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
ð4Þ
where a
i
j
(elasticity) is the automatic response of variables i to j.
Notice that seven parameters are exogenously imposed in order to achieve the full
identi?cation of the ?ve-variable SVAR. Under the recursive approach (Caldara and
Kamps, 2008), not all parameters are imposed exogenously. In practice, the recursive
approach requires a causal ordering of the model variables. The order suggested by
these authors was adopted in this paper. In particular, for a ?ve-variable VAR model,
these authors propose the order g
t
, y
t
, p
t
, t
t
and r
t
based on the following conjectures:
.
Because spending is placed ?rst (g
t
), government expenditures do not react
contemporaneously to shocks to other variables in the system[1].
.
Output is placed second ( y
t
), which implies that it does not react
contemporaneously to tax, in?ation and interest rate shocks but is affected
contemporaneously by expenditure shocks.
.
In?ation is third in the order (p
t
) meaning that in?ation does not react
contemporaneously to tax and interest rate shocks but is affected
contemporaneously by government spending shocks.
.
Tax revenue is in fourth place (t
t
), which implies that it does not react
contemporaneously to interest rate shocks but is affected contemporaneously by
government spending, output and in?ation shocks[2].
.
The interest rate is last (r
t
)[3] because it is assumed that it is affected
contemporaneously by all shocks to the system.
Thus, the recursive approach using this ordering yields an AB matrix:
1 0 0 0 0
2a
g
y
1 0 0 0
2a
g
p
2a
y
p
1 0 0
2a
g
t
2a
y
t
2a
p
t
1 0
2a
g
r
2a
y
r
2a
p
r
2a
t
r
1
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
u
g
t
u
y
t
u
p
t
u
t
t
u
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
¼
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
0 0 0 1 0
0 0 0 0 1
2
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
5
v
g
t
v
y
t
v
p
t
v
t
t
v
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
ð5Þ
b. Exogenous elasticities
The identi?cation technique suggested by Blanchard and Perotti requires exogenous
(and contemporaneous) elasticity of ?scal variables with respect to the macroeconomic
variables. In this process, the adjusted tax revenue is one of the most important
variables. For this case, Lozano and Toro (2007) estimated a
y
t
¼ 1:47 for Colombia,
which is comparable to what international evidence has shown (using the same
JFEP
3,3
210
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
technique)[4]. This estimate was used in our exercises. The other elasticities, also based
on institutional information, were simply calculated as:
lnðxÞ ¼ b
0
þb
1
lnð yÞ ð6Þ
where coef?cient b
1
measures the automatic elasticity of x with respect to y. Table I
shows the results.
Following Perotti (2004), we imposed a non-zero price elasticity of government
expenditure (20.5) since some spending decisions are ?xed in nominal terms or change
with price evolution. In addition, price-tax elasticity is estimated at 1.29, using the
de?ator of GDP, which means that tax revenue is also affected by prices within the
quarter. With respect to the short-term interest rate, the assumption was that it reacts
to output and price evolution, but not in the same quarter, i.e. the contemporaneous
response of interest rate to shocks to taxes is also set at zero. Finally, it was assumed
that government revenue also responds contemporaneously to private consumption
and investment (the elasticity of tax revenue to private consumption is calculated at
1.34 and to private investment at 0.42).
c. Data description and policy issues
Concerning the sources of the data, the macroeconomic variables (real GDP, real
private consumption, real private investment, the three-month nominal interest rate
and GDP-de?ator) stem from the database of the BR. The ?scal database is compiled
on a basis that is approximately accrual, using information from the BR monthly
review, data from the web sites of the DNP, and especially, data from the DANE.
According to national accounting, the general government covers ?scal operations of
both national and sub-national governments, as well as of the social security sector.
Public spending (g
t
) is the result of the sum of public consumption – purchases of goods
and services and compensation of civil servants – and public investment. The net tax
revenue (t
t
) is de?ned as the difference between tax revenues (with social security
contributions) minus transfers to households (including social securitypayments as well
as interest payments on public debt). Figure 1 shows the macro-?scal variables used.
Time behavior of the macro-?scal variables shown in Figure 1 could have been
in?uenced by the reforms, as well as by external shocks. The deep recession recorded
at the end of the 1990s when the Colombian economy dropped to a growth rate
of 24.3 percent was evident in the GDP growth trend (panel A). Prior to that, the
economy had exhibited an extraordinary record of ?ve decades of uninterrupted
positive GDP growth. However, Colombia was affected by the Asian and Russian
?nancial crisis of 1997-1998. As a result, to reduce in?ation and stabilize the private
GDP GDP-de?ator
Government revenue 1.47 1.29
Government spending 0 20.5
Public consumption 0 20.33
Public investment 0 20.24
Income taxes 2.43 0.46
Consumption taxes 1.13 0.21
Source: Authors’ calculations
Table I.
Exogenous elasticities
Effects
of ?scal policy
in Colombia
211
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 1.
Macroeconomic and ?scal
variables: 1980Q1-2007Q4
8.000.000
12.000.000
16.000.000
20.000.000
24.000.000
28.000.000
80 82 85 87 90 92 95 97 00 02 05 07
Panel A: real GDP
(in billons of pesos, 1994 = 100)
60
64
68
72
76
80
80 82 85 87 90 92 95 97 00 02 05 07
Panel B: private consumption
(as % of GDP)
4
8
12
16
20
24
80 82 85 87 90 92 95 97 00 02 05 07
Panel C: private investment
(as % of GDP)
0
5
10
15
20
25
30
35
40
80 82 85 87 90 92 95 97 00 02 05 07
Inflation (using CPI)
Short term interest rate
Panel D: Inflation and interest rate
0
4
8
12
16
20
24
80 82 85 87 90 92 95 97 00 02 05 07
Income taxes
Consumption taxes
Gross tax revenue
Panel F: gross tax revenue
(as % of GDP)
0
5
10
15
20
25
30
80 82 85 87 90 92 95 97 00 02 05 07
Public consumption
Public investment
Total public expenditures
Panel E : public expenditures
(as % of GDP)
240
250
260
270
280
290
300
310
80 82 85 87 90 92 95 97 00 02 05 07
Panel G: real wage index
6
8
10
12
14
16
18
20
80 82 85 87 90 92 95 97 00 02 05 07
Panel H: unemployment rate
JFEP
3,3
212
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
capital in?ows, the BR switched to a ?exible exchange rate regime, and adopted an
in?ation-targeting framework to anchor in?ationary expectations.
Real economic growth recovered to 4 percent a year in 2003-2004 and, later, reached
its highest ?gure for the current decade (7.8 percent in 2007). The recovery and boom
growth phases were driven by the economic authority’s strategy of fostering the
?nancial system, lowering in?ation and interest rates and introducing some economic
reforms which was implemented at the beginning of the newcentury (panel D). In recent
times, private investment has recovered to levels recorded in the middle of the 1990s
while the unemployment rate has dropped to the levels reached in the mid-1980s (panels
C and H)[5].
From the ?scal point of view, the decentralization process, as well as other programs
related to the constitutional changes and modernization of the state pushed government
spending up in the ?rst part of the 1990s (panel E)[6]. At the same time, the ?scal de?cit
grewdespite the recurrent tax reforms implemented during this time. Between 1990 and
2000, there were, at least, eight national tax reforms, the common features of which were
constant changes to value added tax (VAT) coverage and rates (panel F).
After the recession at the end of the 1990s, the government had to make additional
structural reforms. To reduce the actuarial de?cit in the public pension system, two
pension reforms were adopted. To smooth out the transfer of resources from the central
government to local governments to ?nance education, health and other local services,
the transfer system was adjusted twice. Between 2002 and 2006, three additional tax
reforms were introduced. In addition, in 2002, a labor market reform intended to make
it more ?exible was established.
4. The empirical results
The impulse-response functions and multipliers derived from ?scal baseline shocks
and other models are presented in this section. In all cases, the shocks correspond to
one standard deviation and the impulse-response paths are reported for a horizon of
20 quarters. Error bands are calculated by Monte-Carlo simulations based on
1,000 replications. According to the signi?cance test, and to capture institutional
changes, a dummy variable is included in our model in 1994Q1, which coincides with
the jump in the public spending level. The models also include a constant term.
Before analyzing the macroeconomic effects of the ?scal shock in the baseline model,
we brie?y comment on the results in the three-variable SVAR model: g
t
, y
t
, t
t
(Figure 2).
In both cases, that is, an unexpected increase in government spending or an unexpected
reduction in taxation, the GDP clearly responds positively and signi?cantly during the
?rst six quarters; thereafter, the effects tend to disappear steadily in the medium term.
This behavior is precisely coherent with the Keynesian prediction according to which
?scal policy is an effective tool for smoothing out the real business cycle. In addition,
these ?ndings contrast with the negligible results found previously for Colombia.
a. The effects of government spending
Figure 3 shows the impulse-response functions for a government spending shock in
our baseline model (g
t
, y
t
, p
t
, t
t
, r
t
) with the individual columns displaying the results
for the Blanchard-Perotti and recursive identi?cation approaches, respectively. The
results reveal a number of interesting issues. First of all, GDP responds positively and
signi?cantly during the ?rst six quarters and shows a typical hump-shaped form with
Effects
of ?scal policy
in Colombia
213
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
a peak effect in the third quarter at around 0.76 percent. Subsequently, it declines
gradually and becomes temporally negative before returning to zero after the fourth
year. The spending shock itself is moderately persistent with 98 percent of the shock
still present after four quarters.
As illustrated in Table II (?rst line), the cumulative output multiplier ?uctuates
between 1.12 and 1.19 from the short to medium term, which is broadly consistent with
Figure 2.
Fiscal shocks in the
three-variable SVAR
–0.4
–0.3
–0.2
–0.1
0.0
0.1
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.04
0.00
0.04
0.08
0.12
0.16
0.20
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.4
–0.3
–0.2
–0.1
0.0
0.1
0.2
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.6
–0.4
–0.2
0.0
0.2
0.4
0.6
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–1.2
–1.0
–0.8
–0.6
–0.4
–0.2
0.0
0.2
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.2
0.0
0.2
0.4
0.6
0.8
1.0
02 04 06 08 10 12 14 16 18 20
Response of government spending
Note: Responses to an increase in government spending (left column) and net taxes (right column)
JFEP
3,3
214
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 3.
Responses to an increase
in government spending
(baseline model)
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of GDP
Blanchard-Perotti approach Recursive approach
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
2.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.008
–0.004
0.000
0.004
0.008
0.012
0.016
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.2
–0.8
–0.4
0.0
0.4
0.8
1.2
1.6
2.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
0.05
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of government spending
Effects
of ?scal policy
in Colombia
215
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
what is found for Colombia using other methodologies[7]. Recently, Leigh (2008) found
a short-run output multiplier of 1.15 for Colombia, using a Global Integrated Monetary
and Fiscal Model developed by the International Monetary Fund. In order to contrast
results from Colombia at international level, Table II also presents cumulative output
multipliers estimated with analogous technique for various industrialized
economies[8].
Second, both in?ation and nominal interest rates respond positively and
signi?cantly to a positive shock in government spending, particularly after the
second year after the shock. Nonetheless, the size and time persistence of the interest
rate responses are greater than in?ation, which implies that interest rates, in real terms,
go up with increases in government expenditure. These ?ndings are coherent with the
textbook macroeconomic models and have not usually been found in previous
empirical papers.
Third, the short-run dynamics of net taxes differ substantially from the evolution of
expenditures. In contrast to the spending shock, the response of net taxes to this shock
is virtually zero. This unusual result can be associated with the net tax de?nition[9]
and the increase in the amount of interest payments on debt and pension expenditures,
which are discounted from tax revenues.
Finally, the robustness of our baseline results was checked by means of two
alternative identi?cations (the Blanchard-Perotti approach vs the recursive approach);
analternative de?nition of in?ation(GDP-de?ator insteadof CPI); the use of a third-order
lag polynomial instead of a fourth-order lag polynomial[10]; and through the use of other
elasticity values calculated for Colombia. The results do not change substantially.
b. The effects of net taxes
The response of the various macrovariables to a positive shock in net taxes in our
baseline model is shown in Figure 4. Contrary to what was found in the three-variable
SVAR model, but similar to what it has been found in other countries, the GDP
responds positively. Nonetheless, this unexpected result is short lived and has little
signi?cance. In addition, the response of net revenue to its own shock does not present
a conclusive trend and is insigni?cant over time. In?ation and, especially, short-term
interest rates responded with a signi?cant downward drop to an increase in taxes,
which is consistent with intuition. However, the responses of these variables did not
continue beyond the second year after the shock.
Quarters
4th Q 8th Q 12th Q 16th Q 20th Q
Colombia (our results) 1.12 0.59 1.20 2.11 6.01
International evidence
France (Biau and Girard, 2005) 1.90 1.40 1.50 – –
Germany (Heppke-Falk et al., 2006) 3.56 5.96 8.35 – –
Italy (Giordano et al., 2005) 1.50 1.70 1.20 0.90 –
Spain (De Castro, 2006) 1.14 1.04 0.58 20.05 20.83
USA (Blanchard and Perotti, 2002) 0.45 0.54 1.13 1.29 20.97
Source: Elaborated by the authors
Table II.
Cumulative output
multipliers
to a government
expenditure shock
JFEP
3,3
216
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 4.
Responses
to an increase in net taxes
(baseline model)
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.6
–1.2
–0.8
–0.4
0.0
0.4
0.8
02 04 06 08 10 12 14 16 18 20
Response of 3-monthr ate
–0.010
–0.005
0.000
0.005
0.010
0.015
0.020
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.05
–0.04
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
02 04 06 08 10 12 14 16 18 20
Response of GDP
Recursive approach Blanchard-Perotti approach
–0.012
–0.008
–0.004
0.000
0.004
0.008
02 04 06 08 10 12 14 16 18 20
Response of prices
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.010
–0.005
0.000
0.005
0.010
0.015
0.020
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.05
–0.04
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of net taxes
Effects
of ?scal policy
in Colombia
217
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
c. The effects on private consumption and private investment
Figures 5 and 6 show the responses of private consumption and private investment to a
positive shock in both government spending and taxes, respectively. The
impulse-response functions came from the following six-variable VAR model: g
t
, y
t
,
x
t
, p
t
, t
t
, r
t
; where x
t
is the new variable (private consumption or private investment)
that is added in turn to the baseline model. x
t
is placed third in the order which means
that private consumption (or private investment) does not react contemporaneously to
prices, taxes and interest rate shocks but is affected contemporaneously by output and
government spending shocks. However, with this placement in the order, the x
t
component is allowed to affect prices, taxes and interest rate within the same quarter.
The impulse responses show a signi?cant positive response of private consumption
to a spending shock, reproducing the GDP pattern, which is to increase steeply until it
reaches its peak in the ?fth quarter at around 0.65 percent and then decline gradually
to become temporally negative and insigni?cant in the third year. Private investment
presents a similar pattern, even if its response is greater and is shorter lived. Regarding
the tax effect (Figure 6), notice that private investment responds with a signi?cant
downward drop to a positive shock in taxes. After that, it increases gradually to
become transitorily positive but non-signi?cant after the second quarter.
Figure 5.
Responses to an increase
in government spending
Response of private consumption
Note: Using the Blanchard and Perotti identification approach
Response of private investment
0.012
0.12
0.08
0.04
–0.04
–0.08
–0.12
0.00
0.008
0.004
0.000
–0.004
–0.008
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
Figure 6.
Responses to an increase
in net taxes
Response of private consumption
Note: Using the Blanchard and Perotti identification approach
Response of private investment
0.012
0.12
0.08
0.04
–0.04
–0.08
–0.12
0.00
0.008
0.004
0.000
–0.004
–0.008
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
JFEP
3,3
218
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The patterns of response for consumption and investment are in accordance
with previous evidence from VAR analysis. In contrast, the effect on investment
has been less empirically conclusive. From a theoretical point of view, neoclassical
theory broadly predicts that consumption should fall in response to a (temporary)
spending shock while (new) Keynesian models predict that consumption will
increase. Going beyond that, the simple Keynesian theory predicts that the
response of private investment to a tax shock should be the opposite of its response
to a spending shock. The results offered in this section are in line with the latter
approach.
d. The effects of government spending by components
The responses of GDP, private consumption and private investment to a positive
government spending shock are shown by component in Figures 7 and 8. The main
components of government spending include personnel expenditure and other
operating expenditures (public consumption), as well as ?nancial expenditures on
capital formation (public investment). The impulse-response functions emerge from the
following six-variable SVAR model: x
t
, y
t
, z
t
, p
t
, t
t
, r
t
; where x
t
represents the
component of public spending (consumption or investment), and z
t
is added in turn to
capture the private consumption or private investment responses.
We found a strong and signi?cant response of output to a shock to government
investment at the same time as the effect of government consumption is weak and
non-signi?cant[11]. The private investment response after a shock to public
investment is, in general, signi?cant in the short term. Initially, it responds with a
signi?cant upward jump and subsequently declines gradually and becomes temporally
negative before returning to zero after the third year[12].
Figure 7.
Responses to an increase
of public consumption
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
2 4 6 8 10 12 14 16 18 20
Response of public consumption
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.08
–0.04
0.00
0.04
0.08
0.12
2 4 6 8 10 12 14 16 18 20
Response of private investment
Effects
of ?scal policy
in Colombia
219
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
e. The effects of taxes by component
Following the national account criteria, taxes were classi?ed into two components:
consumption and income taxes. To assess the tax effects, a six-variable VAR model
was used (g
t
, y
t
, z
t
, p
t
, x
t
, r
t
) while maintaining the same order as before. In these cases,
x
t
represents the tax component and z
t
is added in turn to capture the private
consumption or private investment responses.
A positive shock to consumption taxes shows little persistence and becomes
non-signi?cant after two quarters (Figure 9). This shock reduces GDP and private
consumption initially, but these responses are non-signi?cant in both cases. Clearly, the
short-terminterest rate reacts negativelyandsigni?cantlytoa shocktoconsumptiontaxes
in line with what one could expect[13]. However, the interest rate response is short lived.
The shock to income taxes (Figure 10) looks a little more persistent than the shock
to consumption taxes. The response of GDP and private consumption are negative and
signi?cant in the ?rst two quarters although these become slightly positive but
non-signi?cant in the medium term. The income tax shocks do appear to have
signi?cant effects on private investment in the short term. Initially, private investment
responds with a signi?cant downward drop and, thereafter, increases gradually to
become transitorily positive (but non-signi?cant) after the third quarter. Once more,
notice that the short-term interest rate reacts negatively to a positive shock to income
taxes and the response is statistically signi?cant for a longer period (seven quarters
after the shock). The fall in interest rates could compensate for the decline in output;
hence its reaction after third quarter.
In summary, shocks to income taxation in Colombia clearly seem to cause more
distortion, mainly by affecting private investment, whereas shocks to consumption
Figure 8.
Responses to an increase
in public investment
–0.10
–0.05
0.00
0.05
0.10
0.15
0.20
2 4 6 8 10 12 14 16 18 20
Response of private investment
–0.12
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
0.20
0.24
2 4 6 8 10 12 14 16 18 20
Response of public investment
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of private consumption
JFEP
3,3
220
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
taxation do not seem to affect activity signi?cantly in the short term. In principle, these
?ndings are in line with what has been found by public ?nance experts who have
generally concluded that not all taxes have the same impact on the economy. VATs that
are basically proportional taxes on consumed income are preferred to personal income
taxes that are oftenappliedwithhighmarginal taxrates onbothconsumptionandsaving.
f. The effects on the labor market
Finally, Figures 11 and 12 show the responses of the labor market to a ?scal shock by
means of six-variable SVAR models. The systems added one variable from the labor
Figure 9.
Responses to an
increase in indirect
(consumption) taxes
–0.006
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.006
–0.004
–0.002
0.000
0.002
0.004
0.006
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
0.20
2 4 6 8 10 12 14 16 18 20
Response of indirect taxes
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
2 4 6 8 10 12 14 16 18 20
Response of 3-month rate
–0.04
–0.02
0.00
0.02
0.04
0.06
0.08
0.10
2 4 6 8 10 12 14 16 18 20
Response of private investment
Effects
of ?scal policy
in Colombia
221
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
market to the ?ve variables of the baseline model, either the unemployment rate, m, or
the real index of minimum wage, w. In the ?rst case, the unemployment rate is de?ned
as the ratio of people classi?ed as unemployed to the total labor force of the seven
major cities (for the 1984Q1-2007Q4 period). In the second, minimum wage is de?ated
using the CPI. Both labor variables are constructed by employing quarterly
information from the BR web site.
The impulse responses are calculated under the recursive identi?cation approach
with the following order: g
t
, y
t
, p
t
, m
t
, t
t
, r
t
, in the ?rst system and g
t
, y
t
, p
t
, t
t
, w
t
, r
t
, in
the second one. This particular order of the variables implies:
Figure 10.
Responses to an increase
in direct (income) taxes
–0.010
–0.005
0.000
0.005
0.010
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–2.5
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
2 4 6 8 10 12 14 16 18 20
Response of 3-month rate
–0.008
–0.004
0.000
0.004
0.008
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.15
–0.10
–0.05
0.00
0.05
0.10
2 4 6 8 10 12 14 16 18 20
Response of private investment
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
2 4 6 8 10 12 14 16 18 20
Response of direct taxes
–0.010
–0.005
0.000
0.005
0.010
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–2.5
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
2 4 6 8 10 12 14 16 18 20
–0.008
–0.004
0.000
0.004
0.008
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.15
–0.10
–0.05
0.00
0.05
0.10
2 4 6 8 10 12 14 16 18 20
Response of private investment
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
2 4 6 8 10 12 14 16 18 20
Response of direct taxes JFEP
3,3
222
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
.
As unemployment rate is fourth in the order, it does not react contemporaneously
to taxes and interest rate, but it is affected contemporaneously by government
spending, output and price shocks.
.
The real minimum wage is affected by spending, output, tax and price shocks
contemporaneously, but it does not react to short-term interest rate shocks.
The spending shock has a positive and signi?cant effect on GDP, replicating the
hump-shaped pattern found initially. Both prices and short-term interest rates move
from a negative to a positive response, which becomes signi?cant after the third year.
Figure 11.
Response of
unemployment rate
to a government
spending shock
–0.6
–0.5
–0.4
–0.3
–0.2
–0.1
0.0
0.1
02 04 06 08 10 12 14 16 18 20
Response of unemployment rate
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.02
–0.01
0.00
0.01
0.02
0.03
02 04 06 08 10 12 14 16 18 20
Response of government spending Effects
of ?scal policy
in Colombia
223
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
All these effects are coherent with what was found in the baseline model. However, the
?nding that was novel came from the unemployment responses (Figure 11). The
unemployment rate reacts negatively and signi?cantly to a positive shock in
government spending in line with what one could expect. The unemployment rate falls
on impact by 0.12 percent, then decreases further to reach a maximum effect of
0.31 percent in the fourth quarter and subsequently stabilizes around 0.25 percent. The
effect on the unemployment rate is remarkably persistent over time. Figure 12 shows
that the response of the real wage index to the spending shock is not statistically
Figure 12.
Response of real minimum
wage to a government
spending shock
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
02 04 06 08 10 12 14 16 18 20
–0.008
–0.004
0.000
0.004
0.008
0.012
0.016
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
2.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of government spending
Response of real wage
JFEP
3,3
224
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cant, but GDP, in?ation and the short-term interest rate react in a manner that is
statistically signi?cant just as the previous case does.
Our results are consistent with the ?ndings of Perotti (2007) and Fata´s and Mihov
(2001), etc. who observed that employment increases after a government spending
shock. However, our ?ndings do not support the positive responses of real wages found
by these authors. From the theoretical point of view, the positive response of
employment is consistent especially with the real business cycle theory and Keynesian
models. They typically predict that an increase in government expenditure will increase
the demand for labor, consumption and output. With lump-sum taxation, the
neoclassical model studied by Baxter and King (1993) also predicts a positive effect of a
shock to government spending on total employment even though the real wage declines
under this approach.
5. Conclusions
The short-term macroeconomic effects of the ?scal policy in Colombia for the
1980-2007 period were studied using a SVAR model which has been widely
recommended by several previous empirical papers. Throughout the period of study,
the policymakers for the country undertook a series of far-reaching economic and
political reforms to modernize the state and strengthen their productive activities.
Fiscal policy played a crucial role in attaining these objectives.
Initially, macroeconomic effects of the ?scal shock were assessed by means
of the simplest model (three-variable SVAR). Either an increase in government spending
or a reduction in taxation signi?cantly expand the GDP during the ?rst six quarters;
thereafter, the effects tend to disappear steadily in the medium term. These ?ndings
contrast with the negligible results found previously for Colombia. In contrast, our
benchmarkisa?ve-variableSVARmodel whichaddsCPI-in?ationandshort-termnominal
interest rates to the previous speci?cation. In addition, we speci?ed six-variable VAR
models, adding in turn private consumption, private investment, the unemployment rate
and the real minimum wage to the last set of variables. We also split ?scal revenue and
expenditure by components to distinguish between their particular effects.
Using the baseline model and the others, the following effects of a positive
government spending shock are found. First, the GDP responds positively and
signi?cantly during the ?rst six quarters and shows a typical hump-shaped form with a
peak effect in the third quarter. Subsequently, it declines graduallyand reaches zero after
the four year. The cumulative output multiplier ?uctuates between1.12 and1.19 fromthe
short to mediumterm, whichis broadlyconsistent withwhat is found for Colombia using
alternative methodologies. Second, as expected, both in?ation and nominal interest rates
also respond positively and signi?cantly, particularly after the second year.
Third, the exercises show a signi?cant positive response by both private
consumption and private investment although the response of the later is greater and
shorter lived. Fourth, with regard to the effects of public spending shown by component,
the strong and signi?cant response of output and private investment to a public
investment shock, at the same time that the effect of government consumption is weak
and non-signi?cant is remarkable. And ?fth, the unemployment rate reacts negatively
and signi?cantly to a positive shock in government spending.
Macroeconomic effects of a positive net tax shock are less conclusive as whole
although we ?nd some remarkable results. In one case, in?ation and especially
Effects
of ?scal policy
in Colombia
225
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
short-term nominal interest rates respond with a signi?cant downward drop, which
could be consistent with intuition. However, the responses of these variables do not
continue beyond the second year. In the other, shocks to income taxation in Colombia
clearly seem to cause more distortion, because they mainly affect private investment,
whereas shocks to consumption taxation do not seem to affect real activities
signi?cantly in the short term.
Two alternative identi?cation techniques and other tools are used in the VARs to
check the robustness of our results. In general terms, the results found in this study are
consistent with real business cycle theory and Keynesian models of both traditional
partial equilibrium and new general equilibrium types. They typically predict that an
increase in government expenditures will increase output, consumption, employment
and real interest rates. Our empirical ?ndings do not support the neoclassical
prescriptions according to which increases in government expenditure crowd out the
private sector through the induced wealth effect. From a policy perspective, our results
support the smoothing role of ?scal policy on output ?uctuations, which implies its
capacity to restore real activity effectively in critical times like the ones currently being
forecast. However, it is essential that ?scal stimulus not be seen by markets as
seriously calling into question medium-term ?scal sustainability. The sustainability
analysis is beyond the scope of this paper and should be the subject of further research.
Notes
1. This assumption also implies that movements in government expenditures, unlike
movements in taxes, are largely unrelated to the business cycle.
2. Placing y
t
and p
t
before t
t
can be justi?ed on the grounds that the output and in?ation
shocks have an instantaneous effect on the tax base and, therefore, a contemporaneous
impact on tax revenues.
3. The placement of the interest rate in the order is justi?ed on two counts. One is on the
grounds of the central bank reaction function, since the interest rate is set as a function of the
output gap and in?ation, and the other is because spending and revenue as de?ned here
(net of interest payments) are not sensitive to interest rate changes.
4. The income tax to GDP elasticity is, on average: 1.3 for OECD; 1.5 for Euro area; 1.1 for new
EU members; 1.85 for the USA; and 0.95 for Germany. These values are taken from Biau and
Girard (2005), Caldara and Kamps (2008) and Heppke-Falk et al. (2006).
5. The expansionary cycle of private consumption and investment recorded in the early 1990s
was fostered by large capital in?ows and ample domestic credit (panels B and C).
6. The social security (pensions and health) and judicial systems in particular demanded
increasing public resources after the new constitution.
7. The cumulative multiplier is calculated as the ratio of the cumulative response of GDP and
the cumulative response of government expenditure for each quarter. Notice that cumulative
multipliers are lower than unit only for the second year when the response function becomes
temporally negative.
8. Notice that even though the GDP response to an increase in public spending is small with
respect to macroeconomic textbook’s ?ndings, it should keep in mind that the latter analysis
does not take into account the role of the other macroeconomic variables (like interest rates,
in?ation, etc.).
9. The net tax revenue results from the difference between gross tax revenue minus interest
payments on public debt minus pension expenditures.
JFEP
3,3
226
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
10. The third- and fourth-order lag polynomial is coherent with the optimal criterion of Akaike
(AIC), Schwarz (SC) and Hannan-Quinn (HQC).
11. The cumulative output multiplier to shocks on public investment, which is estimated in the
simplest model as the ratio of the cumulative response of GDP and the cumulative response
of government investment, rises to 3.64 in the fourth year.
12. The signi?cant response of output to a change in public investment was found previously for
Colombia by Leigh (2008), who concluded that a permanent cut of 0.5 percent points of GDP
in public investment leads to a long-term contraction of 0.86 percent of GDP due to a
contraction in the supply capacity of the economy. However, if public investment is not
productive, a cut in capital expenditures has an effect that is broadly similar to a reduction in
public consumption (GDP declines 0.58 percent).
13. As already mentioned, the shock reduces private consumption and aggregate demand and,
for a given short-term aggregate supply, it also reduces prices and interest rates.
References
Baxter, M. and King, R.G. (1993), “Fiscal policy in general equilibrium”, American Economic
Review, Vol. 83, pp. 315-34.
Biau, O. and Girard, E. (2005), “Politique budge´taire et dynamique e´conomique en France:
l’approche VAR structurel”, Economie et Pre´vision, Nos 169-171, pp. 1-23.
Blanchard, O. and Perotti, R. (2002), “An empirical characterization of the dynamic effects of
changes in government spending and taxes on output”, Quarterly Journal of Economics,
Vol. 107.
Caldara, D. and Kamps, C. (2008), “What are the effects of ?scal shocks? A VAR-based
comparative analysis”, Working Paper Series No. 877, European Central Bank, Frankfurt
am Main.
Cerda, R., Gonza´lez, H. and Lagos, L. (2005), “Is ?scal policy effective? Evidence for an emerging
economy: Chile 1833-2000”, Documento de Trabajo IE-PUC No. 292, Ponti?cia Universidad
Catolica de Chile, Santiago.
De Castro, F. (2006), “The macroeconomic effects of ?scal policy in Spain”, Documento de
Trabajo, No. 0311, Banco de Espan˜a, Madrid.
Fata´s, A. and Mihov, I. (2001), “The effects of ?scal policy on consumption and employment:
theory and evidence”, mimeo, INSEAD, New York, NY.
Gal? ´, J., Lo´pez-Salido, J. and Valle´s, J. (2006), “Understanding the effects of government spending
on consumption”, Journal of the European Economic Association, Vol. 5 No. 1.
Giordano, R., Momigliano, S., Neri, S. and Perotti, R. (2005), “The effects on the economy of
shocks to different government expenditures items: estimates with a SVAR model”, paper
presented at the 7th Workshop on Public Finance, Banca d’Italia, Rome.
Heppke-Falk, K., Tenhofen, J. and Wolff, G. (2006), “The macroeconomic consequences of
exogenous ?scal policy shocks in Germany: a disaggregated SVAR analysis”, Discussion
Paper No. 41, Deutsche Bundesbank, Frankfurt am Main.
Leigh, D (2008), “Achiving a solft landing: the role ?scal policy”, IMF Working Paper No. 0869.
Lozano, I. and Toro, J. (2007), “Fiscal policy throughout the cycle: the Colombian experience”,
Borradores de Econom? ´a No. 434, Banco de la Repu´blica, Bogota´.
Marcellino, M. (2002), “Some stylized facts on non-systematic ?scal policy in the Euro area”,
Discussion Paper No. 3635, Center for Economic Policy Research, Washington, DC.
Effects
of ?scal policy
in Colombia
227
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Perotti, R. (2004), “Estimating the effects of ?scal policy in OECD countries”, mimeo, Bocconi
University, Milan.
Perotti, R. (2005), “Public investment: another (different) look”, mimeo, Bocconi University,
Milan.
Perotti, R. (2007), “In search of the transmission mechanism of ?scal policy”, NBER Working
Papers No. 13143, National Bureau of Economic Research, New York, NY.
Restrepo, J. and Rinco´n, H. (2006), “Identifying ?scal policy shocks in Chile and Colombia”,
Borradores de Econom? ´a No. 397, Banco de la Repu´blica, Bogota´.
S
?
tikova´, R. (2006), “Effects of ?scal policy in the Czech Republic: a SVAR analysis”, Mimeo,
Alternative Approaches to Economic Modelling, IES, Charles University, Prague.
Further reading
Aschauer, D. (1989), “Is public expenditure productive?”, Journal of Monetary Economics, Vol. 23,
pp. 177-200.
About the authors
Ignacio Lozano is a Researcher in the Economics Research Department, Banco de la Repu´blica
(the Central Bank of Colombia). Ignacio Lozano is the corresponding author and can be contacted
at: [email protected]
Karen Rodr? ´guez is a Public Finance Specialist in the Economics Research Department,
Banco de la Repu´blica.
JFEP
3,3
228
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_410619293.pdf
The purpose of this paper is to study the short-term macroeconomic effects of the fiscal
policy in Colombia for the 1980-2007 period using a structural vector autoregression (SVAR) model.
Journal of Financial Economic Policy
Assessing the macroeconomic effects of fiscal policy in Colombia
Ignacio Lozano Karen Rodríguez
Article information:
To cite this document:
Ignacio Lozano Karen Rodríguez, (2011),"Assessing the macroeconomic effects of fiscal policy in
Colombia", J ournal of Financial Economic Policy, Vol. 3 Iss 3 pp. 206 - 228
Permanent link to this document:http://dx.doi.org/10.1108/17576381111152209
Downloaded on: 24 January 2016, At: 21:42 (PT)
References: this document contains references to 19 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 601 times since 2011*
Users who downloaded this article also downloaded:
Matthew Kofi Ocran, (2011),"Fiscal policy and economic growth in South Africa", J ournal of Economic
Studies, Vol. 38 Iss 5 pp. 604-618http://dx.doi.org/10.1108/01443581111161841
Richard C.K. Burdekin, King Banaian, Mark Hallerberg, Pierre L. Siklos, (2011),"Fiscal and monetary
institutions and policies: onward and upward?", J ournal of Financial Economic Policy, Vol. 3 Iss 4 pp.
340-354http://dx.doi.org/10.1108/17576381111182918
Vera Ogeh Soli, Simon Kwadzogah Harvey, Edmond Hagan, (2008),"Fiscal policy, private investment and
economic growth: the case of Ghana", Studies in Economics and Finance, Vol. 25 Iss 2 pp. 112-130 http://
dx.doi.org/10.1108/10867370810879438
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Assessing the macroeconomic
effects of ?scal policy in Colombia
Ignacio Lozano and Karen Rodr? ´guez
Economics Research Department, Banco de la Repu´blica, Bogota´, Colombia
Abstract
Purpose – The purpose of this paper is to study the short-term macroeconomic effects of the ?scal
policy in Colombia for the 1980-2007 period using a structural vector autoregression (SVAR) model.
Design/methodology/approach – The authors’ benchmark is a ?ve-variable SVAR model which
includes government spending, output, tax revenues, in?ation and short-term interest rates.
In addition, the authors speci?ed six-variable VAR models, adding in turn private consumption,
private investment, the unemployment rate and the real minimum wage to the last set of variables.
Two alternative identi?cation techniques are used in the VARs to check the robustness of the results.
Findings – The following effects of a positive government spending shock are found. First, the GDP
responds positively and signi?cantly during the ?rst six quarters. The cumulative output multiplier
?uctuates between 1.12 and 1.19. Second, both in?ation and nominal interest rates respond positively
and signi?cantly. Third, the authors ?nd a signi?cant positive response by both private consumption
and private investment. Finally, the unemployment rate reacts negatively and signi?cantly.
Research limitations/implications – The most surprising result comes from the response of
output to a positive shock in taxes. Nonetheless, the positive respond of the GDP is short lived and has
little signi?cance.
Practical implications – The authors’ results support the smoothing role of ?scal policy on output
?uctuations, which implies its capacity to restore real activity effectively in critical times like the ones
currently being forecast.
Originality/value – The negligible results found previously for Colombia could be related to the
?scal data used, which are not keep coherence with national accounting. To solve these obstacles, a
quarterly ?scal database is assembled on an approximately accrual basis for the general government.
Keywords Colombia, Fiscal policy, Macroeconomics, Public ?nance, Monetary policy,
Keynesian economics
Paper type Case study
1. Introduction
The macroeconomic effects of ?scal policy in Colombia have not received enough
attention from analysts. Little is known about the effects on consumption and
investment resulting from the repeated tax reforms that have been implemented since
the middle of the 1980s. Neither is there conclusive evidence about the effects on output,
employment, prices and interest rates produced by the swift expansion in government
expenditure that occurred during the last few decades. The lack of knowledge in these
aspects does not contribute to the successful execution of ?scal expansion programs as
have been recently implemented by industrial countries and some emerging economies
to compensate the global ?nancial crisis of 2008-2009. Despite the dif?culty in
evaluating these issues, this paper attempts to provide evidence on the subject for
Colombia, using a structural vector autoregression (SVAR) model.
The SVAR technique has been traditionally used to assess the effects of the
monetary policy shocks. However, since the end of the 1990s, this type of model has been
employed to estimate the effects of ?scal policy shocks, but especially for the USA
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JFEP
3,3
206
Journal of Financial Economic Policy
Vol. 3 No. 3, 2011
pp. 206-228
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/17576381111152209
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and for European countries. For emerging economies like Colombia, the use of SVAR
model is scarce, largely owing to the lack of good quality data. This model relies heavily
on the existence of consistent quarterly data over a suf?ciently long period of time.
Using information contained in the Banco de la Repu´blica’s (BR’s) (The Colombia’s
Central Bank) monthly review, data from the National Planning Department (DNP) and,
especially, data from the National Department of Statistics (DANE), a quarterly
database for selected ?scal variables for the 1980:1-2007:4 period was constructed. The
?scal database is assembled on an approximately accrual basis for the general
government, which is coherent with the rest of the macroeconomic variables.
Revenue and expenditures of the Colombian Government are split by components to
differentiate their particular macroeconomic effects. Fromthe revenue side, a distinction
is made between income and consumption taxes since not all taxes give rise to similar
distortions on real activities. As a result of several tax reforms, the tax burden (at the
national level) rose from 7.8 to 16.7 percent of the GDP between 1990 and 2008. Among
the government expenditures, consumption is differentiated from investment spending
since ?scal authorities do not have the same discretion to make decisions about themand
macroeconomic variables could respond differently to innovations in these items. The
size of central government spending has increased from9.6 to 21.8 percent of the GDP in
the last two decades and almost 90 percent corresponds to consumption expenditures.
The consequences of ?scal shocks are evaluated particularly on output, private
consumption, private investment, unemployment, prices, real minimum wages and the
short-term nominal interest rates.
The main results found in this study are consistent with the real business cycle
theory and Keynesian models of both traditional partial equilibrium and new general
equilibrium types. They typically predict that an increase in government expenditures
will increase output, private consumption, employment and real interest rates. The
effects of a positive shock on net taxes are less conclusive as a whole although some
remarkable results were found. The remainder of this paper is organized as follows:
the preview papers on this subject are reviewed in Section 2; the data and the
methodological issues are described in Section 3; results are discussed in Section 4;
?nally, conclusions are drawn in Section 5.
2. Review of the literature
Some empirical research has been done but mainly for industrial countries.
For example, Blanchard and Perotti (2002) used a three-variable baseline VAR
which included government spending, net taxes and private real GDP for US data. The
identi?cation of the variables is obtained by imposing contemporaneous restrictions on
them based on the institutional features of the US tax and expenditure systems. Their
results are consistent with standard Keynesian analysis in that positive public
expenditure shocks and negative tax shocks have signi?cant and positive effects on
GDP and consumption. However, the response of private investment to increased
expenditures is negative (and positive to tax reduction), which is more consistent with
the standard neoclassical model.
A similar identi?cation method has been employed by Perotti (2004) and Gal? ´ et al.
(2006) but for different VAR speci?cations. In the case of Perotti, a ?ve-variable
baseline VAR was used for ?ve OECD and the monetary policy for short-term interest
rates and prices was included. His main results show that the effects of ?scal policy
Effects
of ?scal policy
in Colombia
207
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on GDP tend to be small and substantially weaker over time and that only in the
post-1980 (structural break) period is there some evidence of (small) positive effects of
government spending on interest rates. In the case of Gal? ´ et al., a four-variable baseline
VAR which included employment and real interest rates was used for the USA. Their
results show that output and consumption rise in response to a positive expenditure
shock. The labor variables react similarly.
Another identi?cation technique is employed by Fata´s and Mihov (2001), who used a
?ve-variable baseline VAR for US data and focused on public expenditure effects.
Anincrease inspendingleads to a persistent rise inprivate output withconsumption and
residential investment being the driving factors. The expansionary ?scal policy is also
associated with raising manufacturing wages and increasing total private employment.
In addition, the response of the real interest rate is always positive and signi?cant.
For the USA, Caldara and Kamps (2008) provide new evidence over the 1955-2006
period. Controlling for differences in the way the reduced form models are speci?ed,
they showed that all identi?cation approaches yielded similar results as regards
government spending shocks. GDP, consumption and the real wage all signi?cantly
increased while following a hump-shaped pattern. In contrast, there are strongly
diverging results as regards the effects of tax shocks.
In the context of EU countries, Marcellino (2002) imposes contemporaneous
restrictions to identify a VAR for France, Germany, Italy and Spain. He found
non-homogeneous responses from those countries along with some unexpected effects.
Giordano et al. (2005) for Italy, Biau and Girard (2005) for France, De Castro (2006) for
Spain and Perotti (2005) and Heppke-Falk et al. (2006) for Germany are authors of other
empirical studies that adopted a methodology that is relatively homogeneous to the one
used in our study on Colombia.
The results for UE economies are mixed but, in general:
.
The short-term impact of expenditure shocks is expansionary and, for the
majority of cases, the output multipliers are larger than one.
.
Positive effects on private consumption are passed through by the previous
result.
.
Shocks to net tax revenue have negligible effects on the majority of
macroeconomic variables and, in some cases, the signs are contrary to what is
expected in a Keynesian framework.
.
It is usual for the reaction of in?ation and interest rates to expansionary ?scal
policies to be positive.
.
There is no consensus on the effects on private investment.
There have been few attempts to apply SVAR models to the assessment of the ?scal
shocks in emerging countries. Cerda et al. (2005) used the standard three-variable VAR
for Chile. Their results were not conclusive since they found that a positive expenditure
(tax) shock had a negative (and marginal) effect on output during the ?rst quarter and
afterwards, the effect died out. For the Czech Republic, S
?
tikova´ (2006) initially applied
the usual three-variable SVAR and had signi?cant results. Afterward, they utilized the
?ve-variable SVAR, which included in?ation and short-term interest rate. Surprisingly,
the expansionary effects of spending (and tax cutting) became insigni?cant in the latter
speci?cation and the effects on prices and interest rates were not conclusive.
JFEP
3,3
208
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
In a comparative study of Chile and Colombia, Restrepo and Rinco´n (2006)
employed the standard three-variable SVAR for the 1990-2005 period. Their results
show that, for the Chilean case, an increase in tax revenues (expenditures) had a
transitory negative (positive) effect on GDP. Nevertheless, for Colombia, the results are
at odds with intuition since a tax revenue shock had no impact on GDP and the effect of
an expenditure shock was almost negligible.
We presume that the negligible results for Colombia could be related to the ?scal
data used. In particular, tax revenue and total expenditures were used on cash basis,
corresponded to the non-?nancial public sector – including state-owned companies.
Neither of these de?nitions (basis of recording and the public sector coverage) is
coherent with national accounting. To solve these obstacles, a quarterly ?scal database
is assembled on an approximately accrual basis for the general government. Thus, we
hope that this paper will contribute to the limited amount of evidence that exists for
Colombia by providing a more detailed analysis of ?scal policy effects.
3. Data and empirical methodology
a. The baseline SVAR
Our baseline model is a ?ve-variable VAR model, which included quarterly data on
real GDP ( y
t
), real government spending on goods and services (g
t
), real net tax
revenues (t
t
), in?ation (p
t
) measured by the consumer price index (CPI) and nominal
short-term interest rates (r
t
). In addition, six-variable VAR models were speci?ed, while
adding, in turn, real private consumption (c
t
), real private investment (i
t
),
unemployment rate (n
t
) and the real wage index (w
t
) to the set of variables.
By collecting the ?ve endogenous variables in the k-dimensional vector Y
t
, the
reduced form of the VAR model can be written as:
Y
t
¼ BðLÞY
t21
þ U
t
ð1Þ
where B(L) is lag polynomial, and U
t
is the vector of reduced form innovations with
EðU
t
Þ ¼ 0, EðU
t
U
0
t
Þ ¼
P
U
and EðU
t
U
0
s
Þ ¼ 0 for s – t. To transform the reduced
form model into a structural model, an AB model is usually employed. The AB model
describes the relationship between the reduced form disturbances, U
t
, and the
structural disturbances V
t
:
AU
t
¼ BV
t
ð2Þ
where, it is assumedthat the structural disturbances are not correlatedwitheachother, i.e.
the variance-covariance matrix of structural disturbances S
V
is diagonal. The structural
form of the VAR can be obtained by pre-multiplying the ?rst equation with matrix A:
AY
t
¼ ABðLÞY
t21
þ AU
t
¼ ABðLÞY
t21
þ BV
t
¼ DðLÞY
t21
þ BV
t
ð3Þ
In equation (3), matrix A describes the contemporaneous relationship between the
variables collected in vector Y
t
. Solving this equation for Y
t
, the structural
moving-average representation, whose coef?cients are the structural impulse-response
functions, is obtained.
Identi?cation under the Blanchard-Perotti’s approach is procured by imposing
contemporaneous restrictions based on the institutional features of the tax and
expenditure system, as well as of the timing of tax and expenditure responses to
economic activity. The A and B matrices take the following form:
Effects
of ?scal policy
in Colombia
209
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1 0 0; 5 0 0
2a
g
y
1 0 2a
t
y
0
2a
g
p
2a
y
p
1 2a
t
p
0
0 21:47 21:29 1 0
2a
g
r
2a
y
r
2a
p
r
2a
t
r
1
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
u
g
t
u
y
t
u
p
t
u
t
t
u
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
¼
b
g
0 0 0 0
0 b
y
0 0 0
0 0 b
p
0 0
b
t
g
0 0 b
t
0
0 0 0 0 b
r
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
v
g
t
v
y
t
v
p
t
v
t
t
v
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
ð4Þ
where a
i
j
(elasticity) is the automatic response of variables i to j.
Notice that seven parameters are exogenously imposed in order to achieve the full
identi?cation of the ?ve-variable SVAR. Under the recursive approach (Caldara and
Kamps, 2008), not all parameters are imposed exogenously. In practice, the recursive
approach requires a causal ordering of the model variables. The order suggested by
these authors was adopted in this paper. In particular, for a ?ve-variable VAR model,
these authors propose the order g
t
, y
t
, p
t
, t
t
and r
t
based on the following conjectures:
.
Because spending is placed ?rst (g
t
), government expenditures do not react
contemporaneously to shocks to other variables in the system[1].
.
Output is placed second ( y
t
), which implies that it does not react
contemporaneously to tax, in?ation and interest rate shocks but is affected
contemporaneously by expenditure shocks.
.
In?ation is third in the order (p
t
) meaning that in?ation does not react
contemporaneously to tax and interest rate shocks but is affected
contemporaneously by government spending shocks.
.
Tax revenue is in fourth place (t
t
), which implies that it does not react
contemporaneously to interest rate shocks but is affected contemporaneously by
government spending, output and in?ation shocks[2].
.
The interest rate is last (r
t
)[3] because it is assumed that it is affected
contemporaneously by all shocks to the system.
Thus, the recursive approach using this ordering yields an AB matrix:
1 0 0 0 0
2a
g
y
1 0 0 0
2a
g
p
2a
y
p
1 0 0
2a
g
t
2a
y
t
2a
p
t
1 0
2a
g
r
2a
y
r
2a
p
r
2a
t
r
1
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
u
g
t
u
y
t
u
p
t
u
t
t
u
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
¼
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
0 0 0 1 0
0 0 0 0 1
2
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
5
v
g
t
v
y
t
v
p
t
v
t
t
v
r
t
2
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
5
ð5Þ
b. Exogenous elasticities
The identi?cation technique suggested by Blanchard and Perotti requires exogenous
(and contemporaneous) elasticity of ?scal variables with respect to the macroeconomic
variables. In this process, the adjusted tax revenue is one of the most important
variables. For this case, Lozano and Toro (2007) estimated a
y
t
¼ 1:47 for Colombia,
which is comparable to what international evidence has shown (using the same
JFEP
3,3
210
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
technique)[4]. This estimate was used in our exercises. The other elasticities, also based
on institutional information, were simply calculated as:
lnðxÞ ¼ b
0
þb
1
lnð yÞ ð6Þ
where coef?cient b
1
measures the automatic elasticity of x with respect to y. Table I
shows the results.
Following Perotti (2004), we imposed a non-zero price elasticity of government
expenditure (20.5) since some spending decisions are ?xed in nominal terms or change
with price evolution. In addition, price-tax elasticity is estimated at 1.29, using the
de?ator of GDP, which means that tax revenue is also affected by prices within the
quarter. With respect to the short-term interest rate, the assumption was that it reacts
to output and price evolution, but not in the same quarter, i.e. the contemporaneous
response of interest rate to shocks to taxes is also set at zero. Finally, it was assumed
that government revenue also responds contemporaneously to private consumption
and investment (the elasticity of tax revenue to private consumption is calculated at
1.34 and to private investment at 0.42).
c. Data description and policy issues
Concerning the sources of the data, the macroeconomic variables (real GDP, real
private consumption, real private investment, the three-month nominal interest rate
and GDP-de?ator) stem from the database of the BR. The ?scal database is compiled
on a basis that is approximately accrual, using information from the BR monthly
review, data from the web sites of the DNP, and especially, data from the DANE.
According to national accounting, the general government covers ?scal operations of
both national and sub-national governments, as well as of the social security sector.
Public spending (g
t
) is the result of the sum of public consumption – purchases of goods
and services and compensation of civil servants – and public investment. The net tax
revenue (t
t
) is de?ned as the difference between tax revenues (with social security
contributions) minus transfers to households (including social securitypayments as well
as interest payments on public debt). Figure 1 shows the macro-?scal variables used.
Time behavior of the macro-?scal variables shown in Figure 1 could have been
in?uenced by the reforms, as well as by external shocks. The deep recession recorded
at the end of the 1990s when the Colombian economy dropped to a growth rate
of 24.3 percent was evident in the GDP growth trend (panel A). Prior to that, the
economy had exhibited an extraordinary record of ?ve decades of uninterrupted
positive GDP growth. However, Colombia was affected by the Asian and Russian
?nancial crisis of 1997-1998. As a result, to reduce in?ation and stabilize the private
GDP GDP-de?ator
Government revenue 1.47 1.29
Government spending 0 20.5
Public consumption 0 20.33
Public investment 0 20.24
Income taxes 2.43 0.46
Consumption taxes 1.13 0.21
Source: Authors’ calculations
Table I.
Exogenous elasticities
Effects
of ?scal policy
in Colombia
211
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 1.
Macroeconomic and ?scal
variables: 1980Q1-2007Q4
8.000.000
12.000.000
16.000.000
20.000.000
24.000.000
28.000.000
80 82 85 87 90 92 95 97 00 02 05 07
Panel A: real GDP
(in billons of pesos, 1994 = 100)
60
64
68
72
76
80
80 82 85 87 90 92 95 97 00 02 05 07
Panel B: private consumption
(as % of GDP)
4
8
12
16
20
24
80 82 85 87 90 92 95 97 00 02 05 07
Panel C: private investment
(as % of GDP)
0
5
10
15
20
25
30
35
40
80 82 85 87 90 92 95 97 00 02 05 07
Inflation (using CPI)
Short term interest rate
Panel D: Inflation and interest rate
0
4
8
12
16
20
24
80 82 85 87 90 92 95 97 00 02 05 07
Income taxes
Consumption taxes
Gross tax revenue
Panel F: gross tax revenue
(as % of GDP)
0
5
10
15
20
25
30
80 82 85 87 90 92 95 97 00 02 05 07
Public consumption
Public investment
Total public expenditures
Panel E : public expenditures
(as % of GDP)
240
250
260
270
280
290
300
310
80 82 85 87 90 92 95 97 00 02 05 07
Panel G: real wage index
6
8
10
12
14
16
18
20
80 82 85 87 90 92 95 97 00 02 05 07
Panel H: unemployment rate
JFEP
3,3
212
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
capital in?ows, the BR switched to a ?exible exchange rate regime, and adopted an
in?ation-targeting framework to anchor in?ationary expectations.
Real economic growth recovered to 4 percent a year in 2003-2004 and, later, reached
its highest ?gure for the current decade (7.8 percent in 2007). The recovery and boom
growth phases were driven by the economic authority’s strategy of fostering the
?nancial system, lowering in?ation and interest rates and introducing some economic
reforms which was implemented at the beginning of the newcentury (panel D). In recent
times, private investment has recovered to levels recorded in the middle of the 1990s
while the unemployment rate has dropped to the levels reached in the mid-1980s (panels
C and H)[5].
From the ?scal point of view, the decentralization process, as well as other programs
related to the constitutional changes and modernization of the state pushed government
spending up in the ?rst part of the 1990s (panel E)[6]. At the same time, the ?scal de?cit
grewdespite the recurrent tax reforms implemented during this time. Between 1990 and
2000, there were, at least, eight national tax reforms, the common features of which were
constant changes to value added tax (VAT) coverage and rates (panel F).
After the recession at the end of the 1990s, the government had to make additional
structural reforms. To reduce the actuarial de?cit in the public pension system, two
pension reforms were adopted. To smooth out the transfer of resources from the central
government to local governments to ?nance education, health and other local services,
the transfer system was adjusted twice. Between 2002 and 2006, three additional tax
reforms were introduced. In addition, in 2002, a labor market reform intended to make
it more ?exible was established.
4. The empirical results
The impulse-response functions and multipliers derived from ?scal baseline shocks
and other models are presented in this section. In all cases, the shocks correspond to
one standard deviation and the impulse-response paths are reported for a horizon of
20 quarters. Error bands are calculated by Monte-Carlo simulations based on
1,000 replications. According to the signi?cance test, and to capture institutional
changes, a dummy variable is included in our model in 1994Q1, which coincides with
the jump in the public spending level. The models also include a constant term.
Before analyzing the macroeconomic effects of the ?scal shock in the baseline model,
we brie?y comment on the results in the three-variable SVAR model: g
t
, y
t
, t
t
(Figure 2).
In both cases, that is, an unexpected increase in government spending or an unexpected
reduction in taxation, the GDP clearly responds positively and signi?cantly during the
?rst six quarters; thereafter, the effects tend to disappear steadily in the medium term.
This behavior is precisely coherent with the Keynesian prediction according to which
?scal policy is an effective tool for smoothing out the real business cycle. In addition,
these ?ndings contrast with the negligible results found previously for Colombia.
a. The effects of government spending
Figure 3 shows the impulse-response functions for a government spending shock in
our baseline model (g
t
, y
t
, p
t
, t
t
, r
t
) with the individual columns displaying the results
for the Blanchard-Perotti and recursive identi?cation approaches, respectively. The
results reveal a number of interesting issues. First of all, GDP responds positively and
signi?cantly during the ?rst six quarters and shows a typical hump-shaped form with
Effects
of ?scal policy
in Colombia
213
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
a peak effect in the third quarter at around 0.76 percent. Subsequently, it declines
gradually and becomes temporally negative before returning to zero after the fourth
year. The spending shock itself is moderately persistent with 98 percent of the shock
still present after four quarters.
As illustrated in Table II (?rst line), the cumulative output multiplier ?uctuates
between 1.12 and 1.19 from the short to medium term, which is broadly consistent with
Figure 2.
Fiscal shocks in the
three-variable SVAR
–0.4
–0.3
–0.2
–0.1
0.0
0.1
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.04
0.00
0.04
0.08
0.12
0.16
0.20
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.4
–0.3
–0.2
–0.1
0.0
0.1
0.2
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.6
–0.4
–0.2
0.0
0.2
0.4
0.6
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–1.2
–1.0
–0.8
–0.6
–0.4
–0.2
0.0
0.2
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.2
0.0
0.2
0.4
0.6
0.8
1.0
02 04 06 08 10 12 14 16 18 20
Response of government spending
Note: Responses to an increase in government spending (left column) and net taxes (right column)
JFEP
3,3
214
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 3.
Responses to an increase
in government spending
(baseline model)
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of GDP
Blanchard-Perotti approach Recursive approach
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
2.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.008
–0.004
0.000
0.004
0.008
0.012
0.016
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.2
–0.8
–0.4
0.0
0.4
0.8
1.2
1.6
2.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
0.05
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of government spending
Effects
of ?scal policy
in Colombia
215
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
what is found for Colombia using other methodologies[7]. Recently, Leigh (2008) found
a short-run output multiplier of 1.15 for Colombia, using a Global Integrated Monetary
and Fiscal Model developed by the International Monetary Fund. In order to contrast
results from Colombia at international level, Table II also presents cumulative output
multipliers estimated with analogous technique for various industrialized
economies[8].
Second, both in?ation and nominal interest rates respond positively and
signi?cantly to a positive shock in government spending, particularly after the
second year after the shock. Nonetheless, the size and time persistence of the interest
rate responses are greater than in?ation, which implies that interest rates, in real terms,
go up with increases in government expenditure. These ?ndings are coherent with the
textbook macroeconomic models and have not usually been found in previous
empirical papers.
Third, the short-run dynamics of net taxes differ substantially from the evolution of
expenditures. In contrast to the spending shock, the response of net taxes to this shock
is virtually zero. This unusual result can be associated with the net tax de?nition[9]
and the increase in the amount of interest payments on debt and pension expenditures,
which are discounted from tax revenues.
Finally, the robustness of our baseline results was checked by means of two
alternative identi?cations (the Blanchard-Perotti approach vs the recursive approach);
analternative de?nition of in?ation(GDP-de?ator insteadof CPI); the use of a third-order
lag polynomial instead of a fourth-order lag polynomial[10]; and through the use of other
elasticity values calculated for Colombia. The results do not change substantially.
b. The effects of net taxes
The response of the various macrovariables to a positive shock in net taxes in our
baseline model is shown in Figure 4. Contrary to what was found in the three-variable
SVAR model, but similar to what it has been found in other countries, the GDP
responds positively. Nonetheless, this unexpected result is short lived and has little
signi?cance. In addition, the response of net revenue to its own shock does not present
a conclusive trend and is insigni?cant over time. In?ation and, especially, short-term
interest rates responded with a signi?cant downward drop to an increase in taxes,
which is consistent with intuition. However, the responses of these variables did not
continue beyond the second year after the shock.
Quarters
4th Q 8th Q 12th Q 16th Q 20th Q
Colombia (our results) 1.12 0.59 1.20 2.11 6.01
International evidence
France (Biau and Girard, 2005) 1.90 1.40 1.50 – –
Germany (Heppke-Falk et al., 2006) 3.56 5.96 8.35 – –
Italy (Giordano et al., 2005) 1.50 1.70 1.20 0.90 –
Spain (De Castro, 2006) 1.14 1.04 0.58 20.05 20.83
USA (Blanchard and Perotti, 2002) 0.45 0.54 1.13 1.29 20.97
Source: Elaborated by the authors
Table II.
Cumulative output
multipliers
to a government
expenditure shock
JFEP
3,3
216
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 4.
Responses
to an increase in net taxes
(baseline model)
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.6
–1.2
–0.8
–0.4
0.0
0.4
0.8
02 04 06 08 10 12 14 16 18 20
Response of 3-monthr ate
–0.010
–0.005
0.000
0.005
0.010
0.015
0.020
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.05
–0.04
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of net taxes
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
0.010
02 04 06 08 10 12 14 16 18 20
Response of GDP
Recursive approach Blanchard-Perotti approach
–0.012
–0.008
–0.004
0.000
0.004
0.008
02 04 06 08 10 12 14 16 18 20
Response of prices
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.010
–0.005
0.000
0.005
0.010
0.015
0.020
02 04 06 08 10 12 14 16 18 20
Response of government spending
–0.05
–0.04
–0.03
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of net taxes
Effects
of ?scal policy
in Colombia
217
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
c. The effects on private consumption and private investment
Figures 5 and 6 show the responses of private consumption and private investment to a
positive shock in both government spending and taxes, respectively. The
impulse-response functions came from the following six-variable VAR model: g
t
, y
t
,
x
t
, p
t
, t
t
, r
t
; where x
t
is the new variable (private consumption or private investment)
that is added in turn to the baseline model. x
t
is placed third in the order which means
that private consumption (or private investment) does not react contemporaneously to
prices, taxes and interest rate shocks but is affected contemporaneously by output and
government spending shocks. However, with this placement in the order, the x
t
component is allowed to affect prices, taxes and interest rate within the same quarter.
The impulse responses show a signi?cant positive response of private consumption
to a spending shock, reproducing the GDP pattern, which is to increase steeply until it
reaches its peak in the ?fth quarter at around 0.65 percent and then decline gradually
to become temporally negative and insigni?cant in the third year. Private investment
presents a similar pattern, even if its response is greater and is shorter lived. Regarding
the tax effect (Figure 6), notice that private investment responds with a signi?cant
downward drop to a positive shock in taxes. After that, it increases gradually to
become transitorily positive but non-signi?cant after the second quarter.
Figure 5.
Responses to an increase
in government spending
Response of private consumption
Note: Using the Blanchard and Perotti identification approach
Response of private investment
0.012
0.12
0.08
0.04
–0.04
–0.08
–0.12
0.00
0.008
0.004
0.000
–0.004
–0.008
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
Figure 6.
Responses to an increase
in net taxes
Response of private consumption
Note: Using the Blanchard and Perotti identification approach
Response of private investment
0.012
0.12
0.08
0.04
–0.04
–0.08
–0.12
0.00
0.008
0.004
0.000
–0.004
–0.008
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
JFEP
3,3
218
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The patterns of response for consumption and investment are in accordance
with previous evidence from VAR analysis. In contrast, the effect on investment
has been less empirically conclusive. From a theoretical point of view, neoclassical
theory broadly predicts that consumption should fall in response to a (temporary)
spending shock while (new) Keynesian models predict that consumption will
increase. Going beyond that, the simple Keynesian theory predicts that the
response of private investment to a tax shock should be the opposite of its response
to a spending shock. The results offered in this section are in line with the latter
approach.
d. The effects of government spending by components
The responses of GDP, private consumption and private investment to a positive
government spending shock are shown by component in Figures 7 and 8. The main
components of government spending include personnel expenditure and other
operating expenditures (public consumption), as well as ?nancial expenditures on
capital formation (public investment). The impulse-response functions emerge from the
following six-variable SVAR model: x
t
, y
t
, z
t
, p
t
, t
t
, r
t
; where x
t
represents the
component of public spending (consumption or investment), and z
t
is added in turn to
capture the private consumption or private investment responses.
We found a strong and signi?cant response of output to a shock to government
investment at the same time as the effect of government consumption is weak and
non-signi?cant[11]. The private investment response after a shock to public
investment is, in general, signi?cant in the short term. Initially, it responds with a
signi?cant upward jump and subsequently declines gradually and becomes temporally
negative before returning to zero after the third year[12].
Figure 7.
Responses to an increase
of public consumption
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
2 4 6 8 10 12 14 16 18 20
Response of public consumption
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.08
–0.04
0.00
0.04
0.08
0.12
2 4 6 8 10 12 14 16 18 20
Response of private investment
Effects
of ?scal policy
in Colombia
219
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
e. The effects of taxes by component
Following the national account criteria, taxes were classi?ed into two components:
consumption and income taxes. To assess the tax effects, a six-variable VAR model
was used (g
t
, y
t
, z
t
, p
t
, x
t
, r
t
) while maintaining the same order as before. In these cases,
x
t
represents the tax component and z
t
is added in turn to capture the private
consumption or private investment responses.
A positive shock to consumption taxes shows little persistence and becomes
non-signi?cant after two quarters (Figure 9). This shock reduces GDP and private
consumption initially, but these responses are non-signi?cant in both cases. Clearly, the
short-terminterest rate reacts negativelyandsigni?cantlytoa shocktoconsumptiontaxes
in line with what one could expect[13]. However, the interest rate response is short lived.
The shock to income taxes (Figure 10) looks a little more persistent than the shock
to consumption taxes. The response of GDP and private consumption are negative and
signi?cant in the ?rst two quarters although these become slightly positive but
non-signi?cant in the medium term. The income tax shocks do appear to have
signi?cant effects on private investment in the short term. Initially, private investment
responds with a signi?cant downward drop and, thereafter, increases gradually to
become transitorily positive (but non-signi?cant) after the third quarter. Once more,
notice that the short-term interest rate reacts negatively to a positive shock to income
taxes and the response is statistically signi?cant for a longer period (seven quarters
after the shock). The fall in interest rates could compensate for the decline in output;
hence its reaction after third quarter.
In summary, shocks to income taxation in Colombia clearly seem to cause more
distortion, mainly by affecting private investment, whereas shocks to consumption
Figure 8.
Responses to an increase
in public investment
–0.10
–0.05
0.00
0.05
0.10
0.15
0.20
2 4 6 8 10 12 14 16 18 20
Response of private investment
–0.12
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
0.20
0.24
2 4 6 8 10 12 14 16 18 20
Response of public investment
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
0.012
2 4 6 8 10 12 14 16 18 20
Response of private consumption
JFEP
3,3
220
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
taxation do not seem to affect activity signi?cantly in the short term. In principle, these
?ndings are in line with what has been found by public ?nance experts who have
generally concluded that not all taxes have the same impact on the economy. VATs that
are basically proportional taxes on consumed income are preferred to personal income
taxes that are oftenappliedwithhighmarginal taxrates onbothconsumptionandsaving.
f. The effects on the labor market
Finally, Figures 11 and 12 show the responses of the labor market to a ?scal shock by
means of six-variable SVAR models. The systems added one variable from the labor
Figure 9.
Responses to an
increase in indirect
(consumption) taxes
–0.006
–0.004
–0.002
0.000
0.002
0.004
0.006
0.008
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.006
–0.004
–0.002
0.000
0.002
0.004
0.006
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
0.20
2 4 6 8 10 12 14 16 18 20
Response of indirect taxes
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
2 4 6 8 10 12 14 16 18 20
Response of 3-month rate
–0.04
–0.02
0.00
0.02
0.04
0.06
0.08
0.10
2 4 6 8 10 12 14 16 18 20
Response of private investment
Effects
of ?scal policy
in Colombia
221
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
market to the ?ve variables of the baseline model, either the unemployment rate, m, or
the real index of minimum wage, w. In the ?rst case, the unemployment rate is de?ned
as the ratio of people classi?ed as unemployed to the total labor force of the seven
major cities (for the 1984Q1-2007Q4 period). In the second, minimum wage is de?ated
using the CPI. Both labor variables are constructed by employing quarterly
information from the BR web site.
The impulse responses are calculated under the recursive identi?cation approach
with the following order: g
t
, y
t
, p
t
, m
t
, t
t
, r
t
, in the ?rst system and g
t
, y
t
, p
t
, t
t
, w
t
, r
t
, in
the second one. This particular order of the variables implies:
Figure 10.
Responses to an increase
in direct (income) taxes
–0.010
–0.005
0.000
0.005
0.010
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–2.5
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
2 4 6 8 10 12 14 16 18 20
Response of 3-month rate
–0.008
–0.004
0.000
0.004
0.008
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.15
–0.10
–0.05
0.00
0.05
0.10
2 4 6 8 10 12 14 16 18 20
Response of private investment
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
2 4 6 8 10 12 14 16 18 20
Response of direct taxes
–0.010
–0.005
0.000
0.005
0.010
2 4 6 8 10 12 14 16 18 20
Response of private consumption
–2.5
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
2 4 6 8 10 12 14 16 18 20
–0.008
–0.004
0.000
0.004
0.008
2 4 6 8 10 12 14 16 18 20
Response of GDP
–0.15
–0.10
–0.05
0.00
0.05
0.10
2 4 6 8 10 12 14 16 18 20
Response of private investment
–0.08
–0.04
0.00
0.04
0.08
0.12
0.16
2 4 6 8 10 12 14 16 18 20
Response of direct taxes JFEP
3,3
222
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
.
As unemployment rate is fourth in the order, it does not react contemporaneously
to taxes and interest rate, but it is affected contemporaneously by government
spending, output and price shocks.
.
The real minimum wage is affected by spending, output, tax and price shocks
contemporaneously, but it does not react to short-term interest rate shocks.
The spending shock has a positive and signi?cant effect on GDP, replicating the
hump-shaped pattern found initially. Both prices and short-term interest rates move
from a negative to a positive response, which becomes signi?cant after the third year.
Figure 11.
Response of
unemployment rate
to a government
spending shock
–0.6
–0.5
–0.4
–0.3
–0.2
–0.1
0.0
0.1
02 04 06 08 10 12 14 16 18 20
Response of unemployment rate
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.02
–0.01
0.00
0.01
0.02
0.03
02 04 06 08 10 12 14 16 18 20
Response of government spending Effects
of ?scal policy
in Colombia
223
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
All these effects are coherent with what was found in the baseline model. However, the
?nding that was novel came from the unemployment responses (Figure 11). The
unemployment rate reacts negatively and signi?cantly to a positive shock in
government spending in line with what one could expect. The unemployment rate falls
on impact by 0.12 percent, then decreases further to reach a maximum effect of
0.31 percent in the fourth quarter and subsequently stabilizes around 0.25 percent. The
effect on the unemployment rate is remarkably persistent over time. Figure 12 shows
that the response of the real wage index to the spending shock is not statistically
Figure 12.
Response of real minimum
wage to a government
spending shock
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
02 04 06 08 10 12 14 16 18 20
–0.008
–0.004
0.000
0.004
0.008
0.012
0.016
02 04 06 08 10 12 14 16 18 20
Response of GDP
–0.008
–0.004
0.000
0.004
0.008
0.012
02 04 06 08 10 12 14 16 18 20
Response of prices
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
2.0
02 04 06 08 10 12 14 16 18 20
Response of 3-month rate
–0.02
–0.01
0.00
0.01
0.02
0.03
0.04
02 04 06 08 10 12 14 16 18 20
Response of government spending
Response of real wage
JFEP
3,3
224
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cant, but GDP, in?ation and the short-term interest rate react in a manner that is
statistically signi?cant just as the previous case does.
Our results are consistent with the ?ndings of Perotti (2007) and Fata´s and Mihov
(2001), etc. who observed that employment increases after a government spending
shock. However, our ?ndings do not support the positive responses of real wages found
by these authors. From the theoretical point of view, the positive response of
employment is consistent especially with the real business cycle theory and Keynesian
models. They typically predict that an increase in government expenditure will increase
the demand for labor, consumption and output. With lump-sum taxation, the
neoclassical model studied by Baxter and King (1993) also predicts a positive effect of a
shock to government spending on total employment even though the real wage declines
under this approach.
5. Conclusions
The short-term macroeconomic effects of the ?scal policy in Colombia for the
1980-2007 period were studied using a SVAR model which has been widely
recommended by several previous empirical papers. Throughout the period of study,
the policymakers for the country undertook a series of far-reaching economic and
political reforms to modernize the state and strengthen their productive activities.
Fiscal policy played a crucial role in attaining these objectives.
Initially, macroeconomic effects of the ?scal shock were assessed by means
of the simplest model (three-variable SVAR). Either an increase in government spending
or a reduction in taxation signi?cantly expand the GDP during the ?rst six quarters;
thereafter, the effects tend to disappear steadily in the medium term. These ?ndings
contrast with the negligible results found previously for Colombia. In contrast, our
benchmarkisa?ve-variableSVARmodel whichaddsCPI-in?ationandshort-termnominal
interest rates to the previous speci?cation. In addition, we speci?ed six-variable VAR
models, adding in turn private consumption, private investment, the unemployment rate
and the real minimum wage to the last set of variables. We also split ?scal revenue and
expenditure by components to distinguish between their particular effects.
Using the baseline model and the others, the following effects of a positive
government spending shock are found. First, the GDP responds positively and
signi?cantly during the ?rst six quarters and shows a typical hump-shaped form with a
peak effect in the third quarter. Subsequently, it declines graduallyand reaches zero after
the four year. The cumulative output multiplier ?uctuates between1.12 and1.19 fromthe
short to mediumterm, whichis broadlyconsistent withwhat is found for Colombia using
alternative methodologies. Second, as expected, both in?ation and nominal interest rates
also respond positively and signi?cantly, particularly after the second year.
Third, the exercises show a signi?cant positive response by both private
consumption and private investment although the response of the later is greater and
shorter lived. Fourth, with regard to the effects of public spending shown by component,
the strong and signi?cant response of output and private investment to a public
investment shock, at the same time that the effect of government consumption is weak
and non-signi?cant is remarkable. And ?fth, the unemployment rate reacts negatively
and signi?cantly to a positive shock in government spending.
Macroeconomic effects of a positive net tax shock are less conclusive as whole
although we ?nd some remarkable results. In one case, in?ation and especially
Effects
of ?scal policy
in Colombia
225
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
short-term nominal interest rates respond with a signi?cant downward drop, which
could be consistent with intuition. However, the responses of these variables do not
continue beyond the second year. In the other, shocks to income taxation in Colombia
clearly seem to cause more distortion, because they mainly affect private investment,
whereas shocks to consumption taxation do not seem to affect real activities
signi?cantly in the short term.
Two alternative identi?cation techniques and other tools are used in the VARs to
check the robustness of our results. In general terms, the results found in this study are
consistent with real business cycle theory and Keynesian models of both traditional
partial equilibrium and new general equilibrium types. They typically predict that an
increase in government expenditures will increase output, consumption, employment
and real interest rates. Our empirical ?ndings do not support the neoclassical
prescriptions according to which increases in government expenditure crowd out the
private sector through the induced wealth effect. From a policy perspective, our results
support the smoothing role of ?scal policy on output ?uctuations, which implies its
capacity to restore real activity effectively in critical times like the ones currently being
forecast. However, it is essential that ?scal stimulus not be seen by markets as
seriously calling into question medium-term ?scal sustainability. The sustainability
analysis is beyond the scope of this paper and should be the subject of further research.
Notes
1. This assumption also implies that movements in government expenditures, unlike
movements in taxes, are largely unrelated to the business cycle.
2. Placing y
t
and p
t
before t
t
can be justi?ed on the grounds that the output and in?ation
shocks have an instantaneous effect on the tax base and, therefore, a contemporaneous
impact on tax revenues.
3. The placement of the interest rate in the order is justi?ed on two counts. One is on the
grounds of the central bank reaction function, since the interest rate is set as a function of the
output gap and in?ation, and the other is because spending and revenue as de?ned here
(net of interest payments) are not sensitive to interest rate changes.
4. The income tax to GDP elasticity is, on average: 1.3 for OECD; 1.5 for Euro area; 1.1 for new
EU members; 1.85 for the USA; and 0.95 for Germany. These values are taken from Biau and
Girard (2005), Caldara and Kamps (2008) and Heppke-Falk et al. (2006).
5. The expansionary cycle of private consumption and investment recorded in the early 1990s
was fostered by large capital in?ows and ample domestic credit (panels B and C).
6. The social security (pensions and health) and judicial systems in particular demanded
increasing public resources after the new constitution.
7. The cumulative multiplier is calculated as the ratio of the cumulative response of GDP and
the cumulative response of government expenditure for each quarter. Notice that cumulative
multipliers are lower than unit only for the second year when the response function becomes
temporally negative.
8. Notice that even though the GDP response to an increase in public spending is small with
respect to macroeconomic textbook’s ?ndings, it should keep in mind that the latter analysis
does not take into account the role of the other macroeconomic variables (like interest rates,
in?ation, etc.).
9. The net tax revenue results from the difference between gross tax revenue minus interest
payments on public debt minus pension expenditures.
JFEP
3,3
226
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
10. The third- and fourth-order lag polynomial is coherent with the optimal criterion of Akaike
(AIC), Schwarz (SC) and Hannan-Quinn (HQC).
11. The cumulative output multiplier to shocks on public investment, which is estimated in the
simplest model as the ratio of the cumulative response of GDP and the cumulative response
of government investment, rises to 3.64 in the fourth year.
12. The signi?cant response of output to a change in public investment was found previously for
Colombia by Leigh (2008), who concluded that a permanent cut of 0.5 percent points of GDP
in public investment leads to a long-term contraction of 0.86 percent of GDP due to a
contraction in the supply capacity of the economy. However, if public investment is not
productive, a cut in capital expenditures has an effect that is broadly similar to a reduction in
public consumption (GDP declines 0.58 percent).
13. As already mentioned, the shock reduces private consumption and aggregate demand and,
for a given short-term aggregate supply, it also reduces prices and interest rates.
References
Baxter, M. and King, R.G. (1993), “Fiscal policy in general equilibrium”, American Economic
Review, Vol. 83, pp. 315-34.
Biau, O. and Girard, E. (2005), “Politique budge´taire et dynamique e´conomique en France:
l’approche VAR structurel”, Economie et Pre´vision, Nos 169-171, pp. 1-23.
Blanchard, O. and Perotti, R. (2002), “An empirical characterization of the dynamic effects of
changes in government spending and taxes on output”, Quarterly Journal of Economics,
Vol. 107.
Caldara, D. and Kamps, C. (2008), “What are the effects of ?scal shocks? A VAR-based
comparative analysis”, Working Paper Series No. 877, European Central Bank, Frankfurt
am Main.
Cerda, R., Gonza´lez, H. and Lagos, L. (2005), “Is ?scal policy effective? Evidence for an emerging
economy: Chile 1833-2000”, Documento de Trabajo IE-PUC No. 292, Ponti?cia Universidad
Catolica de Chile, Santiago.
De Castro, F. (2006), “The macroeconomic effects of ?scal policy in Spain”, Documento de
Trabajo, No. 0311, Banco de Espan˜a, Madrid.
Fata´s, A. and Mihov, I. (2001), “The effects of ?scal policy on consumption and employment:
theory and evidence”, mimeo, INSEAD, New York, NY.
Gal? ´, J., Lo´pez-Salido, J. and Valle´s, J. (2006), “Understanding the effects of government spending
on consumption”, Journal of the European Economic Association, Vol. 5 No. 1.
Giordano, R., Momigliano, S., Neri, S. and Perotti, R. (2005), “The effects on the economy of
shocks to different government expenditures items: estimates with a SVAR model”, paper
presented at the 7th Workshop on Public Finance, Banca d’Italia, Rome.
Heppke-Falk, K., Tenhofen, J. and Wolff, G. (2006), “The macroeconomic consequences of
exogenous ?scal policy shocks in Germany: a disaggregated SVAR analysis”, Discussion
Paper No. 41, Deutsche Bundesbank, Frankfurt am Main.
Leigh, D (2008), “Achiving a solft landing: the role ?scal policy”, IMF Working Paper No. 0869.
Lozano, I. and Toro, J. (2007), “Fiscal policy throughout the cycle: the Colombian experience”,
Borradores de Econom? ´a No. 434, Banco de la Repu´blica, Bogota´.
Marcellino, M. (2002), “Some stylized facts on non-systematic ?scal policy in the Euro area”,
Discussion Paper No. 3635, Center for Economic Policy Research, Washington, DC.
Effects
of ?scal policy
in Colombia
227
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Perotti, R. (2004), “Estimating the effects of ?scal policy in OECD countries”, mimeo, Bocconi
University, Milan.
Perotti, R. (2005), “Public investment: another (different) look”, mimeo, Bocconi University,
Milan.
Perotti, R. (2007), “In search of the transmission mechanism of ?scal policy”, NBER Working
Papers No. 13143, National Bureau of Economic Research, New York, NY.
Restrepo, J. and Rinco´n, H. (2006), “Identifying ?scal policy shocks in Chile and Colombia”,
Borradores de Econom? ´a No. 397, Banco de la Repu´blica, Bogota´.
S
?
tikova´, R. (2006), “Effects of ?scal policy in the Czech Republic: a SVAR analysis”, Mimeo,
Alternative Approaches to Economic Modelling, IES, Charles University, Prague.
Further reading
Aschauer, D. (1989), “Is public expenditure productive?”, Journal of Monetary Economics, Vol. 23,
pp. 177-200.
About the authors
Ignacio Lozano is a Researcher in the Economics Research Department, Banco de la Repu´blica
(the Central Bank of Colombia). Ignacio Lozano is the corresponding author and can be contacted
at: [email protected]
Karen Rodr? ´guez is a Public Finance Specialist in the Economics Research Department,
Banco de la Repu´blica.
JFEP
3,3
228
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_410619293.pdf