Transcript Document

6th PEP General Meeting
Lima, Peru, June 9-16, 2007
Tax Changes in Argentina:
A General Equilibrium Analysis
Martín Cicowiez, Javier Alejo,
Luciano Di Gresia, Sergio Olivieri
and Ana Pacheco
Centro de Estudios Distributivos, Laborales y Sociales (CEDLAS)
www.depeco.econo.unlp.edu.ar/cedlas
Outline
•
•
•
•
Introduction
Data
Methodology
Results
Real GDP 1993-2006
(mill. LCU 1993 prices)
340,000
320,000
300,000
280,000
260,000
240,000
220,000
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
200,000
Motivation
• The 2001/02 crisis implied a fall in the GDP of more than
15%. The economy has strongly recovered since then,
reaching levels of activity similar to those in the 1990s.
• The exports and financial transactions taxes that were
introduced during the 2001/02 crises are considered as the
most distortive in the Argentine tax system by many
observers.
• Our objective is to evaluate some of the tax changes that
are being proposed by different observers.
• In this preliminary report we are not trying to “optimize”
the Argentine tax structure.
Poverty 1974-2006
Share of population below the official
moderate poverty line (%)
70
60
50
40
30
20
10
GBA
06 *
04
02
00
98
96
94
92
90
88
86
84
82
80
78
76
74
0
Argentina
Source: Own estimates from microdata of the EPH (<www.depeco.econo.unlp.edu.ar/cedlas>)
Inequality 1974-2005
Gini coefficient household per capita
income
0.55
0.50
0.45
0.40
0.35
GBA
05
04
03
02
01
00
99
98
97
96
95
94
93
92
91
90
89
88
87
86
85
84
83
82
81
80
79
78
77
76
75
74
0.30
Arg
Source: Own estimates from microdata of the EPH (<www.depeco.econo.unlp.edu.ar/cedlas>)
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
Tax Revenue / GDP (%)
1991-2006
29
27
25
23
21
19
17
15
Structure of Tax Revenue
Tax instrument
%TREV
%GDP
2.7
Tariffs
8.3
Export taxes
24.5
Value added tax
6.3
Financial transactions tax
9.2
Turnover tax
4.0
Fuel tax
6.9
Taxes on products
26.5
Income tax
11.5
Factor tax
100.0
Total
Source: Argentina SAM 2005.
0.8
2.4
7.1
1.8
2.7
1.2
2.0
7.7
3.3
28.9
Proposed Tax Changes
• Recently proposed tax changes by different
observers as academics, businessmen, private
organizations, and policy makers:
–
–
–
–
reduction in the export tax;
reduction in the financial transactions tax;
reduction in the value added tax on food and beverages;
reduction in the sub-national turnover tax on agriculture
and manufactures;
– increase in the income tax;
Why Dynamic CGEMicrosimulation?
• The (sequential) dynamic CGE model
captures the economy-wide and growth
effects of tax changes.
• The microsimulation model, in turn, allows
for an assessment of the poverty and
inequality impact of tax changes.
Data Sources: CGE Model
• A 2005 SAM was built for this project,
– based on 1997 IO tables (latest available) but updated to 2005 using
different data sources (national account, government budget data, among
others)
– three labor categories: unskilled; semi-skilled; and skilled
– 25 activities and commodities: 4 agr; 13 mnf; and 8 svc
– nine different tax instruments
– one household
• The SAM is combined with other data in order to calibrate the CGE
model,
–
–
–
–
number of workers in each sector
elasticities -- sensitivity analysis
growth projections for GDP, population, and labor force by skill level
unemployment by skill level
Data Sources: Microsimulation
Model
• We use the Encuesta Permanente de Hogares
(EPH), the main household survey in Argentina.
• The EPH gathers information on individual
sociodemographic characteristics, employment
status, hours of work, wages, incomes, type of job,
education, and migration status.
• There is no alternative to the use of the (urban)
EPH. No attempt was made to reconcile the
household survey data with the national accounts.
We focus on labor income.
The CGE Model
• Recursive dynamic but can be solved in multi-pass or onepass -- possibility of adding forward-looking behavior.
• Labor is perfectly mobile between sectors while capital is
sector-specific.
• Endogenous labor supply and unemployment with a
downward rigid real wage for each type of labor. The
minimum real wage for each skill level varies with the
following determinants (van der Mensbrugghe, 2005):
– the unemployment rate as in a wage curve;
– real living standards captured by the household consumption per
capita; and
– average real factor returns (price of value added).
The CGE Model
• Stone-Geary utility function that allows for commodityspecific income elasticities; labor-leisure choice.
• Commodities are differentiated according to their country
of origin (Armington, 1969).
• Quantitative restrictions on exports of agri-food products
that generate quota rents.
• The behavior of the government income and expenditure
depends on the selected closure rule.
• The behavior of aggregate real investment depends on the
selected closure rule for the savings-investment balance.
The CGE Model
• Between periods updates in: sectoral capital stocks; labor
force by skill level; minimum consumption; transfers
between institutions.
• Between periods the new capital (i.e., investment) is
distributed between sectors according to their relative rate
of return for capital and initial share in total capital stock.
• Transfers between institutions grow exogenously at the
same rate as GDP in the baseline scenario.
• Differences between the simulation scenarios (i.e., policyinfluenced) to the baseline are interpreted as the economywide impact of tax changes.
The CGE Model: Value Added Tax
• The modeling of the value added tax incorporates
rebates for intermediate inputs and investment
purchases (Go et al., 2005). With this treatment,
there is no cascading effect on prices of taxes on
intermediate goods.
• All transactions are taxed at a fixed proportional
rate regardless of whether they are final or
intermediate transactions. Firms can deduct taxes
paid on intermediate inputs.
• Import sales are subject to a VAT while export
sales are not.
CGE Model: Production Side
EXPORTS
DOMESTIC
CET
PRODUCTION
LF
VALUE
ADDED
INTERMEDIATE
INPUTS
CES
LF
CAPITAL
LABOR
TYPE 1
LABOR
ARMINGTON 1
CES
CES
LABOR
TYPE 3
DOMESTIC 1
ARMINGTON N
IMPORTED 1
CGE Model: Consumption Side
UTILITY
STONE-GEARY
LEISURE
ARMINGTON 1
ARMINGTON N
CES
DOMESTIC 1
CES
IMPORTED 1
DOMESTIC N
IMPORTED N
The Microsimulation Model
•
•
At the microsimulation level we produce a counterfactual
household income distribution for each time period of the
simulation.
The microsimulation involves, in each time period, the
following eight steps:
1.
2.
3.
4.
5.
6.
7.
8.
households reweighing to reflect population growth;
labor supply adjustment;
unemployment rate adjustment;
sectoral employment change;
relative wage changes;
average wage changes;
changes in the skill composition of the labor force; and
price changes.
The Microsimulation Model
• Every step of the microsimulation produces a
counterfactual labor income for each worker. This new
labor income is used to compute a counterfactual
household income.
• We use econometric estimations to compute counterfactual
wages for those individuals that change their
characteristics in a simulation.
• The last step involves a change in the poverty line in order
to reflect the price movement of different goods.
• The non-labor income is considered constant throughout
the whole simulation period.
Macro-Micro Interaction
• The two methodologies are used in a sequential
fashion (i.e., top-down approach).
• The link between both modeling stages is made
trough the mapping of changes in wages and
employment, and product prices from the CGE to
the microsimulation.
• We do not need to assure complete consistency
between the data sets used at the two modeling
stages. Only deviations from the benchmark are
transmitted from the CGE model to the
microsimulation model.
Scenarios
• 1. BASE, business-as-usual scenario; reflects the evolution of the
economy in the absence of shocks.
• 2. RETENC, 20% yearly reduction starting in year 2008 of the export
tax that was introduced during the 2001/02 economic crisis.
• 3. IVAALI, 55% reduction in 2008 of the VAT on food; has been
proposed to compensate the increase in the domestic price of food that
is expected as a consequence of RETENC.
• 4. FINANC, 15% in 2008, 20% in 2009, and 25% in 2010 reduction in
the financial transactions tax that was also introduced during the
2001/02 economic crisis.
• 5. GANANC1-2, in order to compensate the loss of tax revenues
induced by the previous tax changes, some analyst have proposed an
increase in the income tax in 2008; two variants: 1) 15%; and 2) 20%.
Scenarios
• 6. PAQUET1-2, combines all the previous tax changes;
two variants: 1) 15% increase in the income tax; and 2)
20% increase in the income tax.
• 7. PAQUET1-CHGCLOS, in order to test the sensitivity of
some results to the selected closure rule, we run the
previous scenario but fixing the government consumption
and flexing the ratio between government savings and
GDP at market prices.
• 8. TOT, 2.5% yearly decrease in the price of exports
starting in 2008.
• 9. PAQUET1-TOT, combines scenarios TOT and
PAQUET1 in order to analyze if the proposed tax changes
are useful to isolate the economy from external shocks.
Closure Rule Baseline
• Calibration run (i.e., endogenous TFP and exogenous
GDP) to impose a GDP growth rate in the baseline
scenario.
• Fix shares of government consumption, investment, and
private consumption in absorption.
• Fix foreign savings and flexible real exchange rate to
equilibrate the current account of the BOP.
• Flexible government savings.
• All tax rates are exogenously imposed.
• Quantitative restrictions on exports of agri-food products
for the period 2006-2007.
Closure Rule Simulations
(changes w.r.t. baseline)
• Exogenous TFP and endogenous GDP.
• Flexible government consumption.
• Fix (at BaU values) ratio between government
savings and GDP at market prices to assure fiscal
solvency -- sensitivity analysis.
• Flexible investment and fix (at BaU values)
savings rates for domestic non-government
institutions.
• Fix foreign savings and flexible real exchange rate
to equilibrate the current account of the BOP.
Results: Real GDP Market Prices
(LCU)
REAL GDP MARKET PRICES
12000
11000
base
gananc2
retenc
paquet1
10000
9000
8000
7000
6000
5000
2005200620072008200920102011201220132014201520162017201820192020
Results: Unemployment Rate (%)
UNEMPLOYMENT RATE
16
15
base
retenc
paquet1
tot
14
13
12
11
10
9
2005200620072008200920102011201220132014201520162017201820192020
Results: Terms Of Trade Shock
Real GDP Market Prices (LCU)
REAL GDP MARKET PRICES
12000
11000
base
retenc
tot
retenc-tot
10000
9000
8000
7000
6000
5000
2005200620072008200920102011201220132014201520162017201820192020
Results: Consumer Price of Food
PQDFOOD
1.1
1.095
base
ivaali
retenc
paquet1
1.09
1.085
1.08
1.075
1.07
1.065
1.06
2005200620072008200920102011201220132014201520162017201820192020
Results: Tax Revenue / GDP (%)
31
29
27
25
23
21
19
17
obs
base
paquet1
2019
2017
2015
2013
2011
2009
2007
2005
2003
2001
1999
1997
1995
1993
1991
15
Results: Government Consumption /
GDP (%)
15.0
14.5
14.0
13.5
13.0
12.5
12.0
11.5
11.0
10.5
obs
base
paquete1
paquet2
2019
2017
2015
2013
2011
2009
2007
2005
2003
2001
1999
1997
10.0
paquete1-chgclos
Sensitivity Analysis w.r.t.
Government Closure Rule
Government Savings / GDP (%)
8
6
4
2
-4
-6
-8
obs
base
paquet1
paquet1-chgclos
2019
2017
2015
2013
2011
2009
2007
2005
2003
2001
1999
-2
1997
0
Results: Poverty
(poverty incidence moderate PL)
38
36
34
32
30
28
26
24
22
base
paquet1
tot
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
20
Results
• The reduction of distortive taxes generates
efficiency gains that translate into a higher GDP.
• As expected, the reduction in the export tax
increases exports and domestic prices.
Specifically, the price of food increases relative to
the baseline having a negative impact on poverty
by moving the poverty line.
• The reduction in the VAT for food products
compensates the increase in domestic prices
generated by the reduction in export taxes.
Results
• The closure rule that we are using for our
simulations generates a decrease in government
consumption as a consequence of the decrease in
tax revenue. However, the ratio of tax revenue to
GDP reaches historical levels.
• When the government closure rule is changed, the
proposed tax changes (see PAQUET1) generate a
decrease in fiscal solvency and a crowding out of
private investment that hurts economic growth.
Results
• Poverty decreases in the PAQUET1
scenario mainly as consequence of the
decrease in unemployment and the increase
in the average wage.
• Notice that we do not consider the positive
effect that government consumption may
have on productivity (e.g., by increasing
infrastructure).
Sensitivity Analysis w.r.t. Elasticities
Values
• The methodology used is based on Harrison and
Vinod (1992). The procedure can be summarized
in the following steps:
–
–
–
–
–
i) randomly select a set of parameters;
ii) solve the model;
iii) store the results;
iv) repeat steps (i) – (iii) several times; and
v) construct confidence intervals for some results.
• For this preliminary report, the model was solved
only 40 times.
Sensitivity Analysis w.r.t. Elasticities
Values
Unemployment Rate
.3
BASE
PAQUET1
kdensity value
.25
.2
.15
.1
.05
0
8
9
10
11
12
UERAT
13
14
15
16