Escolar Documentos
Profissional Documentos
Cultura Documentos
†
Wilfredo Leiva Maldonado
‡
Octávio Augusto Fontes Tourinho
Marcos Valli§
(Preliminary version)
*
The analysis, interpretation, conclusions and opinions contained in this document are
exclusively those of the authors, and do not represent those of the institutions to which they are
or were associated.
†
Of Universidade Católica de Brasília. This author thanks the financial support of for this
research from CNPq, through scholarships PQ 305317/2003-2 and Edital Universal no.
471899/2003-8, and of IPEA, BID and PNUD, through project RedeIPEA no. 31F19/J-CBR
4339/2005.
‡
Of Banco Nacional do Desenvolvimento Econômico e Social. The participation of this author
in this research occurred under the technical cooperation program between BNDES and IPEA,
for years 2006 and 2007.
§
Of Banco Central do Brasil
1 INTRODUCTION
Computable general equilibrium models (CGEs) have been used in several countries to
analyze the costs and benefits of a varied range of public policies, especially those of a
fiscal nature and those directed towards foreign trade, income distribution and
stimulation of economic growth. A systematized account of the main studies can be
found, for example, in the two main surveys published more recently: Gunning and
Keyzer (1995), and Ginsburgh and Keyzer (1997).
In general, these studies have used the comparative static methodology to address
the themes listed above. That approach calculates the impact of economic policy
measures on the stationary state of the economy, leaving aside the analysis of the
transitional dynamics between the current and the new stationary states.
To address the impacts on economic growth it is frequently necessary to use
dynamic versions of the CGEs, which simulate the economic equilibrium for several
successive moments in time, which are linked recursively though the trajectories of the
state variables, that usually are the stocks of the several types of capital and labor
considered in the model. However, it is not indispensable to use a dynamic model for
that purpose because, as shown in Rutherford e Tarr (2005), those issues can be
addressed in the realm of static models, as long as certain conditions are met and
attention to certain details is given throughout the analysis.
Regardless of which type of model is used, however, is of fundamental importance
that the equations of the model reflect the process of physical capital accumulation as
realistically as possible for the model to be able to simulate correctly the workings of
the real economy. It is in this respect that this paper proposes an advance with respect to
conventional CGE’s, by recognizing the bidirectional interaction between the allocation
of physical and financial capital in a modern economy. We show that this effect can be
addressed in the realm of a computable general equilibrium model of the Brazilian
economy which, extended to incorporate the financial system, provides an adequate
framework to treat it endogenously.
The inclusion of financial mechanisms in general equilibrium models has already
been discussed in other papers that studied the economic effects of the globalization of
international capital flows. These are briefly discussed below.
The impact of globalization on economic growth of developing countries was
studied by Prasad, Rogoff, Wei and Kose (2004), exploring, in particular, the effects of
a large and increasing private foreign capital inflow. They conclude for the absence of
strong evidence that it has affected the growth rate of the majority of those countries.
However, they find that countries that had: i) solid financial institutions; ii) good quality
Government, that is politically stable and honors contracts; iii) macroeconomic stability;
iv) fiscal discipline and, v) flexible exchange rates, display a large positive impact of
foreign capital flow on growth. Kose, Prasad and Terrones (2003) show that in the in
the 90´s private consumption in Americas developing countries did not accompany the
increase in global financial integration, as could have been expected if the increased
financial flow had reduced, through diversification, their macroeconomic risk.
Roland-Holst (2001) advances these efforts to model the financial transmission
mechanism, and disaggregates them by type and suggests several alternative
formulations for each one of them, especially for foreign direct investment and reserves
2
flow, which he considers to be the structural elements. Goodhart, Sunirand and
Tsomocos (2004, 2005) emphasize more the modeling of international financial flows
and its effects, and disaggregate net investments by type of investor, discriminating
banks, firms and the rest of the world. In addition, they take into account the possibility
of default, and show that it can generate financial fragility.
Taking the suggestions contained in these papers into account, and extending them,
Maldonado, Tourinho and Valli (2007) propose to treat foreign capital flows in a CGE
model for Brazil as endogenous, using a function linking it to the expected rate of loss
of foreign reserves which is based on empirical evidence for that country over recent
years. To quantify the importance of adding this feature to the model, they simulate the
impacts of the country joining free trade agreements, and compare them with the
response of the model with exogenous capital flows. They conclude that the impacts in
the extended model is significantly larger, lending support that the hypothesis that the
existence of the financial flow amplifies the impact of policies. That feature is a linkage
between the financial and the real sector of the economy, to the extent that foreign
investment is part of the country’s capital formation.
The objective of the research reported in this paper is to extend this line of research
and model the workings of the financial market in the Brazilian economy, is such a way
that it can be included in a larger general equilibrium model, and to determine the
importance of this extension by simulating shocks to financial and fiscal parameters.
We are specifically interested in identifying the real effects of changes in interest
rates. This issue is of special interest in the Brazilian case because of the preponderant
role of the State in the economy, which reinforces that interaction for two reasons. First,
because it is the main issuer of net debt in the country, it regulates the volume of credit
and determines the floor of the basic interest rate, which is the rate of return of public
bonds. Second, it has a large impact in the financing of capital formation in the country,
through the operational policy of public banks that subsidize credit.
To take into account the role of financial capital of firms and individuals, the model
recognizes explicitly that variations in these stocks imply endogenous capital flows that
can both affect the allocation of investment and be affected by it, interacts with other
endogenous variables, and is affected by the model’s parameters. At the aggregation
level employed by CGE models, this implies endogenous capital flows between the
sectors of the economy, and between the country and the rest of the world, which
condition investment and the physical allocation of sector capital and labor and,
therefore, economic growth.
The model, enriched by the financial flows, serves as an instrument to design
strategies to reduce the very high real interest rate of the Brazilian economy, and answer
at least tentatively that recurring issue in academic debate and current economic policy
in Brazil. In addition, it also allows us to calculate the capital movements that occur
when fiscal or monetary policy change, and thereby anticipate problems that may be
produced by their implementation.
The analysis here is developed in three stages. Initially we describe the technology
of the financial markets to collect and distribute savings, intermediating capital in the
economy. This implies relationships between the stocks of financial and physical capital
and their rates of return for the several agents. Next, functions representing these
relations are introduced in a Computable Equilibrium Model (CGE) for the Brazilian
3
economy which had previously been developed, 1 to extend it in the sense of creating a
transmission channel between the financial and productive sectors. Finally, we simulate
some fiscal and monetary policy shocks, and analyze their impacts to evaluate the extent
to which that model of the channel was satisfactory.
The main conclusion is that the existence of a developed capital intermediation
sector, like the one in the Brazilian economy, substantially increases the real impact of
some economic policies. The policies whose impacts are most affected are those related
to the credit supply and to the level of the interest rates.
The paper is composed of this introduction and 4 sections, described below. Section
2 describes how the financial intermediation sector relates to the other entities which are
usually represented in a CGE model, showing how we built the Financial Social
Accounting Matrix (SAM-F). Section 3 describes the model, emphasizing how the
financial market was modeled through a capital intermediation technology, and the
manner in which its supply and demand of capital were incorporated in the model as a
whole. Section 4 shows some simulation experiments. The last session summarizes the
main conclusions.
1
The CGE model is an extension of the model built initially by Sherman Robinson in a consulting conrtract for BNDES in
1996. Its formulation is similar to the model in Dervis, Melo and Robinson (1982), Devarajan, Lewis and Robinson
(1991), and Robinson et al. (1999).
4
structure and components, as well as a description of how to construct it from the
Brazilian national accounts. That paper also presents the SAM for 2003 in the usual
format, without the financial system accounts.
The productive sectors are consolidated in only two categories: the financial and
non-financial enterprises. This large degree of aggregation is one order of magnitude
larger than the one which is usually adopted in CGE models for Brazil, that contemplate
about 40 sectors, and its adoption can be seen as an effort to develop a simple prototype
model of how to incorporate the financial markets in a conventional CGE model. It was
also adopted to allow the recovery of information regarding the flows in the financial
markets, for which detailed sector information does not exist.
The financial sector, as defined in the Brazilian national accounts (IBGE (2004)), is
formed by financial enterprises and institutions that dedicate primarily to financial
intermediation, or to ancillary financial activities strictly linked to it. On the other hand,
the non-financial sector is the set of private and public enterprises that produce
mercantile goods and services.
The two stages of the process of bringing the goods and services to consumers are
represented in the following way: production proper is represented by the activities,
while commercialization (in a broad sense) is represented by the markets, which is a
separate entity in the SAM.
The agent we denominate “families” in the SAM-F, and also in the CGE, includes
not only the families proper, but also the agricultural and cattle-related establishments,
micro-enterprises, the autonomous labor, the rent of owner-occupied residential real
estate, non-mercantile private services (domestic services and non-profit enterprises).
This aggregation requires, no doubt, heroic assumptions, but they are no less dramatic
then those required to consolidate all non-financial enterprises in a single group.
There are also two other agents: government and the rest of the world. The
government produces public goods at the activities at the federal, state and municipal
levels, and corresponds to the public administration sector in the national accounts. Its
main income source is the tax revenue. The rest of the world buys and sells goods and
services from the markets, and also invests in the country.
For each entity listed above there is a line and a column in the SAM-F. Throughout
a column for an entity, are registered all the payments made by it to each entity
(including itself) indicated in the corresponding line. The value of the flow is found in
the intersection of column and the line and, in the real side of the economy, corresponds
to a flow in opposite direction of goods and services. Over a line corresponding to a
given entity, therefore, all payments received by it are found. The equilibrium between
income and expenditure of each agent requires that the total of each column be equal to
the total of the respective line.
The SAM 2003 in Silva, Tourinho e Alves (2006) is consolidated, by summing
columns and rows, to reproduce the institutional framework described above, and yield
the cells in the conventional part of the SAM-F in Table 6.
5
PROPERTY INCOME AND CAPITAL INTERMEDIATION
2
Interest is the remuneration of time deposits, fiduciary titles, and the loans, and their value is contractually established
as a percentage of the capital involved.
3
Or SAM-F admits, por simplicity, that all the dividens distributed to families are transfered to them indirectly, througn
the intermedited capital fund.
4
In the case of simultaneous occurrence of payments and receipts of an agent to/from itself, only the net value is
represented in the matrix.
6
7
The SAM-F also contains lines and columns that represent the accounting of the
agents’ net assets, and show how they convert the result of current account transactions
into changes in the net asset position in the intermediation fund. This represents, in
particular, the financing of the current account deficits of the government and of the rest
of the world by changes in their exposure in the fund.
Table 6 presents the SAM-F for Brazil, in 2003, used in this study.
5
The appendix is available under request to the authors.
6
CES means Constant Elasticity of Substitution.
7
CET means Constant Elasticity of Transformation ,.
8
supply of foreign capital can be seen in this context as the model of the financial
decisions of agent rest of the world. They are discussed below.
LENDING (CET)
Ai Ei
9
and that the technology in the lending of the funds is CET, i.e. the elasticity of
transformation in placement of the funds is also constant, but not necessarily equal to
one. This specification is suggested by the empirical analysis of the time series of the
shares of the different agents in the fund’s borrowing and lending operations. are
displayed in equation (1).
1/ ρ
5
⎛ 5 ⎞
∏ Ai ≥ X ≥ ⎜ ∑ β i Eiρ ⎟
ϕi
(1)
i =1 ⎝ i =1 ⎠
The first order conditions of the restricted optimization in (2) imply the existence
of a Lagrange multiplier λ > 0 such that equations (3) are satisfied:
rA,i Ai = ϕi λ X , i = 1,...,5 (3)
The solution of system of 6 equations composed of the FOC and the production
function yields the fund’s demand for deposits Ai and its minimum cost c( X ) = λX as
a function of rA,i and X . In the second stage the fund maximizes its profit, as indicated
in (4).
⎧ 5
Maximize
⎪⎪ (Ei ), X ∑ =
rE ,i Ei − c( X )
(II) ⎨ i 1 (4)
( )
1/ ρ
⎪Subject to
∑ i=1 βi Ei ≤ X
ρ
5
⎪⎩
The Lagrange multiplier is the same in problems (I) and (II). The FOC of (4) are
given in (5).
ρ −1
⎛E ⎞
rE ,i = λβ i ⎜ i ⎟ , i = 1,...,5 (5)
⎝X⎠
The solution of the system of 6 equations composed of 5 FOC plus the restriction in
(4) yields the optimal level of financial intermediation X ∗ and the optimal level of
10
loans from the fund to each representative agent, Ei∗ . Back substitution of X ∗ into the
FOC (3) yields the optimal borrowing of the fund from each agent: Ai∗ .
∑ Ei = ∑ Ai
i =1 i =1
(9)
The budget equilibrium of the agent “rest of the world” determines the supply of
foreign capital. Because it is part of channel of interaction between the real and
financial parts of the economic system, it also receives specialized treatment. That flow
is endogenous, and is determined according to the formulation in Maldonado, Tourinho
e Valli (2007). It is, on average, an increasing function of the rate of change of the
country’s international reserves.
The main empirical hypothesis is the existence of a stable relation between the flow
of foreign capitals (FCAP) and the expected loss of the country’s international reserves
( RLSFRES e ) in “normal” periods, when the balance of payments is not in a crisis
situation, as suggested by Figure 1, and indicated in equation (10).
8
In our formulation the fund can also lend these resources to the agent itself: this constitutes reinvestment.
11
⎛ FRES te ⎞
sav Row − LSFRES = FCAPt = f ( RLSFRES te ) = f ⎜⎜1 − ⎟
⎟ (10)
⎝ FRES t −1 ⎠
where FRES is the level of international reserves, and the superscript “e” indicates the
expected value of the corresponding variable.9 The function f is represented in
parametric form adopting a flexible functional form whose inclination, concavity and
limiting derivatives approximate the format of the empirical relation.
9
Expected value is used here in a broad meaning, no in a strict sense of mathematical expectation.
10
Limits to the evolution of the interest rate on public debt that extrapolate the equilibrium context desctibed in the text
are, for example, the need to control inflation and the stability in the level of activity
12
rE , gov = rE , gov (14)
4 EXPERIMENTS
The objective of the calibration of the model’s parameters is to make its solution
reproduce the data of the base year and, therefore, the equilibrium of the economy. This
involves the calibration of the conventional part of the CGE-F, which was made using
the data in the SAM-F of Table 6 in the usual manner, described in Robinson et al
(1999) and of the financial part, which is done as described in Appendix A.11
However, it is important to note that it is not sufficient that the basic scenario be a
static equilibrium, but it is also necessary that it also be a dynamic equilibrium with
rational expectations, because in equation (10) the flow of foreign capital depends on
the expected rate of increase of reserves. Since the value of that variable in 2003 is not
the long term equilibrium value, we built the basic scenario is a virtual equilibrium of
the model, were that variable it is set to the value it would have under rational
expectations.
In this section we report on some experiments. The first two show the importance of
the extension proposed here by comparing the response of the model to tariff reductions
that mimic the adhesion of the country to international free trade agreements. The other
three experiments reduce key interest rates to simulate public policy. The first reduces
the short term rate paid by the Government on its debt, the second reduces the rate paid
by non-financial institutions on their debt, and the third reduces the interest rate paid by
foreign institutions on loans extended to them by the domestic capital market. The
results allows us to evaluate the power of the transmission channel between the real and
the financial sectors of the economy.
11
The appendix is available under request to the authors.
13
The simulation of the implementation of free trade agreements in models that have a
financial sector produces larger economic impacts than in those that do not include that
sector. They are better, because it is believed that they are closer to the real world,
where the dynamic effects of the agreements are more important than their direct effects
on country’s flow of external commerce.
The simulation of changes in the reference interest rates of the financial market
show even more bluntly the usefulness of the extension we propose here, because they
not even can be made in CGEs that do not include a model of the financial
intermediation mechanism.
In a broad sense the effects on the real variables in all simulations are relatively
small, but this is a consequence of the nature of CGE models that, as a rule, assume full
employment of factors, as ours does. However, the impacts are useful to identify their
sign and relative magnitude, when comparing different simulations.
Below we discuss how these simulations were made and their results, which are
presented in Appendix B.12
12
The appendix is available under request to the authors.
13
This simulates the increase in export prices that one could expect would result from the reduction in import tariffs of
the commercial partners of the country.
14
that of financial services is stable, as before. Aggregate savings and investment contract
more than before (-2,56%) indicating a larger (negative) impact on growth when the
existence of the financial market is taken into account. Exports of the non-financial
sector increase (1,1%) and imports reduce (-0,2%), both less than in the previous case,
but consistently with the smaller appreciation of the exchange rate in this case.
The drop observed before in Government income is expanded (-0,86%), and
increase in the income of families (0,07%) are both about 15% larger than in the
previous case.
The foreign reserves accumulation is 93% larger than in the base case, and is almost
double what is observed in the model without the financial sector. The increase in the
foreign capital balance (125%) is almost double what it was in the previous case.
Finally, we note there is a reduction in the total stock of capital intermediated by the
fund (-1,4%), and an increase in its average rate of remuneration of 0,01 p.p. This effect
did not exist in the version of the model without the financial market.
15
flow of funds between these agents and the intermediation fund. The financial
institutions borrow and lend more (+2,7% and +2,8%, respectively), and the
government does the opposite ( -2,0% and -5,7%, respectively).
16
5 CONCLUSIONS
In this paper we showed how a computable general equilibrium (CGE) model of the
Brazilian economy was extended to include an explicit model of financial market. The
model was implemented and calibrated for 2003, and used in some comparative statics
exercises designed to highlight the interactions between the financial and real parts of
the economy.
It is an advance in the applied general equilibrium literature, both in Brazil as well
as abroad, and contributes by not only providing an conceptual and analytical
framework for modeling the financial market and its interaction with real economy, but
also by implementing it completely.
The model was built by adding to a standard CGE model a set of equations that
represents the financial market. They establish linkages between the financial and real
sides of the economy, and describe how different agents act in the financial markets,
and how the equilibrium of the extended model (CGE-F) requires that equilibrium is
simultaneously reached in both markets. This makes this interaction truly bidirectional.
The financial market is modeled as an intermediation fund that borrows from some
agents and lends to others at interest rates that are specific to each agent and the nature
of the operation. The intermediation fund operates a technology that transforms funds
borrowed into funds loaned, and its decisions are made in a competitive manner, taking
rates as given. However, the usual general and partial equilibrium conditions are
superimposed on this context, requiring that budget equilibrium of the agents and of the
fund be respected.
After calibration the model was used to examine some public policies, to illustrate
its usefulness. We compared the solution of the model with and without the financial
market to parametric changes that simulate the adhesion of the country to symmetric
free trade agreements (reduction of 1 p.p. in the import and export tariffs). We also
analyze the impact of parametric shocks to three interest rates: a reduction of 1 p.p. in
the basic interest rate of public bonds, in the interest rate for loans to the no-financial
enterprises, and in the reference foreign interest rate. The results indicate that the
existence of the financial transmission channel in the Brazilian economy alters
significantly the impact of public policy, and that financial effects should be taken into
account when formulating them.
In comparing the simulations of symmetric free trade agreements with and without
the financial system we observe that the majority of impacts on quantities is toned
downs by between 10% and 15%, while the effect on the nominal variables (capital and
income) are expanded. This can be justified intuitively realizing that in the extended
model there is a larger set of prices to absorb the impact of the shock.
In the simulations of interest rate changes we note that a reduction of the basic
interest rate has a significative impact on income (+0,7%), on investment (+2,42). The
reduction in the interest rate for loans to non-financial enterprises does not affect the
output, but has an important impact on savings and investment (+12,4%) and, therefore,
on growth. The reduction of the foreign interest rate has a small impact on the domestic
economy. The repercussion of these exogenous shocks on the other interest rates ando n
the flows of the financial market are complex, and have been discussed in the text.
In conclusion, in this paper we showed how a CGE can be extended to incorporate a
model of the financial market that interacts with the real sector of the economy,
significantly increasing its capacity to represent in the simulation of shock and
17
conventional policies, and extending its utility by permitting the analysis of financial
market policies and shocks.
REFERÊNCIAS
18