economies

Article
Leaning against the Wind Policies on Vietnams
Economy with DSGE Model
Phuc Huynh, Trang Nguyen, Thanh Duong * and Duc Pham
John von Neumann Institute, Vietnam National University, HoChiMinh 70000, Vietnam;
[email protected] (P.H.); [email protected] (T.N.); [email protected] (D.P.)
* Correspondence: [email protected]

Academic Editor: David Dapice

Received: 6 July 2016; Accepted: 17 November 2016; Published: 18 January 2017

Abstract: The global financial crisis of 20072008 had a negative impact on many countries, including
Vietnam. Many policies have been applied to stabilize the macro-economic indicators. However, most
of them are based on old qualitative models, which do not help policy makers understand deeply
how each one affects the economy. In this paper, we investigate a quantitative macro-economic
approach and use leaning against the wind policies with the Dynamic Stochastic General Equilibrium
model (DSGE) to find a better way to understand how policies stabilize the Vietnamese economy.
Based on the framework of Gerali et al., we calibrate the hyper-parameter for Vietnam financial
data and do the comparison between the standard Taylor rule and the cases in which we add asset
price and credit elements. The results show that the credit-augmented Taylor rule is better than
the asset-price-augmented one under the technology shock and contrary to the cost-push shock.
Moreover, the extended simulation result shows that combining both asset-price and credit rules on
the model is not useful for Vietnams economy in both types of shock.

Keywords: Taylor rules; monetary policy; leaning against the wind; macro-economy; Vietnam
financial market; DSGE; inflation

JEL Classification: C68; E00; E12; E52; E62

1. Introduction
Monetary policy is an extremely important macro-economic regulation of government in a market
economy. It is a process by which the authority of a country controls the money supply and tends to
target an expected inflation rate or interest rate to ensure price stability, general trust in the currency
and lowering the unemployment rate. Moreover, it also maintains predictable exchange rates with
other currencies. Monetary economics provides insight into how to craft optimal monetary policy.
However, the global financial crisis of 20072008 showed the lack of policy execution in many countries.
The economists realize that the monetary policies are deficient in considering some financial variables
and parameters.
On the other hand, Vietnam was also partially affected by the crisis because the government was
unable to predict that sudden situation and passive in executing monetary policies. By applying tight
monetary policies in the early stage, they created huge shocks and negative effects to the currency
markets, as well as commercial banks. The inflation rate and non-performing loan rate of Vietnam
increased very greatly. Many policies were enforced, such as buying potentially billions of dollars of
bad loans from the banks with official overdue debts (from 3% to 6%) [1]. All of them are based on
qualitative models that do not help policy makers understand deeply how each one affects the economy.
Therefore, we need another optimal solution to stabilize the Vietnamese economy. Although there are
many studies about monetary policy, they are still missing quantitative models to evaluate correctly

Economies 2017, 5, 3; doi:10.3390/economies5010003 www.mdpi.com/journal/economies

Economies 2017, 5, 3 2 of 18

the effectiveness. In this paper, we investigate some macro-economic regulations and propose to use
the Leaning Against The Wind (LATW) policy through the Dynamic Stochastic General Equilibrium
(DSGE) model to find a suitable model for the Vietnamese economy. It was successfully applied in the
European Central Bank. Leaning against the wind is defined as the Central Bank following a rule
where the policy rate is adjusted not only in response to fluctuations of inflation and output, but also
to the change of financial variables (asset prices, credit, etc.) (Leonardo and Federico, 2013) [2].
Lars E. O. Svensson (2014) [3] showed that the LATW was ineffective for Sweden. According to
his paper, Sweden applied the LATW since 2010, and Riskbanks (Swedens central bank) realized that
the inflation target was low and the unemployment was high. Similarly, Fabio Verona et al. (2014) [4]
had evaluated the effective by using LATW policies for the U.S. market. These authors showed that
both the credit-augmented Taylor rule and asset-price-augmented Taylor rule were effective with
various shocks. Furthermore, some other studies regarding LATW, such as Lambertini et al. (2011) [5]
and Christiano et al. (2010) [6], did not have financial frictions in banking and also did not perform
a full grid-search analysis. Thus, in this paper, we simulate various scenarios of the combined Vietnam
data to validate the model.
We make the following contributions: (i) calibrating hyper-parameters for Vietnams economy,
where the data of the interest rate and GPD are collected from Quarters 1/19994/2014; (ii) inspired by
the work of Gerali et al. (2010) [7], we extend their model by combining multiple economic policies.
The results show that interest rate modeling when combining asset price and credit with the standard
Taylor rule is not suitable for the Vietnamese market economy. This is a useful tool to identify problems
and simulate solutions in operating the Vietnamese economy for policy makers. Besides, to estimate
optimal parameters, we simulate each policy in many scenarios. When a shock happens, depending
on the objectives that the Vietnamese economy is approaching, this tool will propose a specific suitable
interest rate quantitatively, instead of qualitatively.
The rest of paper is organized as follows. Section 2 reviews the related literature. Section 3
presents the Dynamic Stochastic General Equilibrium (DSGE) model. Section 4 describes the main
financial channels. Section 5 proposes our simulation method. Section 6 presents how parameters are
calibrated, and Section 7 is the empirical results. Section 8 is the conclusion.

2. Literature Overview
The theoretical basis of the model in this paper originated from the two papers of Bernanke,
Gertler and Gilchrist (1999) [8] and Lacoviello (2005) [9]. In that literature, they assumed that banks did
not have any important role in credit transactions. However, in the real world, banks play a vital role
in modern financial systems. This drawback induced the inaccurate prediction of the global crisis 2007.
In order to overcome this weakness, Gerali et al. (2010) [7] added a new agent into the model banks.
Gerali stated that the banking sector in a Dynamic Stochastic General Equilibrium (DSGE) could
clearly explain the role of banking inter-mediation in the transmission of monetary policy impulses
and analyzed how shocks that originate in credit markets are transmitted to the real economy. On the
other hand, compliance with the Basel Accords imposes capital requirements to exert banking activity.
Gerali changed the standard model by adding credit frictions and borrowed constraints as in Iacoviello
(2005) [9] and a set of real and nominal frictions as in Christiano et al. (2005) [10] and Smets and
Wouters (2003) [11] with an imperfectly-competitive banking sector that collected deposits and then
supplied loans to other agents subject to the requirement of using banking capital.
Based on main idea of Curdia and Woodford (2010) [12] combined with a simplified version of the
above from Gerali et al. (2010) [7], the model included two transmission channel are: the bank lending
channel with the presence of a target level for the banks leverage and the balance-sheet channel,
which was referred from Lacoviello (2005) [9] and assumed that entrepreneurs borrowing capacity
was linked to the value of the assets that they could pledge as collateral.
With regard to frictions in financial intermediates, many recent papers studied this issue, such as
Bernanke and Gertler (2000, 2001) [13,14], Gilchrist and Leahy (2002) [15] and Iacovielllo (2005) [9].
Economies 2017, 5, 3 3 of 18

The main feature of those was only to consider financial frictions from the borrowers side, but the
credit-supply effects were still missing. Therefore, financial intermediates behavior was not expressed
clearly. That was a reason why the framework was enriched by adding the leverage constraint into the
banking sector. This would prevent the implication effect when shocks happen.
Related to the central banks policy interest rate, a paper from Curdia and Woodford (2009,
2010) [12,16] had modified a standard Taylor rule for a central banks policy interest rate to incorporate
either an adjustment for changing interest rate spreads or a response to variations in the aggregate
volume of credit. The results from that papers had shown that modified rules were better than the
standard Taylor rule with different shocks. In our paper, besides the standard Taylor rule, there are
two additional rules, the asset-price augmented rule and the credit augmented rule.
Moreover, Lambertini et al. (2011) [5] and Christiano et al. (2010) [6] did not have financial
frictions in banking and did not perform a full grid-search analysis. Therefore, the papers contribution
to the existent literature was an additional friction in banking and simulation to find out the optimal
parameters for a desirable policy.

3. DSGE Model
We use the model that is from the papers of Gerali et al. (2010) [7] and Leonardo Gambarcorta et al.
(2013) [2]. It introduces an important new agent, bank, into the DSGE model along with financial
frictions (Iacoviello, 2005) [9]. This model is populated by five agents: households, entrepreneurs,
banks, capital-good producers and retailers. Figure 1 describes how the model works.

Figure 1. The DSGE model.

The model starts with banks. They collect deposits from households and lend money to
entrepreneurs. These loans depend on two financial frictions: one from the level of bank leverage and
the other from the value of collateral. Then, entrepreneurs will make use of money to hire household
labor and buy capital from capital-good producers. Finally, some retailers will buy the intermediate
goods from entrepreneurs in a competitive market, brand them at no cost and sell them with a mark-up
over the purchasing cost to households.

3.1. Households
Each household always maximizes its consumption and works less. Therefore, we assume the
household utility function as below:

" #
  ltP (i )1+
max E0 tP log ctP (i ) , (1)
{ctP (i),ltP (i),dtP (i)} t=0 1+
Economies 2017, 5, 3 4 of 18

subject to the budget constraint:

ctP (i ) + dtP (i ) wt ltP (i ) + (1 + rtib1 )dtP1 (i ) + JtR (i ), (2)

where ctP (i ) is consumption, ltP (i ) is labor supply, dtP (i ) is bank deposits, which are paid at a rate equal
to the policy rate rtib , wt is the real wage and JtR (i ) are the retailers profits.
Based on the budget constraint, it is easy to see that the household inflow is the wage that is paid
by entrepreneurs, the interest receive from banks and the profits from selling to the other households.
Therefore, the outflow, which includes consumption and deposits, has to be less than or equal to
the inflow.
By using Lagrange multipliers, the First Order Conditions (FOC) for households are:

1 (1 + rtib )
= Et , (3)
ctP (i ) ctP+1 (i )

wt
ltP (i ) = P
. (4)
c t (i )

3.2. Entrepreneurs
Entrepreneurs maximize their utility function by consumption:
 
max E0 tE log ctE (i ) , (5)
{ctE (i ),ltP,d (i ),btEE } t =0

subject to the budget constraint:

P,d ytE (i )
ctE (i ) + (1 + rtb1 )btEE k E
1 ( i ) + w t l t ( i ) + q t k t ( i ) + btEE (i ) + qkt (1 k )k tE1 (i ), (6)
xt

m E qkt+1 k tE (i )(1 k )
btEE (i ) , (7)
1 + rtb

where ctE (i ) is the entrepreneurs consumption, ltP,d (i ) is the labor demand, k tE (i ) is the entrepreneurs
stock of capital, qkt is the price of capital, ytE is the output of intermediate goods produced by
entrepreneurs, which is the following CobbDouglas:

ytE (i ) = AtE (k tE ) (ltP,d )(1 ) , (8)

xt is the mark-up of the retailer sector, k is the depreciation of capital, btEE (i ) is the amount of the
banks loans taken by the entrepreneurs, m E is a parameter that can be interpreted as the Loan-To-Value
(LTV) ratio chosen by the banks and rtb is the interest rate on the banks loans.
The FOC are:

1 (1 + rtb )
2,E
t (i ) = E , (9)
ctE (i ) ctE+1 (i )

 
2,E
t ( i ) m E qk
t t +1 1 k
E h k   i qkt
k k
+ q t + 1 1 + r t + 1 = , (10)
1 + rtb ctE+1 (i ) ctE (i )
Economies 2017, 5, 3 5 of 18

(1 )ytE (i )
= wt , (11)
ltP,d (i ) xt

where the variable 2,E

t (i ) is the second Lagrange multiplier on the borrowing constraint (and thus
represented the marginal value of one unit of additional borrowing by the entrepreneurs), and 1 is
the output elasticity of labor in the production function.

3.3. Capital Good Producers

Because capital-good producers only sell fixed-capital to entrepreneurs, they maximize their
utility function by the level of the price:

1,E h   i
max E0
It+s
s t1,E
+s
Qkt+s Kt+s Qkt+s 1 k Kt+s1 Pt+s It+s , (12)
s =0 t

subject to:
2 !
i


k
 It+s
Kt+s = 1 K t + s 1 + 1 1 It+s . (13)
2 It+s1

Perfectly-competitive firms buy last-period capital at price Qkt from entrepreneurs and It units
of final goods from retailers at price Pt ; 1,E
t is the entrepreneurs marginal utility of consumption;
Q k
qkt t is the real price of capital; k is the rate of depreciation.
Pt
The FOC is:
" 2 # " 1,E 2 #
i I 
I

I

I

I
t t t t + 1 t + 1 t + 1
1 = qkt 1 1 2 1 + E Et qkt+1 i 1 . (14)
2 It1 It1 It1 1,E It
t
It

3.4. Retailers
Similar to capital-good producers, retailers only sell final goods to households, and so, they
maximize the price according to the utility function:

1,P
"  2 #
P Pt+s (i )
max E0
Pt+s
P 1,P t+s
Pt+s (i )ytE+s (i ) PtW E
+s y t+s (i )
2 Pt+s1 (i )
1 Pt+s ytE+s , (15)
s =0 t

subject to the demand derived from consumers maximization,

 ey
Pt+s (i )
ytE+s (i ) = ytE+s , (16)
Pt+s

1,P
t+s
where P is the households stochastic discount factor, 1,Pt is the households marginal utility of
1,P
t
consumption at time t, ctP is the current consumption, PtW is the wholesale price and Pt (i ) is the retail
y e 1 Pt
price; mk t , mctE = is the real marginal cost, and xt W .
e1 xt Pt
The FOC is:
y y
" #
mk t mk t E tP+1 Yt+1
1 y + y mct P (t 1) t + P Et P ( t +1 1 ) t +1 =0 (17)
mk t 1 mk t 1 tP t

3.5. Banks
Each bank j is composed of two units: a wholesale branch and a retail branch.
Economies 2017, 5, 3 6 of 18

The wholesale branch: maximizes the following utility function subject to a balance-sheet constraint:
!2
Ktb ( j)
max Rbt bt ( j) rtib dt ( j) Ktb ( j), (18)
{bt ( j),dt ( j)} 2 bt ( j )

s.t. bt ( j) = dt ( j) + Ktb ( j). (19)

In this Equation (18), we can see that the wholesale unit loans bt ( j) with the wholesale loan rate Rbt ( j)
by collecting deposits dt ( j) from households, which are paying the interest rate set by the central bank
rtib . Besides, the banks have their own funds Ktb ( j), which are accumulated from reinvested profits.
Moreover, is considered as the target leverage ratio and banks have to pay a certain cost for
deviating from that target. In fact, the target leverage ratio is one of the tools of macro-prudential
policy, which is applied to this paper. Whenever the bank deviates from that ratio, this leads to the
losing of bank profits. Therefore, this will limit the ability of lending while the shocks are happening.
Using Lagrange multipliers for Equations (18) and (19), the FOC is:
! !2
Ktb ( j) Ktb ( j)
Rbt = rtib . (20)
bt ( j ) bt ( j )

The retail branch: We assume that the retail banks operate in terms of monopolistic competition.
Moreover, they buy wholesale loans, differentiate at no cost and resell to final borrowers. The utility
function will be: ! !2
K b ( j) K b ( j)
rtb = rtib t
t
+ b , (21)
bt ( j ) bt ( j )

where rtb is the retail loan rate and the mark-up b is constant.

4. The Transmission Channels

This section will introduce two transmission channels: (i) a collateral channel; and (ii) a
credit-supply channel.
The model includes two financial frictions: (i) entrepreneurs value of assets that affects their
borrowing ability; and (ii) banks set up a target level and pay a cost if they deviate from it. Based on
these financial frictions, we will present them as two additional transmission channels: (i) a collateral
channel, which is linked to the borrowing constraint; and (ii) a credit supply channel, which is linked
to the presence of a positively-sloped loan-supply, which shifts with changes in the policy rate and
bank capital.
The collateral channel: The collateral channel operates based on the fluctuations of asset prices on
the debtors balance-sheet condition, which influences the borrowing constraint. In this channel, the
model assumed that E < P , and this lead to the households being net lenders and the entrepreneurs
net borrowers, as shown in Iacoviello (2005) [9]. Moreover, the paper of Andres et al. (2010) [17] showed
that: (i) the entrepreneurs always consume a fraction 1 E of their net worth; (ii) the entrepreneurs
capital is also a linear function of net wealth; the coefficient is time varying and depends on asset
prices, the LTV ratio and the loan interest rate. The equations are:
 
ctE = 1 E NWtE , (22)
E
Kt = NWtE , (23)
qkt t

where NWtE and t are the entrepreneurs net worth and the entrepreneurs leverage, respectively
(Bt = t Kt ). These two variables are defined as:
 Economies 2017, 5, 3 7 of 18

    Yt
NWtE qkt 1 k Kt1 1 + rtb1 Bt1 + , (24)
xt
m E qkt+1 (1 k )
t . (25)
1 + rtb

From the above equations, we can draw a few main ideas about this model as follows:

(i) Equation (25) shows that the leverage ratio depends positively on the parameter m E and future
asset prices qkt+1 and negatively on the loan interest rate.
(ii) The asset prices and the lending rates from (25) have a crucial impact on the entrepreneurs net
worth, which influences simultaneously both consumption and investment decisions described
through Equations (22) and (23).
(iii) Equations (23)(25) can be interpreted as using past net worth, which is the basis to borrow
money from intermediates with a given fraction in order to buy future capital.
(iv) In detail, from Equations (22) and (24), we see that the current level of asset prices (qkt ) has a
positive relation with net worth; it has a positive impact on consumption, as well. Hence, the
increase of the current level of asset prices will induce the increase of consumption. Conversely,
the impact of the current level of asset prices on investment (Equations (23)(25)) is ambiguous,
because the positive effect of the current level of asset prices will be counteracted by the negative
effect of leverage, which is considered as the cost of purchasing new capital. Expectations on
future asset prices (qkt+1 ) positively impact investment (easy to see from Equations (23) and (25)),
whereas the loan interest rate (rtb ) negatively impacts both consumption and investment.
Therefore, the increase of the loan rate will decrease net worth due to the increase of interest
payments and the decrease of capital accumulation.

The credit-supply channel: Nowadays, in most countries, including Vietnam, banks have been
implementing risk management following the rules of Basel II. Hence, this model assumed that credit
intermediates manage supply conditions to the target capital to asset ratio 1 . Whenever this ratio
deviates, banks will tighten lending standards in order to limit lending amounts and bring this ratio
back to the desired level. Equation (21) can be written in linear form as follows:

3 b 3 bb
rtb = rtib + Bt K , (26)
1 + rib 1 + rib t
Many of the main ideas we can draw are as follows:

(i) The loan supply schedule B bt has a positive effect on the loan rate. The implication is that the
increase of the lending amount will lead to a higher loan rate due to higher default risk.
(ii) The multiplier of loan supply has a positive correlation for both the level of the banks target
capital-to-asset ration () and the cost for deviating from that target .
(iii) From Equation (19), the level of bank capital has a positive effect on the loan supply, which
depends positively on the banks profits. The capital will be accumulated through an increase
in the banks profit; banks have numerous capital to lend, and this leads to the shift in the loan
supply in the subsequent period. Moreover, the increase of bank profits will reduce the loan rate
(in the subsequent period) for any given level of loans to the economy.
(iv) In particular, the loan supply is inversely proportional to the level of the policy rate. Hence,
whenever the central bank implements an easy policy, this will lead to the increase of loan supply.
This feature is consistent with the existence of a bank-lending channel in the model.

In fact, the interaction between the credit-supply channel and the collateral channel will determine
the equilibrium level of the interest rate. In detail, the changes in credit supply will impact the

1 According to Basel 2regulations, referred to as the Capital Adequacy Ratio (CAR).

 Economies 2017, 5, 3 8 of 18

entrepreneurs net wealth, consumption and investment decisions. This will increase the implication
effect, because the effect of different shocks increases or decreases the prices of assets as described in
the collateral channel.

5. Simulations

5.1. Monetary Policies

In this section, we will test the efficiency of an LATW by empirical research with real data collected
from the Vietnamese economy through the Dynare package. This answers the key question Could
LATW policies apply for Vietnam?. Hence, first of all, we will introduce policies used to stabilize the
economy. We assume that the central bank follows a general-form Taylor rule as follows:
 h i
rtib = ibe
e rtib1 + 1 ib
bt + y Y bt + q qbkt .
bt + B B (27)

We take into account policies one-by-one; specifically, the three policies will be:

(i) The standard Taylor rule:

 h i
rtib = ibe
e rtib1 + 1 ib
bt + y Y
bt . (28)

(ii) The Taylor rule augmented with asset prices:

 h i
rtib = ibe
e rtib1 + 1 ib bt + q qbkt .
bt + y Y (29)

(iii) The Taylor rule augmented with credit:

 h i
e rtib1 + 1 ib
rtib = ibe bt + y Y
bt + B B
bt . (30)

5.2. Shocks
To determine the policies that are desirable, we propose two shocks in the DSGE model: (i) the
technology shock; and (ii) the cost-push shock.

5.2.1. Technology Shock

The technology shock is a shock to total factor productivity, i.e., to the variable (AtE ) in the
production function:
ytE = AtE (k tE ) (l P,d )(1 ) . (31)

We assume that AtE follows an AR(1) process:

AtE = A AtE1 + A
t . (32)

We set A = 0.95, and the variance of AtE equals one percent, which are taken from Gerali et al.
(2010) [7]. Technology shocks are events in a macroeconomic model that change the production
function. The term shock is defined as a sudden change in the economy. A technology shock is when
there is a sudden change in technology to either benefit or worsen economic activity. This type of
shock has big effects on companies that are solely dependent on technology as their main source of
labor or production, such as manufacturing plants or oil/energy extraction.

5.2.2. Cost-Push Shock

The second shock considered in this model is the cost-push shock (the so-called cost-push
inflation), defined as a shock to the elasticity of substitution between varieties in the goods market.
A situation that has been often cited of this was the oil crisis of the 1970s. Petroleum is so important to
Economies 2017, 5, 3 9 of 18

industrialized economies with no suitable alternative available, so a large increase in its price can lead
to the increase in the price of most products, raising the inflation rate.
y
In this model, the cost-push shock is modeled through the firms mark-up mk t as follows:
y y y
mk t = y mk t1 + t . (33)
y
We set y = 0.50, and the variance of t is calibrated, so that the variance of mk t equals one
percent, which are taken from Gerali et al. (2010) [7].

5.3. Loss Function

In order to identify the desirable LATW policies in terms of macroeconomic stabilization, we
consider assumptions about the central banks preferences. In particular, we assume that the central
banks objective is the minimization of the weighted sum of the variance of inflation and output:

Loss = Var ( ) + Var (Y ). (34)

We set the weighting parameter to vary within the range [0,2], i.e., we allow for a broad range
of values for the relative weight of inflation versus output stabilization. The increase of value shows
that the central bank focuses on stabilizing output Y. For each value of , we calculate the value of the
loss function for each rule and then pick the minimum value as the best policy outcome attainable
under that rule.

6. Calibration
In this section, we will introduce the specific parameters that are used for the simulation. A few
of them are calibrated by the Vietnamese data. Due to the limit of the ability of the computer, we
only simulate some scenarios to obtain a general view of the application of LATW policies in Vietnam.
In particular, parameters y , B and q are changed from zero to two with a step size of one; parameter
is changed with the values being: 0.5, 1.5, 3 and 4.5. The value of the households discount factor
P is 0.996, which implies a steady-state policy rate of roughly 2% (annualized). The entrepreneurs
discount factor E is set at 0.975, which is from Iacoviello (2005) [9]. The inverse of the Frisch elasticity
is set at one (Gali, 2008) [18]. The share of capital in the production function () and the depreciation rate
of physical capital (k ) are set at 0.20 and 0.05, respectively, in order to match the investment-to-GDP
ratio and the entrepreneurs share in consumption in Gerali et al. (2010) [7]; which equal 0.11 and 0.09,
respectively. The
 elasticity
 of substitution across goods y is set at six, implying that the mark-up in the
y
goods market is 1.20. The degree of price stickiness p is set at 28.65, the value estimated by
y 1
Gerali et al. (2010) [7]. As mentioned, the degree of monetary policy inertia is set at 0.77. As regards the
parameters related to the financial frictions/banking sector, we follow Gerali et al. (2010) [7]. The LTV
ratio set by the banks m E is set at 0.35, which is similar to what Christensen et al. (2007) [19] estimated
for Canada and to the average ratio of long-term loans to the value of shares and other equities for
non-financial corporations in the European area, as in Gerali et al. (2010) [7]. The target capital-to-asset
ratio , the cost for managing the bank capital position b and the elasticity of substitution across loan
varieties eb (which determines the steady-state loan spread) are set at 9%, 0.049 and 3, respectively,
as in Gerali et al. (2010) [7]. The bank capital adjustment cost equals 11, the estimated value in
Gerali et al. (2010) [7].

7. Empirical Results
In this section, we discuss some results obtained by simulating in different scenarios, including
technology shock and cost-push shock. For each scenario, we try four groups of policies. The first one
is the Taylor rule, which only takes into account the interest rate and inflation rate parameters. This is
the baseline or standard rule of our model. We further augment the standard rule with the following
Economies 2017, 5, 3 10 of 18

extra parameters: asset price, credit and both of them. The simulation results will be shown in the
next sections.

7.1. Scenario 1: Technology Shock

Figure 2 shows that, after the technology shock, the increase in asset prices has led to the relaxation
in the borrowing constraint of the entrepreneurs. Furthermore, under the standard rule, both the
policy and the loan rate decrease. The decrease of the loan rate increases entrepreneurs lending
demand, which increases banks balance-sheet and leverage (the increase in leverage has activated the
credit-supply channel), which also induces the strong increase of the investment and consumption of
entrepreneurs and households (due to higher wages).

Figure 2. Impulse response functions after a technology shock.

Under the credit-augmented Taylor rule, on the contrary, both the policy rate and loan rate
increase when the shock hits the economy. This shows the counteracting effect of leaning against the
wind, whereby the central bank could prevent the over-extension of banks, as well as the amplification
of asset values. In addition, the tightening of monetary policy has affected the increase of bank
lending rates, which induces a smaller expansion of bank leverage. In turn, in this case, borrowers
financing conditions improve significantly less (because the increase of the loan rate has decreased the
entrepreneurs net worth), so that the increase in investment, consumption and output is lower. Thus,
the reduction of output volatility under leaning against the wind has lost both the amplification effect
of the traditional collateral channel and the credit supply channel.
Meanwhile, under the asset-price-augmented rule, both the policy rate and loan rate significantly
decrease; this has led to the strong increase of output, which the implication effect did not dampen. Thus,
the asset-price Taylor rule proved virtually ineffective. Looking at Table 1, the asset-price-augmented
Economies 2017, 5, 3 11 of 18

rule is ineffective; the optimal parameter of q is zero. On the contrary, the credit-augmented rule is
a desirable policy after technology shock; we can see this clearly through Figure 3. From Figure 3, we
can see that the gains 2 under the standard Taylor rule and the credit-augmented rule are positive
numbers; this shows that the credit-augmented rule is better than the standard Taylor rule. Moreover,
we can see from this figure that the gains increase as increases. To explain this, in this model, the
increase in output (due to the positive shock) has increased the entrepreneurs net worth (Equation (24)).
The borrowing conditions is loosened, and entrepreneurs will borrow more money, which leads to the
rapid expansion of credit. Hence, the credit-augmented rule is effective in this case.

Table 1. Losses and gain under both the standard Taylor rule and the asset-price-augmented rule, after
a technology shock.

Output Weight Standard Taylor Rule Asset-Price-Augmented Rule

y Loss pi y q Loss Gain
0 3 0 0.0008743 3 0 0 0.00008743 0
0.1 0.5 0 0.00044837 0.5 0 0 0.00044837 0
0.2 0.5 0 0.00071837 0.5 0 0 0.00071837 0
0.3 0.5 0 0.00098837 0.5 0 0 0.00098837 0
0.4 0.5 0 0.00125837 0.5 0 0 0.00125837 0
0.5 1.5 2 0.00145042 1.5 2 0 0.00145042 0
0.6 1.5 2 0.001460504 1.5 2 0 0.001460504 0
0.7 1.5 2 0.001470588 1.5 2 0 0.001470588 0
0.8 1.5 2 0.001480672 1.5 2 0 0.001480672 0
0.9 1.5 2 0.001490756 1.5 2 0 0.001490756 0
1.0 1.5 2 0.00150084 1.5 2 0 0.00150084 0
1.1 1.5 2 0.001510924 1.5 2 0 0.001510924 0
1.2 1.5 2 0.001521008 1.5 2 0 0.001521008 0
1.3 1.5 2 0.001531092 1.5 2 0 0.001531092 0
1.4 1.5 2 0.001541176 1.5 2 0 0.001541176 0
1.5 1.5 2 0.00155126 1.5 2 0 0.00155126 0
1.6 1.5 2 0.001561344 1.5 2 0 0.001561344 0
1.7 1.5 2 0.001571428 1.5 2 0 0.001571428 0
1.8 1.5 2 0.001581512 1.5 2 0 0.001581512 0
1.9 1.5 2 0.001591596 1.5 2 0 0.001591596 0
2.0 1.5 2 0.00160168 1.5 2 0 0.001591596 0
Output Weight Standard Taylor Rule Credit-Augmented Rule
y Loss pi y q Loss Gain
0 3 0 0.0008743 3 0 0 0.00008743 0
0.1 0.5 0 0.00044837 3 0 2 0.000366897 18.17093026
0.2 0.5 0 0.00071837 3 0 2 0.000464364 35.35865919
0.3 0.5 0 0.00098837 3 0 2 0.000561831 43.15580198
0.4 0.5 0 0.00125837 3 0 2 0.000659298 47.60698364
0.5 1.5 2 0.00145042 3 0 2 0.000756765 47.82442327
0.6 1.5 2 0.001460504 3 0 2 0.000854232 41.51114958
0.7 1.5 2 0.001470588 3 0 2 0.000951699 35.28445765
0.8 1.5 2 0.001480672 3 2 1 0.001025644 30.73118152
0.9 1.5 2 0.001490756 3 2 1 0.00102885 30.98471514
1.0 1.5 2 0.00150084 3 2 1 0.001032055 31.23484182
1.1 1.5 2 0.001510924 3 2 1 0.001035261 31.48162978
1.2 1.5 2 0.001521008 3 2 1 0.001038466 31.72514543
1.3 1.5 2 0.001531092 3 2 1 0.001041672 31.96545341
1.4 1.5 2 0.001541176 3 2 1 0.001044877 32.2026167
1.5 1.5 2 0.00155126 1.5 2 0 0.001048083 32.43669662
1.6 1.5 2 0.001561344 3 2 1 0.001051288 32.66775291
1.7 1.5 2 0.001571428 3 2 1 0.001054494 32.89584378
1.8 1.5 2 0.001581512 3 2 1 0.001057699 33.12102595
1.9 1.5 2 0.001591596 3 2 1 0.001060905 33.34335472
2.0 1.5 2 0.00160168 3 2 1 0.00106411 33.56288397

2 Gain is the percentage difference between the minimum loss under the standard Taylor rule and the augmented Taylor rule
(100 (Loss|STR Loss|ATR) /Loss|STR). A positive number means that the augmented rule performs better than the
standard one.
Economies 2017, 5, 3 12 of 18

Figure 3. % gain under the standard Taylor rule and that credit-augmented rule.

7.2. Scenario 2: Cost-Push Shock

From Figure 4, after an adverse inflation shock, the output significantly decreases under the
standard rule compared to under the asset price-augmented rule. We see that the different rules have
led to the difference in output volatility. Under the standard rule, both the policy and the lending rates
rise over time; the banks balance sheets (leverage) will be reduced; leading to a sharp deterioration of
entrepreneurs financing conditions and, hence, the fall in investment.

Figure 4. Impulse response functions after the cost-push shock.

Under the rule including asset prices, instead, both the loan rate and policy rate are larger at
first, but decrease later; the monetary policy is eased on impact; this has contributed to limiting
the deterioration of the entrepreneurs financing conditions by sustaining banks balance sheets and
Economies 2017, 5, 3 13 of 18

avoiding a major disruption of credit supply. The output and investment increase, as well. We confirm
that through Figures 5 and 6, the loss function value is smaller, and the gain is larger under the
asset price-augmented rule than others. The value of is larger, and the gain and the loss under the
asset-price-augmented rule is better (more detail in Tables 1 and 2). This shows that the LATW policies
are effective at stabilizing the economy for various shocks.

Table 2. Losses and gain under both the standard Taylor rule and the credit-augmented rule, after the
cost-push shock.

Output Weight Standard Taylor Rule Asset-Price-Augmented Rule

y Loss pi y q Loss Gain
0 1.5 0 0.00002388 1.5 0 0 0.00002388 0
0.1 3 0 0.000264782 3 0 0 0.000264782 0
0.2 3 0 0.000504782 3 0 1 0.000435828 13.66015428
0.3 3 0 0.000744782 3 0 1 0.000453812 39.06780776
0.4 3 0 0.000984782 3 0 1 0.000471796 52.0913258
0.5 3 0 0.001224782 3 0 1 0.00048978 60.01084275
0.6 3 0 0.001464782 3 0 1 0.000507764 65.33518298
0.7 3 0 0.001704782 3 0 1 0.000525748 69.16039705
0.8 3 0 0.001944782 3 0 1 0.000561716 72.0414936
0.9 3 0 0.002184782 3 0 1 0.00102885 74.28960876
1.0 3 0 0.002424782 3 0 1 0.0005797 76.09269617
1.1 3 0 0.002664782 3 0 1 0.000597684 77.5709983
1.2 3 0 0.002904782 3 0 1 0.000615668 78.80501876
1.3 3 0 0.003144782 3 0 1 0.000633652 79.85068599
1.4 3 0 0.003384782 3 0 1 0.000651636 80.7480659
1.5 3 0 0.003624782 1.5 2 1 0.00066962 81.52661319
1.6 3 0 0.003864782 3 0 1 0.000687604 82.20846609
1.7 3 0 0.004104782 3 0 1 0.000705588 82.81058531
1.8 3 0 0.004344782 3 0 1 0.000723572 83.346184
1.9 3 0 0.004584782 3 0 1 0.000741556 83.82570862
2.0 3 0 0.004824782 3 0 1 0.00075954 84.25752708
Output Weight Standard Taylor Rule Credit-Augmented Rule
y Loss pi y q Loss Gain
0 1.5 0 0.00002388 1.5 0 0 0.00002388 0
0.1 3 0 0.000264782 3 0 0 0.000264782 0
0.2 3 0 0.000504782 3 0 0 0.000504782 0
0.3 3 0 0.000744782 3 0 0 0.000744782 0
0.4 3 0 0.000984782 3 0 0 0.000984782 0
0.5 3 0 0.001224782 3 0 0 0.001224782 0
0.6 3 0 0.001464782 3 0 0 0.001464782 0
0.7 3 0 0.001704782 3 0 0 0.001704782 0
0.8 3 0 0.001944782 3 0 0 0.001944782 0
0.9 3 0 0.002184782 3 0 0 0.002184782 0
1.0 3 0 0.002424782 3 0 0 0.002424782 0
1.1 3 0 0.002664782 3 0 0 0.002664782 0
1.2 3 0 0.002904782 0.5 0 2 0.000615668 0
1.3 3 0 0.003144782 0.5 0 2 0.002945406 6.769049836
1.4 3 0 0.003384782 0.5 0 2 0.002979668 13.59594425
1.5 3 0 0.003624782 0.5 0 2 0.00301393 20.267624
1.6 3 0 0.003864782 0.5 0 2 0.003082454 26.78932298
1.7 3 0 0.004104782 0.5 0 2 0.000705588 33.16604238
1.8 3 0 0.004344782 0.5 0 2 0.000723572 39.40256347
1.9 3 0 0.004584782 0.5 0 2 0.003150978 45.50345956
2.0 3 0 0.004824782 0.5 0 2 0.00318524 51.47310721
Economies 2017, 5, 3 14 of 18

Figure 5. Loss function of the Leaning Against The Wind (LATW) policies after the cost-push shock.

Figure 6. % gain after the cost-push shock.

According to the report of the General Statistics Office of Vietnam, the increase of oil price from
17% up to 24% and electricity prices up to 15.2% in March 2011 led to the inflation increases from 12.8%
in the first quarter of 2011 (up 60% compared with the prior year) up to 19.83% in the fourth quarter of
2011 (up 38.35% compared with the prior year) as described in Figure 7, also the policy rate increases
from 12% in the first quarter of 2011 (up 59.97% compared with the prior year) to 15% in the fourth
quarter of 2011 (up 120.28% compared with the prior year), as described in Figure 8. However, the
goal of the Vietnamese monetary policy is to curb inflation through controlling the interest rate. Thus,
GDP slowly grew from the first quarter of 2011 to the first quarter of 2012, as in Figure 9. The results
realistically explained the responses described in Figure 4.

Figure 7. Vietnamese inflation rate from the first quarter of 2010 to the fourth quarter of 2014.
Economies 2017, 5, 3 15 of 18

Figure 8. Vietnam Central Bank policy rate from the first quarter of 2010 to the fourth quarter of 2014.

Figure 9. Vietnam GDP from the first quarter of 2010 to the first quarter of 2015.

7.3. Combining Asset Price and Credit

From the previous results, a question we investigated in this paper is: can combining both asset
price and credit improve the model in both scenarios of technology shock and cost-push shock? In the
scenario of technology shock, even combined with the two new policies, the Taylor rule still has the
best result with the lowest loss. Note that, in Figure 10, the loss functions of the credit-augmented
rule and standard rule are identical. The loss function of the asset price and credit rule is drawn in
the secondary axis of the figure. In order to compare with the standard Taylor rule, we compute the
percentage of gain after the technology shock, as shown in Figure 11.

Figure 10. Loss function of the LATW policies after the technology shock.
Economies 2017, 5, 3 16 of 18

Figure 11. % gain after the technology shock.

In the scenario of cost-push shock, the asset price-augmented rule is still the best rule with the
lowest loss. Figure 12 shows the loss functions of the four methods. Similar to Figure 10, the loss
function of the asset price and credit rule is drawn in the secondary axis.
The standard Taylor rule has been used for a long time in stabilizing the economy. For example,
in the context of a positive shock, it has proven effective for the economy of Vietnam. When a negative
shock occurs, the asset-price-augmented rule proved to be more effective. However, the combination
of the asset-price and credit rule on the model is shown not to be useful for Vietnams economy in
both types of shock. This can be seen in Figure 13.

Figure 12. Loss function of the LATW policies after the cost-push shock.

Figure 13. % gain after cost-push shock.

Economies 2017, 5, 3 17 of 18

8. Conclusions
This paper showed the effectiveness of LATW policies for application to Vietnam. The Vietnamese
monetary policy for stabilizing the economy through controlling the interest rate based on output
and inflation is not enough. To improve significantly the performance of the economy, the Vietnam
Central Bank needs to consider adding to the factor the amount of the loan for the technology shock
(or the positive shock) or the factor of the entrepreneurs asset price for the cost-push shock (or
negative shock).
Interestingly, the results of the model applied on the Vietnamese economy are similar to the results
of the model by Gerali et al. (2010) ([7]) applied to the European economy. The results showed that
output grew very slowly after the cost-push shock (the increase of oil and electricity prices) in the year
2011; this happens to coincide with the considered model. Thus, the LATW policy could be a desirable
policy to apply in the future to stabilize and improve Vietnamese economy performance.

Acknowledgments: This research is funded by Vietnam National University Ho Chi Minh City (VNU-HCM)
under Grant Number B2015-42-01.
Author Contributions: All authors contributed equally to the paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Anh, D.N. Non-Performing LoansCase Study in Vietnam: Causes, Consequences, and Effects.
Bachelors Thesis, Uppsala University, Uppsala, Sweden, 2014.
2. Gambbarcorta, L.; Signoretti, F.M. Should Monetary Policy Lean against the Wind? An Analysis Based on a DSGE
Model with Banking; Economic Working Papers, Bank for International Settlements; Norges Bank: Oslo,
Norway, 2013; Volume 418.
3. Svensson, L.E.O. Why Leaning against the Wind Is the Wrong Monetary Policy for Sweden; Stockholm School of
Economics, Stockholm University: Stockholm, Sweden, 2014.
4. Fabio, V.; Manuel, M.F. Martins and Ines Drumond. In Financial Shocks and Optimal Monetary Policy Rules;
Bank of Finland: Helsinki, Finland, 2014; Volume 418.
5. Lambertini, L.; Mendicino, C.; Punzi, M.T. Leaning against boombust cycles in credit and housing prices.
J. Econ. Dyn. Control 2013, 37, 15001522.
6. Christiano, L.; Ilut, C.; Motto, R.; Rostagno, M. Monetary Policy and Stock Market Booms; National Bureau of
Economic Research: Cambridge, MA, USA, 2010.
7. Gerali, A.; Neri, S.; Sessa, L.; Signoretti, F.M. Credit and Baking in a DSGE model of the Euro Area. J. Money
Credit Bank. 2010, 42, 4470.
8. Bernanke, B.S.; Gertler, M.; Gilchrist, S. The Financial Accelerator in a Quantitative Business Cycle Framework.
In Handbook of Macroeconomics; Taylor, J.B., Woodford, M., Eds.; North-Holland: Amsterdam, The Netherland,
1999; pp. 13411393.
9. Iacoviello, M. House Prices, Borrowing Constraints and Monetary Policy in the Business Cycle. Am. Econ. Rev.
2005, 95, 739764.
10. Christiano, L.; Eichenbaum, M.; Evans, C. Nominal Rigidities and the Dynamic Effects of a Shock to Monetary
Policy. J. Political Econ. 2005, 113, 146.
11. Smets, F.; Wouters, R. An Estimated Dynamic Stochastic General Equilibrium Model of the Euro Area. J. Eur.
Econ. Assoc. 2013, 1, 11231175.
12. Curdia, V.; Woodford, M. Credit Spreads and Monetary Policy. J. Money Credit Bank. 2010, 42, 335.
13. Bernanke, B.; Gertler, M. Monetary Policy and Asset Price Volatility; National Bureau of Economic Research:
Cambridge, MA, USA, 2000.
14. Bernanke, B.S.; Gertler, M. Should Central Banks Respond to Movements in Asset Prices? Am. Econ. Rev.
2001, 91, 253257.
15. Gilchrist, S.; Leahy, J.V. Monetary Policy and Asset Prices. J. Monet. Econ. 2002, 49, 7597.
16. Crdia, V.; Woodford, M. Credit Frictions and Optimal Monetary Policy; National Bureau of Economic Research:
Cambridge, MA, USA, 2015.
Economies 2017, 5, 3 18 of 18

17. Andrs, J.; Arce, O.; Thomas, C. Banking competition, collateral constraints, and optimal monetary policy.
J. Money Credit Bank. 2013, 45, 87125.
18. Gali, J. Monetary Policy, Inflation, and the Business Cycle: An Introduction to the New Keynesian Framework;
Princeton University Press: Princeton, NJ, USA, 2008.
19. Christensen, I.; Corrigan, P.; Mendicino, C.; Nishiyama, S. An Estimated Open-Economy General Equilibrium
Model with Housing Investment and Financial Frictions; Bank of Norway: Mimeo, Norway, 2007.

c 2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).