Pamungkas, Raditiyo Harya-Kajian Ekonomi dan Keuangan Vol 20 No. 1 (April 2016)


KAJIAN EKONOMI & KEUANGAN
http://fiskal.depkeu.go.id/ejournal



Estimation of Indonesia’s Fiscal Reaction Function
Estimasi Fungsi Reaksi Fiskal Indonesia

Raditiyo Harya Pamungkas ∗α

Abstrak
Kebijakan fiskal merupakan faktor penting dalam strategi pengelolaan
indikator ekonomi makro. Setelah mengalami krisis keuangan, banyak negara
melaksanakan kebijakan fiskal dengan sangat berhati-hati untuk menjaga
tingkat utang. Tulisan ini menganalisis perilaku kebijakan fiskal Indonesia
melalui pendekatan fiscal reaction function, yang menggambarkan
bagaimana pemerintah merespons rasio utang terhadap Produk Domestik
Bruto (PDB) dengan menciptakan keseimbangan primer dalam Anggaran
Pendapatan dan Belanja Negara (APBN). Data yang digunakan adalah data
triwulanan dari tahun 1990 hingga 2014 dengan menggunakan analisis
Autoregressive Distributive Lag (ARDL) bound test didahului dengan
breakpoint unit root test sesuai dengan karakteristik data yang tidak
stationer pada level yang sama. Dapat diperoleh kesimpulan bahwa
pemerintah Indonesia telah merespons peningkatan rasio utang terhadap
PDB dengan meningkatkan surplus primer, yang mengindikasikan perilaku
yang rasional terhadap peningkatan akumulasi utang. Indikator-indikator
yang berpengaruh secara signifikan pada fiscal reaction function Indonesia
adalah tingkat bunga riil, nilai tukar nominal Rupiah terhadap Dolar Amerika
Serikat, dan pemilihan umum. Sebagai tambahan, untuk menjaga rasio utang
terhadap PDB pada level yang aman, pemerintah juga harus
mempertimbangkan variabel lain selain utang untuk mencapai kebijakan
fiskal yang berkelanjutan, terutama untuk mengelola guncangan pada
perekonomian.
Abstract
Fiscal policy is a core factor in managing macroeconomic indicators strategy.
Following several financial crises, both advanced and emerging countries
undertook prudent fiscal policies to maintain debt sustainability. This paper
investigates the fiscal policy behaviour of Indonesia through a fiscal reaction
function, which represents how the government reacts to the debt to GDP
ratio by the creation of primary balance in the budget. Breakpoint unit root
test is conducted due to the stationarity characteristics of data variables,
hence the widely used Autoregressive Distributive Lag (ARDL) bound test is
employed using quarterly data from 1990 to 2014. These results indicate that
the government of Indonesia has reacted to the increase in debt to GDP by
generating the primary surplus due to increase in debt accumulation which
shows the well-behaved fiscal policy to maintain debt sustainability. In
Indonesia’s fiscal reaction function, real interest rate, nominal exchange rate
to US$, and election significantly determine the primary balance behaviour.
In addition to maintaining a debt to GDP ratio at a low level, the government
should also consider the other variables other than debt to achieve
sustainability of fiscal policy especially in managing shocks.

©2016 Badan Kebijakan Fiskal Kementerian Keuangan RI
* Email:
[email protected]
α Pusat Kebijakan Ekonomi Makro,
Badan Kebijakan Fiskal,
Kementerian Keuangan. Ged. R.M.
Notohamiprodjo Lt. 8, Jl. Dr.
Wahidin No.1, Jakarta 10710

Riwayat artikel:

Diterima 6 April 2017
Direvisi 15 April 2017
Disetujui 25 April 2017


Keywords: fiscal reaction function;
ARDL model; fiscal sustainability


JEL Classification: E62; H63; C32

- Pamungkas, Raditiyo Harya


2
1. INTRODUCTION
Fiscal sustainability is a challenge for most countries in the world, both for advanced and
emerging countries. Ensuring that the fiscal policy is on the right track is a daunting task requiring
rigorous assessment. Learning from the case of Greece, poor management of debt leads to default
which can cause another problem since Greece is part of the Euro area and such a financial crisis can
create effects to its counterparts (Michaelides et al., 2014, p. 18). In other words, fiscal discipline or
ability to manage debt plays an important role in maintaining debt sustainability (Yoshino & Vollmer
2014, p. 342).
The risk of fiscal crisis been faced by industrialized countries and could be experienced by
emerging countries as well because some emerging countries still depend on foreign countries to
support economic growth. In some circumstances, this can be a vicious problem, especially under
global economic uncertainty. Some emerging countries may suffer from sudden reversal of funds.
Funds from foreign countries are preferable to fill the fund gap.
As part of the macroeconomic strategy, monetary policy is not the sole policy to support
economic growth. Solow (2005, p. 512) initiated a discussion on the role of fiscal policy. Monetary
policy is not the only policy that can help correct the economy. It needs fiscal policy to deal with
medium-term problems caused by various shocks. Indonesia, as an emerging country should have a
strategy to deal with debt sustainability. For emerging countries, the fiscal sustainability is expected
to be put in high priority. Holding the conceptual of fiscal sustainability, emerging countries have the
space for economic development.
Fiscal sustainability captures the framework of static and dynamic budget constraint
theoretically (Akyüz, 2007). Based on static budget constraint, current government spending is
financed by revenue and fund from borrowing. The theory of intertemporal budget constraint can be
determined as the present value of future primary balance is equal to outstanding debt accumulation
which satisfies the concept of solvency. Hence, the value of primary balance should be positive if there
is a debt in present time. It ensures the fiscal stability of emerging countries which have high risk
perception and vulnerability in the economy.
Considering the nature of emerging countries that still need sources of the fund including foreign
borrowing, Indonesia has to create a strategy to improve the economy without jeopardizing
sustainability, especially for creating fiscal space in the budget to provide allocation budget to develop
the economy. This requires a technical strategy to assess debt sustainability.
A fiscal reaction function is one of a commonly used method to assess debt sustainability. This
method is derived from a basic model of intertemporal budget constraint (Bohn, 1998, p. 951) which
investigates the reaction of primary surplus with respect to a change in debt accumulation of the US.
This model became popular and many scholars used it for debt sustainability assessment for
industrialized, developing, and even low-income countries. Furthermore, this model satisfies the
dynamic concept of sustainability and emphasizes the primary surplus with respect toward
indebtedness.
In this study, the fiscal reaction function of Indonesia is investigated with the identification of
fundamental determinant which also affect the reaction of fiscal policy. Quarterly data from 1990 to
2014 is employed to ascertain the nexus between debt accumulation and primary balance. This study
also evaluates the behavior of fiscal policy in Indonesia in accordance with the concept of fiscal
sustainability. The finding shows that Indonesia has a sustainability strategy to ensure a safe level of
debt accumulation. The real interest rate, nominal exchange rate, and elections are identified as the
variables which affect the fiscal policy reaction in response to indebtedness.
The structure of this paper is as follows. The remainder of reviews of previous studies is in
Section 2. Section 3 explains the method and data. This is followed by the results and analysis of
empirical research in Section 4. Section 5 concludes and proposes several policy recommendations.

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 3



2. PREVIOUS STUDIES
2.1. Fiscal Reaction Function
Basic concepts of fiscal reaction function can be understood from the Ricardian regime of fiscal
policy. The Ricardian regime is defined as a regime with the well behaved government which ensures
the liabilities constantly by generating revenue in the future to match an increase in government
spending in the previous period (Afonso, 2008, p. 314). Capturing the time dimension involves the
intertemporal approach in explaining the debt dynamics conception. This model also includes the
limitation of the budget. The concept of the intertemporal budget constraint is motivated by tax
smoothing models which explain the significant relationship between debt and primary surplus
(Barro, 1979, p. 949). Fischer and Easterly (1990, p. 135) emphasize the primary deficit as a key factor
to understand the debt dynamic intuitively. A persistent primary deficit may lead to the debt
explosion since its number is higher than the amount of seignorage, hence the number debt ratio can
rise without limit. The basic concept of public debt dynamics ensures the importance of primary
balance towards debt accumulation.
A fiscal reaction function model has been proposed by Bohn (1998, p. 950) which shows how the
government reacts to public debt accumulation in the case of the United States. The model originates
from an intertemporal budget constraint of government. The dynamics of models the government
finance is:
D
r>5=(D
r−S
r)×(1+R
r>5) (1)
where D denotes debt, S denotes tax minus non-interest spending, and R is the interest rate.
Debt in the next period is the difference between debt and non-interest spending and is the so-called
primary balance multiplied by the gross interest factor. Equation 1 shows that the primary balance is
an important factor in reducing recent debt. If the primary surplus is small, the debt in the next period
will be higher. Consequently, the debt is accumulated and tends to explode in the long run, if it is not
repaid by primary surpluses. Following this, the model has been developed by incorporating output or
income in the equation to adjust for the growing economy with growing revenues and expenditure,
hence the form is in ratio. Thus, the equation is written as:
d
r>5=(d
r−s
r)×x
r>5 (2)
where d denotes ratio of debt to output (debt to GDP ratio) and s denotes primary balance to
GDP ratio. While x represents the ratio of gross return of government debt to the gross growth rate of
output.
In this study, the model of primary balance is developed. The aim is to find the systematic
relationship between primary balance and debt accumulation. Additionally, other determinants are
incorporated in the model to avoid inconsistency in the result and estimation due to omitted variables
(Bohn, 1998, p. 951).
A fiscal reaction function has been used in debt assessment for many countries. Bohn (1998)
found that there is a positive relationship between primary balances and debt to GDP ratio in the US
(p. 956). The primary balances will increase in response to an increase in debt to GDP ratio. Other
authors have used a similar baseline model for estimation of the fiscal reaction function for the
emerging countries. Ostry and Abiad (2005) conducted panel-data analysis for emerging countries.
They incorporated other determinants as independent variables which are potentially important in
emerging countries including commodity prices, oil prices, inflation, and elections which are
important for presidential countries. They found that emerging countries respond positively to debt
accumulation. Other relevant economic indicators include business cycle, inflation, commodity and oil
prices. Mendoza and Ostry (2005, p. 1088) also found the similar results. The response of primary
balance is 3.6 basis points while the response of the industrialized economy groups is estimated 2
basis points. Although there is a different level of response between emerging and industrialized

- Pamungkas, Raditiyo Harya


4
countries, the response of primary balances to debt is generally positive (Mendoza & Ostry, 2008, p.
1088). In other words, the government reacts to a debt increase by using primary balance adjustment.
Research on fiscal reaction function has also been developed for others specific country case. De-
Mello (2005) found a positive relationship of primary balance and debt accumulation in Brazil.
Similar results were found by Budina and Wijnbergen (2008, p. 136) for the case of Turkey, Burger et
al. (2012, p. 217) for South Africa, Nguyen (2013) in the case of India, Asiama et al. (2014) for the case
of Ghana and Lestari (2014) for the case of Indonesia. All the models have been developed from Bohn’s
model with additional variables. The difference is in the variable determinants which are specified
specific characteristic countries.
Lestari (2014) found that Indonesia has run a sustainable strategy for debt. For every 1% increase
of debt, the primary balances increase 0.046%. However, simplifying the model may lead to
inaccuracies while the budget Indonesia can be affected by the other factors due to its characteristics.
Nguyen (2013) found that interest rates level affects India’s fiscal policy in addition to debt and the
output gap. Asiama et al. (2014) examined the electoral effect captured in the model using dummies
for the parliamentary and presidential election. This study seeks to analyse the behaviour of
Indonesia’s fiscal policy through primary balances of state budget towards the debt position and
identify other factors that affect the behaviour of fiscal policy. The other determinant should also
contribute to the behaviour of government toward debt in Indonesia. These factors include interest
rate, inflation, exchange rate, oil and non-oil commodity price, output gap, and electoral effects.
A model of fiscal reaction function of Indonesia has been developed by Lestari (2014) employing
a Vector Error Correction Model (VECM). Using annual data from 1992 to 2012, she found a positive
relationship of primary balance with lagged debt. The model specification is a primary balance as the
dependent variable and debt to GDP ratio and the output gap is the independent variables. For every
increase of 1% of debt to GDP ratio, the primary balance reacts by increasing 4.6 basis points, while
the coefficient of output is positive and statistically insignificant which represents weak pro-cyclical
fiscal policy.
The results show that the government of Indonesia has already run a debt sustainable strategy. A
primary balance increases to respond the debt accumulation in the previous year. By generating the
primary balance, the government pays down the debt and maintains the debt through fiscal policy at a
safe level. This response is a strategy to avoid debt exploding in the long run. In addition, the factors
that determine Indonesia’s primary balances are the lagged of primary balances themselves, debt
accumulation, exchange rate, real interest rate, and elections. These determinants are also important
and need to be controlled because of they also potentially affect the response of the primary balance,
especially for the variable which can produce shocks.
2.2. Debt Profile of Indonesia
In the 1990s, Indonesia had the relatively low debt ratio to GDP hovering around 20-30%.
However, this increased in the period of Asian financial crisis in 1998. In this period, Indonesia and
Thailand were the countries severely hit by the crisis. Moreover, there was a multidimensional crisis
because of political instability. The nominal debt accumulated of Indonesia increased from 49% to
around 90% in 2000. This nominal change of debt, especially for external debt, was caused by the
large depreciation of Rupiah to US Dollar. In this period, Indonesia also changed the exchange regime
from pegging the currency to free-floating.
After having experienced the Asian financial crisis, Indonesia conducted a macro strategy to pay
down the debt. There was a significant improvement in managing debt to achieve sustainability. The
debt to GDP ratio gradually declined and continued to decrease until around 24% in 2014 from the
high level of debt in the early 2000s. The prudent action of the government to reduce the debt shows
the effort of the government of Indonesia to maintain the level of debt during the crisis periods.
The government also diversified the debt structure both of external and domestic debt. The
diversification of debt source from domestic with domestic currency denomination can help reduce

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 5


the currency risk due to excess depreciation that may lead to uncontrolled debt. Indonesia officially
passed the 2003 State Financial Law which stated that the ratio debt to GDP can not exceed 60% and
budget deficit’s limit is maximum 3% ratio to GDP. This threshold is set in accordance with
Maastricht Treaty which is for the Euro area countries. If the level of debt or budget deficit of
Indonesia exceeds the limits, there would be a revision of the budget or special occasion to deal with
this problem. This threshold can guide Indonesia to follow sustainability of fiscal policy.
FIGURE-1: Debt to GDP Ratio Indonesia, 1990-2014

Source: Ministry of Finance of Indonesia (2015)

3. METHOD
3.1. Model and Specification
In this paper, a model of primary balance is developed from Bohn’s fiscal reaction function model.
The aim is to find the systematic relationship between primary balance and debt accumulation in the
case of Indonesia, considering other fundamental factors both economic and non-economic. These
additional factors avoid bias due to omitted variables. The reduced form of the model is specified as:
s
r=ρ.d
r+α.Z
r+ε
r (3)
where s denotes primary balance or revenue minus non-interest expenditure, d is debt
accumulation to GDP ratio, Z denotes the vector of the other determinants which varies across
countries and ε is for the error terms. In this study, eight chosen variables are incorporated in the
model to explain the behavioural impacts. These variables are considered as potential determinants
which drive the reaction of the primary balances. Thus, the final fiscal reaction function becomes:
PBR
r=α
4+α
5DEBTR
r?5+β
5GAP
r+β
6INFL
r+β
7RIR
r+β
8LNER
r+β
9WTI
r+∆
5DEL+
∆
6DAFC+ ∆
7DGFC+ε
r (4)
where PBR denotes primary balance to GDP ratio, DEBTR is debt accumulation to GDP ratio,
GAP represents output gap, INFL is inflation, RIR denotes real interest rate, and WTI represents the
West Texas Intermediate crude oil price. There are also 3 dummy variables; DEL which denotes the
dummy for election, DAFC is the dummy for Asian Financial Crisis, and DGFC is the dummy for
Global Financial Crisis. In addition, each of coefficients represents the impact of each relevant
variable in the model and α
4 denotes the intercept. The coefficient of lagged debt to GDP ratio is
denoted by α
5. For other additional determinants, the impacts are represented by the coefficients β
5,
β
6, β
7, β
8, and β
9for the variables of the output gap, inflation, real interest rate, nominal exchange rate
and WTI oil price respectively. The coefficients ∆
5, ∆
6, ∆
7 represent the implication of the
0,0%
10,0%
20,0%
30,0%
40,0%
50,0%
60,0%
70,0%
80,0%
90,0%
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

- Pamungkas, Raditiyo Harya


6
presidential and parliamentary elections, the Asian Financial crisis, and the Global Financial Crisis
respectively.
Other variables are: output gap which is added to right hand side to accommodate the
government policy of demand stabilization in the short term (Bohn, 1998, p. 951)as well as inflation
and interest rate which are added to capture the monetary impact on primary balance. It
acknowledges the relation between fiscal and monetary policies (Nguyen 2013). In this model, the
interest rate is used in the form of real interest rate.
In Indonesia’s budget structure, exchange rate and oil price are also considered. The exchange
rate change can lead to pressure because of Indonesia, as an emerging country, still requires more
foreign funds. In other words, it captures the impact of external debt services. The exchange rate
variable is the value of the domestic currency relatively compared to the US dollar. Since Indonesia
has also gained revenue from natural resources, especially oil, an oil price indicator is considered to be
incorporated in the fiscal reaction function model. The WTI crude oil price is chosen as a
representative benchmark as a proxy for the crude price oil in Indonesia.
In this study, Indonesia’s fiscal reaction function model considers the electoral effects because
the government of Indonesia has a presidential system where every five years Indonesia carries out
elections for the president and for the parliament members also. These elections have the potential to
result in budget overruns. The other dummy variables capture the effects of the crisis period on fiscal
balance. The Asian and Global Financial Crises are incorporated because they could affect the budget
structure in these periods and in the post-crises periods.
3.2. Data
Quarterly data 1990Q1 to 2014Q4 is employed in this study. These data are obtained from the
Ministry of Finance of Indonesia, Bank Indonesia, Badan Pusat Statistik Indonesia or Statistic
Indonesia Agency. Some supporting data is taken from Bloomberg and Federal Reserve Economic
Data (FRED). Table 1 shows the overall data sources.
TABLE-1: Source of Data
Data Source
Primary Balance Ministry of Finance
Nominal Debt Ministry of Finance
Nominal GDP Ministry of Finance
Output Gap Ministry of Finance generated by Hodrick-Prescott (HP) Filter
Inflation Statistics Indonesia
Exchange Rate Bank Indonesia
Real Interest Rate FRED
WTI oil price Bloomberg

The primary balances and debt data are taken from the national budget of Indonesia. Primary
balance is defined as total revenue minus total expenditure without interest expenditure while debt is
nominal debt accumulation in 1 quarter. The primary balance and debt data are used in a ratio to GDP.
The output gap is incorporated as an independent variable. It is obtained by generating value from the
nominal GDP, by using the Hodrick-Prescott (HP) Filter. Output gap data is obtained by deducting
the nominal GDP from this value. This study also uses selected data for the other determinants. For
the monetary data, inflation and real interest rate are in percentage terms while the exchange rate is in
the form of the logarithm of its nominal value. The data of WTI oil price benchmark is in US dollar
terms.
In this study, the dummy for election is constructed as having a value of 1 in the fiscal year of the
election. This is because of the characteristic quarterly data and annual budget cycles. The Asian

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 7


Financial Crisis is in the period from 1997Q3 to 1999Q4 while for the Global Financial Crisis is from
2008Q2 to 2009Q2.

4. RESULTS AND DISCUSSION
The first step of analysis in this study is a unit root test. The unit root test is conducted to know
whether data is stationary, which is useful for determining which cointegration test will be used in
the next step. In order to deal with non-stationary data, a breakpoint unit root test is employed. The
paper uses the Augmented Dickey Fuller and breakpoint test within the framework used by
Volgesang and Perron (1998, pp. 1075-1078). The framework of cointegration analysis in this paper
follows Asiama et al. (2013) and Nguyen (2013) who use the Autoregressive Distributive Lag (ARDL).
The ARDL is widely used for time series data analysis. This method of analysis was introduced by
Pesaran et al. (2001, p. 315) to accommodate the analysis of cointegration with different levels of
integration without having a bias in the result of estimation.
4.1. Unit Root Test and Breakpoint Test
Standard unit root tests are conducted before the estimation of the fiscal reaction function. The
tests check whether the variables are stationary or not. The first test is the Augmented Dickey Fuller
(ADF) test and the result can be seen in Table 2.
TABLE-2: Augmented Dickey-Fuller Stationary Result Test
Variables
5% critical
value
(with
trend)
5% critical
value
(without
trend)
t-statistic Stationary
primary balance ratio (PBR)
-3.456 -2.891
-3.983 I(0)
debt ratio (DEBTR) -1.120 I(1)
real interest rate (RIR) -5.804 I(0)
log nominal exchange rate (LNER) -1.544 I(1)
output gap (OUTGAP) -5.007 I(0)
inflation (INFL) -6.873 I(0)
WTI crude oil price (WTI) -2.810 I(1)
Source: Author’s estimation
Table 2 shows that there are three set of data that are not stationary at the level, namely debt
ratio (DEBTR), log nominal exchange rate (LNER), and WTI crude oil rice (WTI). Therefore, another
unit root test has to be considered. Following this, the stationarity is checked by the unit root test
allowing for one trend and one intercept break. A breakpoint unit root test indicates a potential break
in the non-stationary data. In this model, the breakpoint selection is obtained by minimizing Dickey-
Fuller t-statistics and the break type is an innovational outlier. All these procedures can be run in E-
Views 9 software.
TABLE-3: Break point unit root result test (allowing one trend and one intercept)
Variables
5%
critical
value
t-statistic Break Stationary
debt ratio (DEBTR)
-5.176
-6.067 1999-Q1 I(0)
log nominal exchange rate (LNER) -8.453 1997-Q4 I(0)
WTI crude oil price (WTI) -5.259 2004-Q3 I(0)
Source: Author’s estimation
The breakpoint test result is represented in Table 3. From this table, it can be seen that the t-
statistic of all variables is less than the 5% critical value which shows that by considering a structural
break, the data is stationary at level. After this, a new variable as the break is introduced which

- Pamungkas, Raditiyo Harya


8
contains the break date from all individual breakpoint selection tests. The date breaks considered as
the time will be incorporated in a break variable. However, the date break of debt ratio and exchange
rate, cover the period of the Asian Financial Crisis. Hence, the break consists of only the date break of
WTI variables. After this test, the data is available for the next analysis in estimating the fiscal
reaction function by using ARDL along with the new variable of a break.
4.2. Fiscal Reaction Function
The ARDL is commonly used in macroeconomic time series analysis. This model works well
with stationary data. In this ARDL analysis, the data are analysed considering one structural break.
The lag length criteria and model selection is considered by Akaike Information Criteria (AIC). This
study also examines the existence of the long run relationship among the relevant variables. The
results are represented in Tables 4 and 5. The ARDL analysis results in the optimal selection with the
order ARDL (3, 4, 3, 0, 0, 1, 0). The order of the model represents the optimal model for each variable.
From the result, it can be seen that the fiscal reaction function of Indonesia is determined by the
lagged the dependent variable itself, the lagged debt to GDP ratio, real interest rate, exchange rate
(domestic currency to dollar US), and elections, which are all statistically significant. While output
gap, inflation, and WTI oil price are statistically insignificant. The bound test shows that the model
cannot reject the null hypothesis which means the existence of the long run relationship. The F-
statistic is larger than the critical value of upper or I(1) bound at a significance level of 1%. The result
is presented in the Appendix.
4.2.1. Lagged Primary Balance and Debt to GDP Ratio
The lagged primary balance to GDP ratio is negatively correlated with the primary balance to
GDP ratio. This shows that the previous debt influences the following fiscal policy. If there is an
increase in the previous primary balance, the primary balance decreases. The government reacts in a
stability inducing manner three periods after the fact. The coefficient of the first period lag is positive.
TABLE-4: Results of ARDL analysis
Variable Coefficient Standard Error t-statistic
PBR(-1) -0.411*** 0.108 -3.791
PBR(-2) -0.425*** 0.095 -4.461
PBR(-3) -0.270*** 0.101 -2.676
DEBTR -0.023 0.016 -1.411
DEBTR(-1) 0.023 0.020 1.135
DEBTR(-2) -0.032 0.022 -1.471
DEBTR(-3) 0.003 0.023 0.113
DEBTR(-4) 0.047*** 0.015 3.090
GAP 0.000 0.000 0.526
GAP(-1) 0.000 0.000 0.018
GAP(-2) -0.000 0.000 -0.255
INFL -0.001 0.000 -1.085
RIR 0.002** 0.001 2.379
LNER -0.084*** 0.025 -3.383
LNER(-1) 0.076*** 0.027 2.848
WTI -0.000 0.000 -1.099
DEL -0.019*** 0.007 -2.952
DAFC 0.058*** 0.016 3.514
DGFC 0.027** 0.012 2.326
BREAK -0.003 0.021 -0.142
C 0.078 0.054 1.430
Notes: *,**, & *** denote 10, 5, & 1% significance level
Source: Author’s estimation

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 9


The government responds to the debt to GDP ratio aggressively through increasing the primary
balances. Primary balances react to the increase of debt in the four previous quarter periods. This
shows that the level of debt in 1 previous year is responded by the increase in the primary balances.
When the debt to GDP ratio increases by 1%, the primary balances to GDP ratio respond by
increasing 0.047%. This is similar to the result of Lestari (2014) who finds a reaction of 0.046%. This
number is even higher than the reaction compared to other countries. An increase in debt by 1% of
GDP is associated with an increase in 0.030% in Brazil (De-Mello 2005), 0.040% for South Africa
(Burger et al. 2012), and 0.016% in Ghana (Asiama et al. 2014). This number is also higher than the
result found by Mendoza and Ostry (2008) in their panel-data estimation of fiscal reaction function
for emerging countries. Their research found the reaction for emerging countries is 0.036% on average.
The strong response of primary balance towards debt accumulation represents the fiscal policy
requirement in the vulnerable and riskier economy (Mendoza & Ostry 2008, p. 1093). The reaction
becomes critical because it also denotes the strength of the budget to prevent a debt explosion. The
debt to GDP ratio reduction can be determined by the size of a fiscal adjustment (Ardagna 2004, p.
1047).
For the long run, there is a positive relationship between primary balances and the debt ratio but
with a lower magnitude around 0.8 basis points. It can be concluded that the government of Indonesia
has run the fiscal sustainability strategy to react properly towards the indebtedness. Even though the
relationship in the long term is weak, the positive sign ensures the effort of government to generate
the primary surplus and pay down the debt. This describes the behaviour of primary balances to avoid
debt explosion in the long run in accordance with the principle of fiscal sustainability. The
relationship shows that primary surplus ensures that there are more resources after paying the
interest payment to pay down the debt principal (Feridhanusetyawan & Pangestu 2003, p. 149).
TABLE-5: Long Run Relationship Results
Variable Coefficient Standard Error t-statistic
Debt to GDP ratio 0.008** 0.003 2.554
Output Gap 0.000*** 0.000 2.829
Inflation -0.000 0.000 -1.094
Real Interest Rate 0.001** 0.000 2.357
Log Nominal Exchange Rate -0.004 0.003 -0.917
WTI oil price 0.000 0.000 -1.116
Dummy Election -0.009*** 0.003 -3.240
Dummy AFC 0.027*** 0.008 3.586
Dummy GFC 0.013** 0.005 2.426
Break -0.001 0.010 -0.142
C 0.037 0.026 1.442
Notes: *,**, & *** denote 10, 5, & 1% significance level
Source: Author’s estimation
4.2.2. Other Determinants
In addition to debt variable, this study also examines other fundamental determinants which
potentially affect the primary balances behaviour. These other determinants must be incorporated in
the model to avoid an inconsistency on results estimation due to the omitted variables (Bohn 1998, p.
951). For a specific country case, the determination of additional variables is important because it may
differ across countries (Celasun et al. 2006). From six additional variables which are incorporated in
the model, only three independent variables are statistically significant and influence the independent
variable. These variables are the real interest rate, exchange rate, and election year, while output gap,
inflation, and an oil price of the WTI are insignificant which is associated with weak impacts.
Real interest rate is positively correlated with the primary balance ratio to GDP. The direction
between these two variables is similar because the nexus is positive. If the real interest rate increases
by 1%, the primary balance ratio will also increase by 0.002%. This shows that the primary balance

- Pamungkas, Raditiyo Harya


10
reacts to the higher real interest rate which leads to higher interest of debt services. This indicates
that greater fiscal effort is needed to guarantee sustainability fiscal. In addition, conversely, if the real
interest rate is low, it is associated with lower borrowing cost. Likewise, in the short-term, real
interest rate also affect in the long term.
Regarding of exchange rate, the nominal exchange rate affects the primary balance. Depreciation
of Indonesian Rupiah to the US dollar by 1% leads to worsening by 0.03% of the primary balance.
Although the coefficient is significant only at 10% level, currency depreciation influences the budget
through revenue and expenditure. Depreciation of the currency will improve the revenue, especially
from natural resources. However, on the spending side, fuel subsidies which are from import can be an
important factor causing excess on government spending. Generally due to depreciation, an increase
in nominal spending exceeds the nominal total revenue. Currency depreciation also changes the
structure of external debt which is denominated in foreign currencies. Therefore, the exchange rate of
the Rupiah to the US Dollar is a useful indicator in budget alteration. However, in the long run, the
impact of the exchange rate is not significant. This means that the effect of changes in exchange rate
appears only in the short run. This can be in the form of impacts. Since the exchange rate may
fluctuate, it is necessary to manage the shocks to minimise the risk.
Fiscal policy in Indonesia is not driven by inflation. In the short run and long run, inflation has a
negative correlation to the primary balance but it is insignificant. The negative sign of inflation shows
that the higher inflation, the lower the primary balances. Inflation can be an alternative to reduce
debt. Inflation may reduce the debt burden through the nominal value. Hence, if the debt decreases, it
does not require fiscal effort. Thus, the relationship between primary balance and inflation is expected
to be negative.
The output gap is not significant in determining the primary balance in the short run, but in the
long run, it has a positive coefficient although the magnitude is small. Additionally, the positive
coefficient of output gap represents pro-cyclical fiscal policy. During the Global Financial Crisis,
Indonesia decided to boost spending to improve the economy.
The variable price of crude oil WTI does not affect the primary balances. The WTI crude price
oil is assumed as a proxy for Indonesia’s oil price. This variable is accounted for in the model because
oil is an important part of revenue in the total revenue structure of Indonesia’s budget. However, the
price of oil does not impact on the primary balance. This might be caused by the difference of price
change pattern between Indonesian oil price and the WTI. Additionally, the component of the budget
which relates to oil prices is revenue from oil and gas and fuel subsidies on the expenditure side. The
sign of the negative coefficient of price WTI means that an increase in oil price tends to decrease the
primary balance because of the fuel subsidies from imports, while the revenue is affected less. This is
in line with Indonesia’s position as an oil importer country since 2004. A higher oil price is more likely
to depress the primary balances rather than increase a surplus.
4.2.3. Dummy Variables
In this model, three dummy variables are introduced to cover the impact of the presidential and
parliamentary elections, the period of Asian Financial Crisis, and the Global Financial Crisis. From the
results, it can be seen that during the year of the elections, the primary balances are lower than in the
year without election. Vergne (2009, p. 75) argues that public spending is affected by elections.
Elections lead to budget overrun because there is an additional expenditure. Government spending
increases for election rather than lower tax revenue (Schuknech, 2000, p. 115). A deficit is possible due
to lower in direct tax revenue (Katsimi & Sarantides, 2012, p. 356) or indirect taxes (Ehrhart, 2013, p.
201). In the election, governments increase expenditure which leads to a deficit for the developing
countries (Shi & Svensson, 2006, p. 1367; Ebeke & Olcer, 2013).
In the period of the Asian and Global Financial Crises, the primary balance shows a higher effort
to stabilize the debt. During the Asian Financial Crisis, Indonesia maintained prudent fiscal policies
even though they were severely affected rather than Thailand (Rosengard 2014). Generating a primary

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 11


surplus becomes a preferable strategy to reduce debt ratio to exit from the crisis rather than high
economic growth. However, a comparison of primary balance response between the two periods of
crisis shows that the reaction to Global Financial Crisis was lower than in the period of the Asian
crisis. Sangsubhan and Basri (2012, p. 250) suggest that in the 2008 crisis the impacts were limited
due to political stability and the floating exchange rate regime. Indonesia’s economy was more
resilient due to greater domestic demand (Hur et al., 2010).

5. CONCLUSION AND POLICY RECOMMENDATIONS
Since fiscal sustainability has become a critical issue for developing countries, it is necessary to
rigorously assess the debt sustainability to ensure its strategy to prevent a debt explosion in the long
run. A fiscal reaction function is a common instrument for debt assessment which investigates the
systematic relationship between primary balance and debt accumulation. Using the ARDL analysis,
this paper estimates the fiscal reaction function of Indonesia and indentifies the others fundamental
factors determinants other than debt accumulation as the independent variables. The findings show
that the government of Indonesia has run a sustainability strategy by responding to debt
accumulation through generating surplus to stabilize the debt. The response represents the effort of
fiscal policy in the short run and long run. This reaction avoids a debt explosion in the long run which
can lead to a crisis because the primary surplus counteracts the higher debt accumulation in the
previous period. A primary surplus represents the ability of government to pay down the debt
principal. It also ensures the debt ratio does not increase without limit.
In Indonesia’s fiscal reaction function, other determinants are also considered. Fundamental
factors that also contribute to the primary balance model are the real interest rate, exchange rate, and
elections. In the long run, the fiscal reaction function is determined by output gap while exchange rate
does not affect primary balance. By assessing the other fundamental factors of fiscal reaction function,
the government of Indonesia can identify the type of appropriate policy to undertake regarding the
fiscal reaction function, especially for a policy of shocks management.
The evidence indicates that government needs to undertake appropriate policy and maintain the
fiscal sustainability strategy considering the vulnerability of developing countries. This debt
assessment can be useful to support a fiscal policy proposal, such as the strategy on tax policy and
government spending structure. It also helps the government manage debt to achieve fiscal
sustainability. A debt management policy should be a high priority to ensure the fiscal sustainability
and reduce the possibility of the crisis. Indonesia has to maintain the macro strategy to support
economic growth. Fiscal discipline and fiscal reform through strict rules and tight debt management
will help the government keep the debt low. Imposing debt threshold in the law ensures attention to
raising debt which leads to better management of expenditure and revenue. However, tight rules
about debt may limit economy growing as high as the level of potential output. It requires an accurate
threshold calculation to determine the maximum debt to boost economic growth optimally. It is also
necessary to calculate the fiscal fatigue to measure the strength of the budget in general.
In addition, the Government of Indonesia has to put attention intensively to the important
variables for fiscal reaction function, especially for managing external and internal shocks.
Information about essential variables may help the government assess the sustainability problem, for
example in an optimal allocation structure of foreign currency debt, and deciding policy action that
should be undertaken under the certain circumstances. The detrimental variables should be
highlighted for which lead to worsening the primary balance. The control for significant variables can
lead to better management of the budget process. In general, this result can be used for future research
related fiscal policies, especially to overcome the practical problems in the fiscal sustainability.

6. REFERENCES
Afonso, A. (2008). Ricardian fiscal regimes in the European Union. Empirica, 35 (3), 313-334.

- Pamungkas, Raditiyo Harya


12
Akyüz, Y. (2007). Debt sustainability in emerging markets: a critical appraisal. United Nations,
Department of Economics and Social Affairs (UNDESA), viewed 24 September 2015 at
http://www.un.org/esa/desa/papers/2007/wp61_2007.pdf.
Ardagna, S. (2004). Fiscal stabilizations: when do they work and why. European Economic
Review, 48(5), 1047-1074.
Asiama, J, Akosah, N., & Owusu-Afriyie, E. (2014). An assessment of fiscal sustainability in Ghana.
Working Paper of Bank of Ghana, No. WP/BOG-2014/09, Bank of Ghana, viewed 25 July 2015, at
http://www.bog.gov.gh/privatecontent/Publications/Staff_Working_Papers/%20Fiscal%20Susta
inability%20in%20Ghana.pdf.
Barro, R. J. (1979). On the determination of the public debt. The Journal of Political Economy, 940-971.
Bohn, H. (1998). The behavior of US public debt and deficits. Quarterly Journal of Economics, 113(3), 949-
963.
Budina, N., & Van Wijnbergen, S. (2009). Quantitative approaches to fiscal sustainability analysis: a
case study of Turkey since the crisis of 2001. The World Bank Economic Review, 23(1), 119-140.
Burger, P., Stuart, I., Jooste, C., & Cuevas, A. (2012). Fiscal sustainability and the fiscal reaction
function for South Africa: assessment of the past and future policy applications. South African
Journal of Economics, 80(2), 209-227.
Celasun, O., Debrun, X., & Ostry, J.D. (2006). Primary surplus behavior and risks to fiscal sustainability in
emerging market countries: a 'fan-chart' approach. viewed 12 September 2015 at
https://www.imf.org/external/pubs/ft/wp/2006/wp0667.pdf.
De-Mello, L. (2008). Estimating a fiscal reaction function: the case of debt sustainability in Brazil.
Applied Economics, 40(3), 271-284.
Ebeke, C., & Ölçer, D. (2013). Fiscal policy over the election cycle in low-income countries. IMF
Working Paper No. WP/13/153, International Monetary Fund, viewed 4 October 2015 at
https://www.imf.org/external/pubs/ft/wp/2013/wp13153.pdf.
Ehrhart, H. (2013). Elections and the structure of taxation in developing countries. Public Choice, 156(1-
2), 195-211.
Feridhanusetyawan, T., & Pangestu, M. (2003). Managing Indonesia's debt. Asian Economic
Papers, 2(3), 128-154.
Fischer, S., & Easterly, W. (1990). The economics of the government budget constraint. The World Bank
Research Observer, 5(2), 127-142.
Hur, S.K., Jha, S., Park, D., & Quising, P. (2010). Did fiscal stimulus lift developing Asia out of the
global crisis? A preliminary empirical investigation. Asian Development Bank Economics Working Paper
Series No. 215, Asian Development Bank, viewed 3 October 2015 at
http://www.adb.org/sites/default/files/publication/28270/economics-wp215.pdf.
Katsimi, M., & Sarantides, V. (2012). Do elections affect the composition of fiscal policy in developed,
established democracies? Public Choice, 151(1-2), 325-362.
Lestari, T. (2014). Can Indonesia’s fiscal policy be sustained, with exploding debt? Working Paper in
Economics and Development Studies, No. 201415, Department of Economics, Padjadjaran University,
viewed 28 July 2015 at http://lp3e.fe.unpad.ac.id/wopeds/201415.pdf.
Mendoza, E.G., & Ostry, J.D. (2008). International evidence on fiscal solvency: is fiscal policy
“responsible”? Journal of Monetary Economics, 55(6), 1081-1093.

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 13


Michaelides, P.G., Papageorgiou, T., & Tsionas, E.G. (2014). Is the Greek crisis in the EMU
contagious? Applied Economics Letters, 21(1), 13-18.
Nguyen, T. (2013). Estimating India’s fiscal reaction function. ASARC Working Paper, No. 2013-05, The
Australian National University, Australia South Asia Research Centre, viewed 30 July 2015 at
https://crawford.anu.edu.au/acde/asarc/pdf/papers/2013/WP2013_05.pdf.Ostry, M. J.D., & Abiad,
M.A. (2005). Primary surpluses and sustainable debt levels in emerging market countries. IMF
Policy Discussion Paper No. PDP/05/6, International Monetary Fund, viewed 2 October 2015 at
https://www.imf.org/external/pubs/ft/pdp/2005/pdp06.pdf.
Pesaran, M.H., Shin, Y., & Smith, R.J. (2001). Bounds testing approaches to the analysis of level
relationships. Journal of Applied Econometrics, 16(3), 289-326.
Rosengard, J.K. (2004). Will bank bailouts bust budgets? Fiscalisation of the East Asian financial
crisis. Asian‐Pacific Economic Literature, 18(2), 19-29.
Sangsubhan, K., & Basri, M.C. (2012). Global financial crisis and ASEAN: fiscal policy response in the
case of Thailand and Indonesia. Asian Economic Policy Review, 7(2), 248-269.
Schuknecht, L. (1996). Political business cycles and fiscal policies in developing
countries. Kyklos, 49(2), 155-170.
Shi, M., & Svensson, J. (2006). Political budget cycles: do they differ across countries and why? Journal
of Public Economics, 90(8), 1367-1389.
Solow, R.M. (2005). Rethinking fiscal policy. Oxford Review of Economic Policy, 21(4), 509-514.
Vergne, C. (2009). Democracy, elections and allocation of public expenditures in developing
countries. European Journal of Political Economy, 25(1), 63-77.
Vogelsang, T.J., & Perron, P. (1998). Additional tests for a unit root allowing for a break in the trend
function at an unknown time. International Economic Review, 39(4), 1073-1100.
Yoshino, N., & Vollmer, U. (2014). The sovereign debt crisis: why Greece, but not Japan? Asia Europe
Journal, 12(3), 325-344.

- Pamungkas, Raditiyo Harya


14
APPENDIX

ARDL Results Estimation
Dependent Variable: PBR
Method: ARDL
Date: 11/12/15 Time: 03:33
Sample (adjusted): 1991Q1 2014Q4
Included observations: 96 after adjustments
Maximum dependent lags: 4 (Automatic selection)
Model selection method: Akaike info criterion (AIC)
Dynamic regressors (4 lags, automatic): DEBTR GAP INFL RIR LNER WTI

Fixed regressors: DEL DAFC DGFC C
Number of models evalulated: 62500
Selected Model: ARDL(3, 4, 3, 0, 0, 1, 0)

Variable Coefficient Std. Error t-Statistic Prob.*


PBR(-1) -0.410558 0.107641 -3.814140 0.0003
PBR(-2) -0.425512 0.094671 -4.494643 0.0000
PBR(-3) -0.269823 0.100281 -2.690674 0.0088
DEBTR -0.022679 0.016023 -1.415402 0.1611
DEBTR(-1) 0.022563 0.019787 1.140330 0.2578
DEBTR(-2) -0.032589 0.021710 -1.501082 0.1375
DEBTR(-3) 0.002972 0.023079 0.128755 0.8979
DEBTR(-4) 0.046452 0.014932 3.110877 0.0026
GAP 8.54E-08 1.66E-07 0.515180 0.6079
GAP(-1) 6.64E-09 2.23E-07 0.029861 0.9763
GAP(-2) -5.71E-08 2.16E-07 -0.264137 0.7924
GAP(-3) 3.65E-07 1.69E-07 2.155457 0.0343
INFL -0.000465 0.000425 -1.095731 0.2767
RIR 0.001551 0.000643 2.411067 0.0184
LNER -0.083619 0.024572 -3.402999 0.0011
LNER(-1) 0.076189 0.026585 2.865845 0.0054
WTI -0.000207 0.000186 -1.108748 0.2711
DEL -0.019449 0.006330 -3.072348 0.0030
DAFC 0.057758 0.016268 3.550389 0.0007
DGFC 0.026765 0.011453 2.337024 0.0221
C 0.077675 0.054099 1.435801 0.1552

R-squared 0.512839 Mean dependent var 0.012969
Adjusted R-squared 0.382929 S.D. dependent var 0.023936
S.E. of regression 0.018803 Akaike info criterion -4.918971
Sum squared resid 0.026516 Schwarz criterion -4.358020
Log likelihood 257.1106 Hannan-Quinn criter. -4.692226
F-statistic 3.947658 Durbin-Watson stat 1.932549
Prob(F-statistic) 0.000007

*Note: p-values and any subsequent tests do not account for model
selection.

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 15


Result of Long Run Relationship Model

Long Run Coefficients

Variable Coefficient Std. Error t-Statistic Prob.

DEBTR 0.007939 0.003088 2.570644 0.0121
GAP 0.000000 0.000000 2.882491 0.0051
INFL -0.000221 0.000200 -1.104585 0.2729
RIR 0.000736 0.000308 2.389690 0.0194
LNER -0.003528 0.003843 -0.918061 0.3615
WTI -0.000098 0.000087 -1.125776 0.2638
DEL -0.009235 0.002719 -3.396375 0.0011
DAFC 0.027427 0.007567 3.624594 0.0005
DGFC 0.012710 0.005213 2.437828 0.0171
C 0.036885 0.025474 1.447905 0.1518



Bound Test Result

ARDL Bounds Test
Date: 11/12/15 Time: 03:34
Sample: 1991Q1 2014Q4
Included observations: 96
Null Hypothesis: No long-run relationships exist


Test Statistic Value k

F-statistic 12.51965 6


Critical Value Bounds

Significance I0 Bound I1 Bound

10% 2.12 3.23
5% 2.45 3.61
2.5% 2.75 3.99
1% 3.15 4.43

- Pamungkas, Raditiyo Harya


16
Correlogram Result
Date: 11/12/15 Time: 03:35
Sample: 1990Q1 2014Q4
Included observations: 96
Q-statistic probabilities adjusted for 3 dynamic regressors

Autocorrelation Partial Correlation AC PAC Q-Stat Prob*

.|. | .|. | 1 0.015 0.015 0.0225 0.881
.|. | .|. | 2 -0.041 -0.041 0.1880 0.910
.|. | .|. | 3 -0.021 -0.019 0.2306 0.973
.|. | .|. | 4 0.055 0.054 0.5377 0.970
*|. | *|. | 5 -0.094 -0.098 1.4599 0.918
.|. | .|. | 6 -0.027 -0.019 1.5335 0.957
*|. | *|. | 7 -0.069 -0.075 2.0380 0.958
.|* | .|* | 8 0.191 0.189 5.9538 0.652
*|. | *|. | 9 -0.092 -0.104 6.8689 0.651
*|. | *|. | 10 -0.178 -0.174 10.326 0.412
*|. | *|. | 11 -0.157 -0.158 13.038 0.291
.|* | .|* | 12 0.130 0.102 14.919 0.246
*|. | *|. | 13 -0.121 -0.109 16.588 0.219
*|. | *|. | 14 -0.081 -0.086 17.336 0.239
.|. | *|. | 15 -0.064 -0.082 17.817 0.272
.|* | .|. | 16 0.127 0.043 19.700 0.234
*|. | *|. | 17 -0.129 -0.130 21.687 0.197
*|. | *|. | 18 -0.173 -0.179 25.278 0.117
.|. | .|. | 19 -0.018 -0.006 25.317 0.150
.|** | .|* | 20 0.215 0.092 31.052 0.055
.|. | .|. | 21 0.017 0.009 31.087 0.072
*|. | **|. | 22 -0.165 -0.230 34.536 0.043
.|. | *|. | 23 -0.042 -0.078 34.760 0.055
.|* | .|. | 24 0.152 0.030 37.772 0.037
.|. | .|* | 25 0.058 0.111 38.221 0.044
.|. | .|. | 26 -0.038 -0.039 38.414 0.055
.|. | .|. | 27 0.072 -0.001 39.125 0.062
.|* | *|. | 28 0.107 -0.091 40.713 0.057
.|. | .|. | 29 0.011 0.014 40.731 0.073
*|. | *|. | 30 -0.159 -0.076 44.348 0.044
.|. | .|. | 31 0.017 0.024 44.389 0.056
.|* | .|. | 32 0.177 0.063 49.005 0.028
.|. | .|. | 33 0.024 -0.024 49.092 0.035
*|. | .|. | 34 -0.116 -0.055 51.131 0.030
.|. | *|. | 35 -0.034 -0.114 51.310 0.037
*|. | *|. | 36 -0.075 -0.174 52.182 0.040


*Probabilities may not be valid for this equation specification.

Kajian Ekonomi & Keuangan Vol 20 No. 1 (April 2016) - 17


Normality Test Result
0
2
4
6
8
10
-0.0375-0.0250-0.01250.00000.01250.02500.03750.0500
Series: Residuals
Sample 1991Q1 2014Q4
Observations 96
Mean -1.19e-16
Median -0.001225
Maximum 0.048086
Minimum -0.040496
Std. Dev. 0.016707
Skewness -0.075272
Kurtosis 3.096575
Jarque-Bera 0.127960
Probability 0.938024


Serial Correlation Test Result
Breusch-Godfrey Serial Correlation LM Test:

F-statistic 0.211291 Prob. F(4,71) 0.9314
Obs*R-squared 1.129312 Prob. Chi-Square(4) 0.8896



Heteroskedasticity Test Result

Heteroskedasticity Test: Breusch-Pagan-Godfrey


F-statistic 1.400430 Prob. F(20,75) 0.1494
Obs*R-squared 26.10292 Prob. Chi-Square(20) 0.1624
Scaled explained SS 16.70127 Prob. Chi-Square(20) 0.6723



Model Selection Summary Table
-4.920
-4.915
-4.910
-4.905
-4.900
-4.895
-4.890
-4.885
A
R
D
L
(
3
,

4
,

3
,

0
,

0
,

1
,

0
)
A
R
D
L
(
4
,

4
,

3
,

0
,

0
,

1
,

0
)
A
R
D
L
(
3
,

4
,

3
,

0
,

0
,

1
,

1
)
A
R
D
L
(
3
,

4
,

3
,

0
,

1
,

1
,

0
)
A
R
D
L
(
3
,

4
,

3
,

1
,

0
,

1
,

0
)
A
R
D
L
(
3
,

4
,

3
,

0
,

0
,

2
,

0
)
A
R
D
L
(
3
,

4
,

4
,

0
,

0
,

1
,

0
)
A
R
D
L
(
4
,

4
,

3
,

0
,

1
,

1
,

0
)
A
R
D
L
(
4
,

4
,

3
,

0
,

0
,

1
,

1
)
A
R
D
L
(
3
,

4
,

3
,

0
,

1
,

1
,

1
)
A
R
D
L
(
3
,

4
,

3
,

0
,

0
,

2
,

1
)
A
R
D
L
(
4
,

4
,

3
,

1
,

0
,

1
,

0
)
A
R
D
L
(
3
,

4
,

3
,

0
,

0
,

1
,

2
)
A
R
D
L
(
4
,

4
,

3
,

0
,

0
,

2
,

0
)
A
R
D
L
(
3
,

4
,

3
,

1
,

0
,

1
,

1
)
A
R
D
L
(
4
,

4
,

4
,

0
,

0
,

1
,

0
)
A
R
D
L
(
3
,

4
,

3
,

0
,

1
,

2
,

0
)
A
R
D
L
(
3
,

4
,

4
,

0
,

0
,

1
,

1
)
A
R
D
L
(
3
,

4
,

3
,

0
,

0
,

3
,

0
)
A
R
D
L
(
3
,

4
,

3
,

1
,

1
,

1
,

0
)
Akaike Information Criteria (top 20 models)

- Pamungkas, Raditiyo Harya


18
Fitted and Residual Graph
-.06
-.04
-.02
.00
.02
.04
.06
-.08
-.04
.00
.04
.08
92 94 96 98 00 02 04 06 08 10 12 14
Residual Actual Fitted