Global Economics and Management Review 19 (2014) 16–25

www.elsevierciencia.com/gemrev

Original

The dynamics of insurance sector development, banking sector

development and economic growth: Evidence from G-20 countries
Rudra Prakash Pradhan a,∗ , Sahar Bahmani b , Marepalli Uday Kiran c
a
Vinod Gupta School of Management, Indian Institute of Technology, Kharagpur 721302, India
b
Department of Economics, University of Wisconsin, Parkside, Kenosha, WI 53144, USA
c
Department of Electronics and Communication, Birla Institute of Technology, Mesra 835215, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper examines the mutual relationship between banking sector development, insurance sector
Received 18 November 2014 development, and economic growth in the G-20 countries between 1980 and 2012. Our results demon-
Accepted 16 May 2015 strate that there is a long-run equilibrium relationship between these three variables. We then use a panel
Available online 27 June 2015
vector auto-regression model to reveal the nature of Granger causality among these three variables. As
expected, we find that both banking sector development and economic growth Granger cause insurance
JEL classification: sector development.
G10
G20 © 2014 INDEG/PROJECTOS- Inst. para o Desenvolvimento da Gestão Empresarial/Projectos. Published
O2 by Elsevier España, S.L.U. All rights reserved.
O16
O20
O53

Keywords:
Insurance sector development
Banking sector development
Economic growth
G-20 countries

Introduction by empirically testing the co-integrating and causal relationship

between insurance sector development, banking sector develop-
A growing literature in financial history has emphasized that the ment and economic growth. With the broad use of insurance
transition to higher and sustained economic growth rates was pre- market in the 1980s, we deploy cross-country panel data to gather
ceded by the emergence of a modern financial system, sometimes enough observations to analyze the causal relationships between
referred to as financial diffusion. The diffusion of well-managed the three.
public finances, stable money, a central bank, a banking system, The insurance sector1 has influenced the economy in every
securities markets, and a sound insurance system all play a crucial aspect (Beck & Webb, 2003; Beenstock, Dickinson, & Khajuria, 1986;
role in promoting economic development (Andersson, Eriksson, & Boon, 2005; Chang, Cheng, Pan, & Wu, 2013; Chen, Cheng, Pan,
Lindmark, 2010; Billio, Getmansky, Lo, & Pelizzon, 2012). However, & Wu, 2013; Han, Li, Moshirian, & Tian, 2010; Kugler & Ofoghi,
the insurance sector has gained less attention in the finance-growth 2005; Lee, 2011; Nektarios, 2010; Outreville, 1996; Pagano, 1993;
dilemma (Lee, Lee, & Chiu, 2013; Ward & Zubruegg, 2000). Sümegi & Haiss, 2008; Wasow & Hill, 1986). The insurance sector
In this study, we expand the works of Adams, Andersson,
Andersson, and Lindmark (2009), Allen and Santomero (2001),
Haiss and Sumegi (2008), Horng, Chang, and Wu (2012), Hussels, 1
Insurance, like other financial services, has grown in quantitative importance
Ward, and Zurbruegg (2005), Lee (2013) and Liu and Lee (2014) as part of the general development of financial institutions. The governments of
many developing countries have historically held the view that the financial systems
they inherited could not adequately serve their countries’ developmental needs.
Therefore, during the past several years, they have directed considerable efforts
∗ Corresponding author. toward changing the structure of these financial systems and controlling their oper-
E-mail addresses: [email protected] (R.P. Pradhan), ations in order to channel savings to investments, which are crucial components of
[email protected] (S. Bahmani), [email protected] (M.U. Kiran). development programs (Outreville, 1990; UNCTD, 1988).

http://dx.doi.org/10.1016/j.gemrev.2015.05.001
2340-1540/© 2014 INDEG/PROJECTOS- Inst. para o Desenvolvimento da Gestão Empresarial/Projectos. Published by Elsevier España, S.L.U. All rights reserved.
 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25 17

contributes to economic growth, both as a financial intermediary Table 1

Definition of variables.
and as a provider of risk transfer and indemnification, by allowing
different risks to be managed more efficiently and by mobiliz- Variable code Variable definition
ing domestic savings (Ward & Zubruegg, 2000). The relationship GDP Per capita economic growth: percentage change in
between insurance sector development2 and economic growth3 per capita gross domestic product.
has been extensively documented in the financial literature using BRM Broad money supply: expressed as a percentage of
an array of econometric techniques, such as cross-country, time gross domestic product.
PSC Private sector credit: expressed as a percentage of
series, panel data, and firm level studies: for example, Arena (2008),
gross domestic product.
Avram, Nguyen, and Skully (2010), Chang, Cheng, et al. (2013), LID Life insurance density: direct domestic life
Chang, Lee, and Chang (2013), Chen, Lee, and Lee (2012), Ching, premiums per capita in USD.
Kogid, and Furuoka (2010), Curak, Loncar, and Poposki (2009), Enz NID Non-life insurance density: direct domestic
non-life premiums per capita in USD.
(2000), Haiss and Sumegi (2008), Han et al. (2010), Lee (2011), Lee
TID Total Insurance density: direct domestic premiums
and Chiu (2012), Lee, Kwon, and Chung (2010), Lee, Chang, and (both life and non-life) per capita in USD.
Chen (2012), Lee, Lee, et al. (2013), Chiu, Tsong, Yang and Chang LIP Life insurance penetration: direct domestic life
(2013), Ward and Zubruegg (2000), and Webb, Grace, and Skipper premiums as a percentage of gross domestic
(2005). By and large, the empirical evidence has demonstrated that product.
NIP Non-life insurance penetration: direct domestic
there is a positive long-run association between the indicators of
non-life premiums as a percentage of gross
insurance sector development and economic growth. Many of the domestic product.
papers suggest that a well-developed insurance sector is growth- TIP Total insurance penetration: direct domestic
enhancing, and, therefore, consistent with the proposition of ‘more premiums (both life and non-life) as a percentage
of gross domestic product.
insurance, more growth’.
This study establishes that insurance sector development has Note 1: All monetary measures are in real US dollars.
a long-run equilibrium relationship with banking sector devel- Note 2: Variables above are defined in the World Development Indicators and pub-
lished by the World Bank and in World Insurance published by Sigma Economic
opment and economic growth. This paper establishes a highly Research & Consulting, Switzerland.
significant positive impact of economic growth and banking sec- Note 3: The coverage of these variables is 1988–2012.
tor development on insurance sector development for the G-20 Note 4: Insurance penetration, defined as the ratio insurance premium to gross
countries, implying that both economic growth and banking sec- domestic product, associates the scale of the insurance market with that of the econ-
omy, but ignores the population factor. On the contrary, insurance density, defined
tor development play a critical role in boosting insurance sector
as the premium per capita, takes population into consideration, but neglects eco-
development in the economies of the G-20. nomic development.
The rest of the paper is organized as follows. The next section Note 5: Insurance penetration means direct domestic premiums (for life/non-
describes the data and the model. The estimation strategy and the life/total) in USD expressed as a percentage of gross domestic product.
empirical results are discussed in “The estimation strategy and Note 6: This paper uses these six variables, one at a time, to characterize insurance
market development.
empirical results” section. The final section concludes.
Note 7: This paper uses BRM as a representative to banking sector development.
However, PSC has also been simultaneously used for some comparative analysis
Data and model only, particularly in place of BRM.

Annual data from 1980 to 2012 for the G-20 countries were
obtained from the World Development Indicators of the World Bank
insurance penetration as a percentage of gross domestic prod-
and Sigma/Economic Research & Consulting, Switzerland. The G-
uct (LIP), non-life penetration as a percentage of gross domestic
20 consists of 19 member countries plus the European Union (EU),
product (NIP), total insurance penetration as a percentage of gross
which is represented by the President of the European Council
domestic product (TIP), and broad money supply as a percentage
and by the European Central Bank. Although we look at the G-
of gross domestic product (BRM7 ). Table 1 presents the detailed
20, within this group of industrialized and developing economies,
discussion of these variables.
we observed only 17 member countries, which were used for our
All three variables were converted into their natural logarithms
analysis.4 The member countries are Argentina, Brazil, China, India,
for estimation purposes.
Indonesia, Mexico, the Russian Federation, Saudi Arabia, South
We used the following model to detect the long-run and short-
Africa, Turkey, Australia, Canada, Italy, Japan, the Korean Republic,
run causal relationship between insurance sector development
the United Kingdom, and the United States.
(ISD), banking sector development (BSD), and per capita economic
The variables used were the growth rate of real per capita
growth (GDP).
income as a percentage, life insurance5 density as a premium per
capita (LID), non-life insurance6 density as a premium per capita ISDit = ˛ISDit + ˇ1ISDi GDPit + ˇ2ISDi BSDit + εit (1)
(NID), total insurance density as a premium per capita (TID), life
where, LID, NID, TID, LIP, NIP and TIP are used as proxies for ISD;
BRM is used as a proxy for BSD; i = 1, 2,. . ., 17 represent each country
2
Insurance market development refers to a process that marks improvements in in the panel; t = 1, 2, . . ., T (1980–2012) refers to the time period;
the quantity, quality and efficiency of the insurance sector. The parameters ˇ1ISD and ˇ2ISD represent the long-run elasticity
3
The broad literature focusing on the various determinants of economic growth estimates of ISD with respect to GDP and BSD, respectively. The task
is not surveyed here. See, for example, Barro and Sala-i-Martin (1995).
4
To include the EU, the twentieth member, would have meant double-counting
four countries: France, Germany, Italy, and the United Kingdom. However, later
7
we exclude both France and Germany due to the non-availability of data for broad BRM, as a representative to banking sector development, is a cardinal to foster
money supply. long-run economic growth. There are many channels through which BRM (or private
5
Life insurance, in its general form, is guaranteed to pay a specific amount of sector credit, another banking sector development indicator) can contribute to long-
indemnification to a beneficiary after the insured’s death or to the insured who run economic growth. These channels are (a) rendering information about possible
lives beyond a certain age. investments, so as to allocate capital efficiently; (b) supervising firms and exerting
6
In contrast to life insurance, non-life insurance includes all other types of insur- corporate governance; (c) diversifying risk; (d) mobilizing and pooling savings; (e)
ance, such as homeowner’s insurance, motor vehicle insurance, marine insurance, facilitating the exchange of goods and services; and (f) easing technology transfer
liability insurance, etc. (Chen et al., 2013). (see, for instance, Pradhan, Arvin, Norman & Hall, 2014).
18 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25

is to estimate the parameters in Eq. (1) and conduct panel tests on of standard errors that can be used for inference. According to Kao
the causal nexus between these three variables. It is postulated that and Chiang (2000), both FMOLS and DOLS estimators have normal
ˇjISD > 0 (for j = 1 and 2), which suggests that increases in per capita limiting properties. The models start with the estimation of the
economic growth (GDP) and banking sector development (BSD) will following regression equation.
likely cause an increase in insurance sector development (ISD).
Ki
We expect the existence of bidirectional causality between 
these three variables [ISD–GDP–BSD]. Fig. 1 depicts the bidirec- Yit = ˛i + ˇi Xit + ik Xit − k + εit (2)
tional causal relationship between insurance sector development, k=−Ki

banking sector development, and per capita economic growth

where, Yit represents log ISD and Xit represents log GDP and log
[H1A, B –H3A, B ].
BSD. Both Yit and Xit are cointegrated with slopes ˇi , which may or
may not be homogenous across i.
The estimation strategy and empirical results Following Eq. (2), let it = ε̂it Xit be a stationary vector consist-
ing of the estimated residualsfromthe cointegrating
 regression.

T T 
Two types of tests were performed here: a panel cointegration Also let ˝it = lim T → ∞ T −1 
t=1 it

t=1 it
be the
test and a panel Granger causality test. In conducting these tests,
long-run covariance for this vector process which can be decom-
the first step of identifying the order of integration at which the
posed into ˝it = ˝it0 + i + i , where ˝it0 is the contemporaneous
variables attain stationarity is essential. Three sets of unit root tests
covariance and ii is a weighted sum of autocovariances.
were used for this purpose: the Levin–Lin–Chu (LLC; Levin, Lin, &
Thus, the panel FMOLS estimator will be given by
Chu, 2002), Augmented Dickey Fuller (ADF) and Phillips Perron (PP)
⎡
−1
T ⎤
panel unit root tests (Choi, 2001). These tests are detailed in several

N T 
advanced econometric textbooks and are not described here due to ˆ∗
ˇFMOLS = N
−1 ⎣ (Xit − X̄i )
2
(Xit − X̄i )Yit∗ − T ˆ i ⎦
space constraints.
i=1 t=1 t=1
Table 2 reports the results of the unit root tests for each variable.
(3)
Evidently, all of the eight time series variables were non-stationary
in their levels. However, they are all stationary at their first differ- where Yit∗ ˆ 21i /˝
= (Yit − Ȳ ) − (˝ ˆ 22i )Xit and ˆ i = ˆ 21i + ˆ 0 −
21i
ence at the 1% level of significance. In other words, all the variables ˆ 21i /˝
(˝ ˆ 22i )(T̂22i /˝
ˆ0 )
22i
are integrated of order one (denoted by I (1)) at the individual coun- On the contrary, with the DOLS approach, as developed by Kao
try and panel levels. Naturally, I (1) meets the requirements of the and Chiang (2001) and Mark and Sul (2003), the concerning esti-
cointegration test. mator is coming from Eq. (2) which includes advanced and delayed
In the subsequent step, we deployed the Johansen cointegration values (XiT ) in the cointegrated relationship, in order to eliminate
test (Maddala & Wu, 1999), using both trace statistic and maximum the correlation between the regressor and the error terms. Hence
eigen values, at the panel setting to check for the existence of coin- the panel DOLS estimator can be defined as
tegration between ISD (LID/NID/TID/LIP/NIP/TIP), BSD and GDP. The ⎡
−1
T ⎤
discussion of this test is not available here due to space constraints 
N T 
and can be obtainable upon request. ˆ∗
ˇDOLS = N
−1 ⎣ Zit Zit Zit Ŷit ⎦ [4]
We investigate six different cases by using each insurance sector i=1 t=1 t=1
indicator (LID/NID/TID/LIP/NIP/TIP) separately with BSD and GDP.
The results of this test are reported in Table 3. The estimated results where, Zit = [Xit − X̄, Xit−ki , ....Xt+Ki ] is the vector of the regres-
indicate that there exists one significant cointegrating vector in sor, and Ŷit = [Yit − Ȳ ].
each case. It means that our three variables (ISD, BSD and GDP) To carry out tests on the cointegrated vectors, it is consequently
are linked together by a long-run equilibrium relationship. In other necessary to use methods of effective estimation.
words, we can argue that insurance sector development has a long- The results of this section are presented in Table 4.
run equilibrium relationship with banking sector development and We are mostly interested in knowing the nature of the relation-
economic growth. ship, be it positive or negative, of the variables. It can be seen that
Having confirmed the existence of cointegration of our panel, both GDP and BSD exercise a significant positive influence on ISD in
the next step is to estimate the associated long-run cointegration the long-run. The finding of the presence of a highly significant pos-
parameters. Although the ordinary least squares (OLS) estimators itive impact of economic growth and banking sector development
of the cointegrated vectors are super-convergent, their distribu- on insurance sector development for the G-20 countries implies
tion is asymptotically biased and depends on nuisance parameters that both economic growth and banking sector development play
associated with the presence of serial correlation in the data (see, a critical role in boosting insurance sector development in the
for instance, Pedroni, 2001). Many types of problems existing in economies of the G-20. For instance, the panel long-run growth
the time series analysis may also arise for the panel data analysis elasticity is 0.62 in Case 1 (see Table 4, row 1 and column 1). This
and tend to be more marked, even in the presence of heterogeneity implies that a 1% increase in economic growth can increase insur-
(Kao and Chiang, 2001). For this reason, several estimators have ance market development by 0.62%. This is exactly the same in Case
been proposed. The study uses two panel cointegration estima- 2 and 0.65%, 0.23%, 0.21% and 0.27% in Cases 3–6. Similarly, a 1%
tors: the between group fully modified OLS (FMOLS8 ) and dynamic increase in banking sector development can increase the insurance
OLS (DOLS9 ). Both FMOLS and DOLS provide consistent estimates market development by 0.39–2.05%, in all the occasions (Cases 1–6;
Table 4).
Engle and Granger (1987) demonstrated that when variables are
8
FMOLS is a non-parametric approach, taking into account the possible correla- cointegrated, an error correction model necessarily describes the
tion between the error term and the first differences of the regressor as well as the data-generating process. Therefore, on the basis of the unit root and
presence of a constant term to deal with corrections for serial correlation (see, for cointegration test results above, the vector error-correction mod-
instance, Maeso-Fernandez et al., 2006; Pedroni, 2000, 2001).
9
DOLS is a parametric approach, which adjusts the errors by augmenting the
els, VECMs, were used to determine the causal relations between
static regression with leads, lags, and contemporaneous values of the regressor in the variables. In other words, we sought to determine what vari-
first differences (see, for instance, Kao & Chiang, 2000). able caused the other in the presence of the third variable. We were
 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25 19

H2A

H1A H3A

GDP
ISD BSD

H1B H3B

H2B

Note 1: BSD: Banking sector development; ISD: Insurance sector development; and GDP: Per capita economic growth rate.
Note 2:
H1A, B: Insurance sector development Granger-causes economic growth and vice versa
H2A, B: Insurance sector development Granger-causes banking sector development and vice versa.
H3A, B: Banking sector development Granger-causes economic growth and vice versa

Fig. 1. Possible causality between insurance sector development, banking sector development and economic growth.

able to determine this causal link for both the short-run and the 
p

q

long-run. Following the Arellano and Bond (1991) and Holtz-Eakin, ISDit = ISDj + 1ISDik ISDit−k + 2ISDik BSDit−k
Newey, and Rosen (1988) estimation procedures, we deployed the k=1 k=1
following VECMs to trace the causal links between insurance devel-
opment, banking sector development and economic growth.

r
+ 3ISDik GDPit−k + ISDi ECTit−1 + ε3it (7)
k=1

p

q

GDPit = GDPj + 1GDPik GDPit−k + 2GDPik BSDit−k where,  is the first difference operator; i represents the country
k=1 k=1 in the panel (i = 1, 2, . . ., N); t denotes the year in the panel (t = 1, 2,

r . . .., T); p, q, and r are lag lengths for the differenced variables of the
+ 3GDPik ISDit−k + GDPi ECTit−1 + ε1it (5) respective equations and can be determined by the Engle–Granger
approach;
k=1
The subscript ‘i’ was removed from the equation when we con-
ducted individual country analysis; εit is a normally distributed
random error term for all i and t with a zero mean and a finite

p

q heterogeneous variance;
BSDit = BSDIj + 1BSDik BSDit−k + 2BSDik GDPit−k The ECTs are error-correction terms, derived from the cointe-
k=1 k=1
grating equations. The ECTs represent the long-run dynamics, while
differenced variables represent the short-run dynamics between

r
the variables. Put differently, ECTs indicate the extent of the devi-
+ 3BSDik ISDit−k + BSDi ECTit−1 + ε2it (6) ation from the long-run equilibrium which was present in the
k=1 previous period. The coefficients of ECT fulfill the role of the

Table 2
Results of panel unit root tests.

LLC ADF PP

Test Statistics Level Fist difference Level Fist difference Level Fist difference

GDP −0.40 −22.9 *

15.9 447.0 *
18.5 246.4*
BRM 4.40 −11.9* 9.64 219.1* 11.5 351.8*
LID 4.94 −10.6* 5.26 175.0* 3.87 238.4*
NID 4.14 −9.35* 1.92 161.3* 1.41 242.9*
TID 5.32 −9.71* 1.48 163.1* 1.02 233.0*
LIP −0.21 −11.8* 42.7 199.5* 62.7 261.3*
NIP −0.08 −13.3* 45.1 235.6* 64.2 346.1*
TIP 2.42 −12.1* 11.4 203.9* 13.4 285.6*

Note 1: GDP: per capita economic growth rate; BRM: broad money supply; LID: life insurance density; NID: non-life insurance density; TID: total insurance density; LIP: life
insurance premium; NIP: non-life insurance premium; TIP: total insurance premium.
*
Rejection of the null hypothesis at the 0.01 level.
20 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25

Table 3
Results of panel cointegration test.

Hypothesized no. of CE (s) Trace statistic Prob. Max-eigen statistic Prob.

Case 1: GDP, BRM, LID

None (k ≤ 0)* 102.5 0.00 85.85 0.00
At most 1 (k ≤ 1) 48.51 0.15 42.15 0.15
At most 2 (k ≤ 2) 39.02 0.25 39.02 0.25

Case 2: GDP, BRM, NID

None (k ≤ 0)* 104.4 0.00 87.33 0.00
At most 1 (k ≤ 1) 49.83 0.04 45.75 0.04
At most 2 (k ≤ 2) 35.19 0.41 35.19 0.41

Case 3: GDP, BRM, TID

None (k ≤ 0)* 104.3 0.00 86.23 0.00
At most 1 (k ≤ 1) 50.00 0.04 45.50 0.04
At most 2 (k ≤ 2) 38.03 0.29 38.03 0.29

Case 4: GDP, BRM, LIP

None (k ≤ 0)* 96.03 0.00 81.98 0.00
At most 1 (k ≤ 1) 43.87 0.12 40.24 0.12
At most 2 (k ≤ 2) 34.21 0.46 34.21 0.46

Case 5: GDP, BRM, NIP

None (k ≤ 0)* 134.8 0.00 112.7 0.00
At most 1 (k ≤ 1) 58.44 0.05 51.44 0.05
At most 2 (k ≤ 2) 34.46 0.46 34.46 0.46

Case 6: GDP, BRM, TIP

None (k ≤ 0)* 91.55 0.00 72.14 0.00
At most 1 (k ≤ 1) 50.64 0.03 45.39 0.03
At most 2 (k ≤ 2) 35.77 0.39 35.77 0.39

Note 1: GDP: per capita economic growth rate; BRM: broad money supply; LID: life insurance density; NID: non-life insurance density; TID: total insurance density; LIP: life
insurance premium; NIP: non-life insurance premium; TIP: total insurance premium.
*
Rejection of the null hypothesis at the 0.01 level.

adjustment parameter, which show the proportion of the disequi- that there is Granger causality running from BSD to GDP, or from
librium that is recovered during the subsequent period. On the GDP to BSD. Similarly, if the joint null hypothesis of 3GDP = 0 in
contrary, the coefficients of the lagged-first differences provide an Eq. (5) (or 3ISD = 0 in Eq. (7)) is rejected, then Granger causal-
indication of the short-run relationship between the endogenous ity is running from ISD to GDP, or from GDP to ISD. Again, if the
variables (Harris & Sollis, 2006; Enders, 2004). joint null hypothesis of 3BSD = 0 in Eq. (6) (or 2ISD = 0 in Eq. (7)) is
For short-run causal relationships, if the null hypothesis of rejected, then Granger causality is running from ISD to BSD, or from
2GDP = 0 in Eq. (5) (or 2BSD = 0 in Eq. (6)) is rejected, this implies BSD to ISD. We use both Akaike Information Criterion and Schwarz

Table 4
Panel FMOLS and DOLS results.

Dependent variables Independent variables Panels

FMOLS DOLS

Coefficients t-Statistic Coefficients t-Statistic

Case 1: GDP, BRM, LID

LID GDP 0.62 2.53* 1.51 3.00*
BRM 2.05 9.50* 2.00 8.54*

Case 2: GDP, BRM, NID

NID GDP 0.62 4.42* 0.97 3.51*
BRM 1.56 13.3* 1.51 11.9*

Case 3: GDP, BRM, TID

TID GDP 0.65 3.84* 1.22 3.71*
BRM 1.80 12.7* 1.75 11.4*

Case 4: GDP, BRM, LIP

LIP GDP 0.23 2.24* 0.86 2.27*
BRM 0.88 5.55* 0.86** 4.88*

Case 5: GDP, BRM, NIP

NIP GDP 0.21 3.55* 0.32 2.84*
BRM 0.39 7.99* 0.37 6.99*

Case 6: GDP, BRM, TIP

TIP GDP 0.27 2.85* 0.51 2.58*
BRM 0.82 6.69* 0.82 5.45*

Note 1: GDP: per capita economic growth rate; BRM: broad money supply; LID: life insurance density; NID: non-life insurance density; TID: total insurance density; LIP: life
insurance premium; NIP: non-life insurance premium; TIP: total insurance premium.
Note 2: * and ** denote rejection of the null hypothesis at the 0.01 and 0.05 levels.
 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25 21

Table 5
Results of panel granger causality test.

Dependent Independent variables (possible ECT−1 coefficient variable (for possible

sources of causation) long run causality)

Case 1: GDP, BRM, LID

GDP BRM LID ECT−1
GDP – 10.89* 5.72* −0.62*
[–] [0.00] [0.01] (−6.78)
BRM 14.4* – 4.12** 0.25
[0.00] [–] [0.05] (5.19)
LID 11.9* 5.69* – −0.23**
[0.00] [0.01] [–] (−3.21)

Case 2: GDP, BRM, NID

GDP BRM NID ECT−1
GDP – 8.26* 1.80 −074*
[–] [0.00] [0.77] (−6.81)
BRM 12.1* – 3.99** 0.25
[0.00] [–] [0.05] (4.41)
NID 5.94* 9.60* – −0.07
[0.01] [0.00] [–] (−1.07)

Case 3: GDP, BRM, TID

GDP BRM TID ECT−1
GDP – 8.15* 1.54 −0.66*
[–] [0.00] [0.80] (−6.09)
BRM 14.5* – 5.80** 0.26
[0.00] [–] [0.05] (4.63)
TID 9.57* 9.60* – −0.03
[0.01] [0.00] [–] (−0.40)

Case 4: GDP, BRM, LIP

GDP BRM LIP ECT−1
GDP – 9.15* 1.50 −0.59*
[–] [0.00] [0.82] (−6.98)
BRM 12.6** – 3.16** 0.20
[0.00] [–] [0.05] (4.67)
LIP 12.3* 7.07* – −0.16**
[0.00] [0.00] [–] (−3.10)

Case 5: GDP, BRM, NIP

GDP BRM NIP ECT−1
GDP – 8.72* 6.12* −0.69*
[–] [0.00] [0.00] (−7.64)
BRM 13.0* – 3.89** 0.21
[0.00] [–] [0.05] (4.53)
NIP 16.6* 22.3* – −0.05
[0.00] [0.00] [–] (−1.67)

Case 6: GDP, BRM, TIP

GDP BRM TIP ECT−1
GDP – 7.20* 1.79 −0.63*
[–] [0.00] [0.77] (−7.06)
BRM 13.6* – 3.94** 0.21
[0.00] [–] [0.05] (4.77)
TIP 2.87** 9.18* – −0.01
[0.05] [0.01] [–] (−0.06)

Note 1: GDP: per capita economic growth rate; BRM: broad money supply; LID: life insurance density; NID: non-life insurance density; TID: total insurance density; LIP: life
insurance premium; NIP: non-life insurance premium; TIP: total insurance premium.
Note 2: VECM: vector error correction model; ECT: error correction term.
Note 3: Values in squared brackets represent probabilities for F-statistics.
Note 4: Values in parentheses represent t-statistics.
Note 5: Basis for the determination of long run causality lies in the significance of the lagged ECT coefficient.
*
Rejection of the null hypothesis at the 0.01 level.
**
Rejection of the null hypothesis at the 0.05 level.

Information Criterion to determine the optimal lag lengths of the From Table 5, when we considered GDP as the dependent
VECMs. variable, we found the existence of long-run causal relationships
For long-run causal relationships, the null hypothesis ( GDP = 0, between economic growth, insurance sector development and
BSD = 0, and ISD = 0) needed to be rejected. The above tests banking sector development. The estimated lagged ECTs (Cases
can be achieved via a Wald test. The results of the Granger 1–6) all carry negative signs, as expected. This implies that the
causality test are shown in Tables 5 and 6. Table 5 reports change in the level of economic growth (GDP) rapidly responds
both short-run and long-run causality, based on the signifi- to any deviation in the long-run equilibrium (or short-run dis-
cance of the lagged ECT coefficients, while Table 6 reports the equilibrium) for the t − 1 period. In other words, the effect of an
summary of short-run causality, based on the significance of instantaneous shock to insurance market development and bank-
Wald-statistics. ing sector development on economic growth will be completely
22 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25

Table 6
The summary of short-run granger causality.

Causal relationships tested ISD vs. GDP BSD vs. GDP ISD vs. BSD
in the models

Case 1: GDP–BRM–LID LID ↔ GDP BRM ↔ GDP LID ↔ BRM

Case 2: GDP–BRM–NID NID ← GDP BRM ↔ GDP NID ↔ BRM
Case 3: GDP–BRM–TID TID ← GDP BRM ↔ GDP TID ↔ BRM
Case 4: GDP–BRM–LIP LIP ← GDP BRM ↔ GDP LIP ↔ BRM
Case 5: GDP–BRM–NIP NIP ↔ GDP BRM ↔ GDP NIP ↔ BRM
Case 6: GDP–BRM–TIP TIP ← GDP BRM ↔ GDP TIP ↔ BRM

Note 1: GDP: per capita economic growth rate; BRM: broad money supply; LID: life insurance density; NID: non-life insurance density; TID: total insurance density; LIP: life
insurance premium; NIP: non-life insurance premium; TIP: total insurance premium.
Note 2: X → Y means variable X Granger causes variable Y; X ← Y means variable Y Granger causes X; and X ↔ Y means both variables Granger cause each other; NA: no
causality happening between the two variables.

adjusted in the long-run. The return to equilibrium, however, require various new financial services, thus leading to the creation
occurs at different rates: 62% in Case 1, 74% in Case 2, 66% in Case of modern financial institutions and related financial services.
3, 59% in Case 4, 69% in Case 5, and 63% in Case 6. In Case 3, we establish the presence of bidirectional causality
The long-run causal relationships between economic growth, between broad money supply and economic growth [BRM ↔ GDP]
insurance sector development and banking sector development and total insurance density and broad money supply [TID ↔ BRM].
also exist when LID and LIP were used as the dependent vari- Moreover, we find the unidirectional causality from economic
ables. The speeds of adjustment for these two cases are 23% and 16% growth to total insurance density [GDP → TID]. These results are
for Case 1 and Case 4, respectively. However, when we considered very similar to those of Case 2 [NID–GDP–BRM].
BSD or ISD (for NID, TID, NIP and TIP) as the dependent In Case 4, we detect the continuation of bidirectional
variable, we found non-uniform results. That means the ECTs are causality between broad money supply and economic growth
not statically significant. The statistical insignificance of the ECTs [BRM ↔ GDP] and life insurance penetration and broad money sup-
suggests that economic growth (or BSD or ISD) does not respond to ply [LIP ↔ BRM]. In addition, we find unidirectional causality from
deviations from long-run equilibrium. economic growth to life insurance penetration [GDP → LIP]. These
From these results, the common insight is that both insurance results are very similar to those of Case 2 [NID–GDP–BRM] and Case
sector development and banking sector development drive eco- 3 [TID–GDP–BRM].
nomic growth in the G-20 countries. In Case 5, we come across the presence of bidirectional causality
In contrast to the long-run Granger causality results, our study between broad money supply and economic growth [BRM ↔ GDP],
reveals a wide spectrum of short-run causality results between the non-life insurance penetration and economic growth [NIP ↔ GDP],
three variables (ISD, BSD and GDP). These results are summarized and between non-life insurance penetration and broad money
in Table 6 and are presented below. supply [NIP ↔ BRM]. These results are very similar to Case 1
In Case 1, we find the existence of bidirectional causality [LID–GDP–BRM].
between broad money supply and economic growth [BRM ↔ GDP], In Case 6, we find the existence of bidirectional causality
life insurance density and economic growth [LID ↔ GDP], and between broad money supply and economic growth [BRM ↔ GDP]
between life insurance density and broad money supply and total insurance penetration and broad money supply
[LID ↔ BRM]. This gives the evidence of two-way Granger causal- [TIP ↔ BRM]. Furthermore, we find unidirectional causality from
ity (feedback) between BRM and GDP, LID and GDP, and LID and economic growth to total insurance penetration [GDP → TIP].
BRM. That means there is support for both the supply-leading10 These results are very similar to Case 2 [NID–GDP–BRM], Case 3
hypothesis and the demand-following11 hypothesis. The inference, [TID–GDP–BRM] and Case 4 [LIP–GDP–BRM].
particularly for the LID–GDP nexus, is that insurance sector devel- The above discussion, as per Arellano and Bond (1991) and
opment and economic growth are endogenous. This indicates that Holtz-Eakin et al. (1988) estimation procedures, is one way of
they mutually cause each other and that their reinforcement may checking the direction of Granger causality between insurance
have important implications for the development of financial and sector development, banking sector development and economic
economic policies. growth. Nonetheless, this estimation procedure does not provide
In Case 2, we find the existence of bidirectional causality much information on how each variable responds to innovations
between broad money supply and economic growth [BRM ↔ GDP] in other variables, or whether the shock is permanent or not. To
and non-life insurance density and broad money supply avoid this shortcoming, we employ generalized impulse response
[NID ↔ BRM]. Furthermore, we find the unidirectional causality functions (GIRFs), developed by Koop, Pesaran, and Potter (1996),
from economic growth to non-life insurance density [GDP → NID]. to trace the effect of a one-off shock to one of the innovations
This supports the feedback relationship between banking sector on the current and future values of the endogenous variables.
development and economic growth as well as between bank- The GIRFs offer additional insight into how shocks to insurance
ing sector development and insurance sector development. In development can affect and be affected by each of the other
addition, it supports the demand-following hypothesis between two variables: economic growth and banking sector development.
insurance sector development and economic growth, implying These results are not available here due to space constraints but
that as real per capita growth increases, investors and savers will can be obtained upon request. Furthermore, we have also tested
the results between economic growth, insurance sector develop-
ment and banking sector development, where we replace private
sector credit with broad money supply. The findings are almost
10
The supply-leading hypothesis, for the finance-growth nexus, implies more similar, like the case of the inclusion of broad money supply. We
financial services (banking/insurance) leads to more economic growth.
11
The demand-following hypothesis implies more economic growth leads to an
just report the Granger causality results of this portion in the
increase in the demand for financial services. Appendix A.
 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25 23

Even though the majority of the results12 are similar, still premium. However, for life insurance density and non-life insur-
some minor differences exist, particularly between insurance sec- ance premium, they reinforce each other, e.g., the existence of a
tor development and economic growth. In general, it is worthwhile feedback relationship between economic growth and insurance
to note that our panel short-run Granger causality results seem sector development.
to be in line with some of the previous studies. For instance, the The main message from this study for policy-makers is that
findings of bidirectional Granger causality between banking sector inferences drawn from research on economic growth that excludes
development and economic growth is well supported by the find- the dynamic interrelation between insurance sector development
ings of Pradhan, Arvin, Norman and Hall (2014), Pradhan, Arvin, and banking sector development may be unreliable. Policymakers
Norman, and Nishigaki (2014),13 Wolde-Rufael (2009),14 Craigwell, need to eagerly reform the financial system in these G-20 countries
Downes, and Howard (2001),15 and Ahmed and Ansari (1998).16 in order to strengthen the cooperative relationship between the
The results of bidirectional Granger causality between banking sec- insurance sector and the banking sector and achieve their interac-
tor development and insurance sector development are in line with tive effects on economic growth.
Liu and Lee (2014),17 and Vadlamannati (2008).18 The finding of Following the results of this empirical investigation, the impli-
unidirectional Granger causality from economic growth to insur- cation of studies of this sort is that, a policy framework aimed
ance sector development is well supported by the findings of Horng at increasing banking sector development and economic growth
et al. (2012),19 and Vadlamannati (2008). Furthermore, the finding might help increase insurance sector development. The develop-
of bidirectional Granger causality economic growth and insurance ment of the insurance sector can also be utilized properly for
sector development is in line with the findings of Chang, Cheng, obtaining more banking sector development and achieving sustain-
et al. (2013), Chang, Lee, et al. (2013),20 Vadlamannati (2008), and able economic growth in the G-20 countries. For instance, an active
Ward and Zubruegg (2000).21 and competitive insurance sector can enable these economies in
stimulating savings, catering an alternate source of investment,
Conclusion reinforcing the development of the stock market and bond mar-
ket, mitigating risks associated to volatility in capital inflows, and
The financial system has become noticeably more complex over can shift government burden in supporting large pension schemes
the past couple of decades as the separation between hedge funds, to employee insurance-supported retirement schemes.
mutual funds, insurance companies, banks, and brokers/dealers has The important requirement is that the government of these
blurred thanks to financial diffusion and deregulation. While such respective countries should encourage financial products innova-
complexities are an unavoidable consequence of competition and tion and the diffusion of combining insurance markets with the
economic growth, it is accompanied by certain advantages, includ- banking sector, in order to promote collaborative developments of
ing much greater interdependence (Billio et al., 2012). both the insurance and the banking sector. Furthermore, for achiev-
This paper investigates the interdependence between insurance ing sustainable economic growth, it is advantageous to bring more
market development, banking sector development, and economic reform in both the insurance sector and the banking sector in order
growth. Using panel data relating to G-20 countries from 1988 to improve information flow, better service delivery and to enhance
to 2012, we find that both banking sector development and eco- competition. It is the conjoint interaction between these variables
nomic growth exercise a significant positive effect on insurance that distinguishes our study and guides future research on this
sector development. The study additionally finds that insurance topic.
market development, banking sector development, and economic
growth are cointegrated. The panel Granger causality tests fur- Acknowledgments
ther confirm that both banking sector development and economic
growth Granger-cause insurance sector development in the long We thank the Editor and an anonymous referee for their com-
run. However, in the short-run, we detect the bidirectional causal- ments and suggestions.
ity between banking sector development and economic growth and
between banking sector development and insurance sector devel-
opment. We also notice economic growth Granger causes insurance Appendix A.
sector development, particularly for non-life insurance density,
total insurance density, life insurance premium and total insurance See Table A1.

12
The definition of the variables is not always uniform across the studies.
13
The study dealt with 26 ASEAN Regional Forum countries for the period
1961–2012.
14
The study was in Kenya for the period 1966–2005.
15
The study was in Barbados for the period 1974–1998.
16
The study dealt with three Asian countries: India, Pakistan and Sri Lanka for the
period 1973–1991.
17
The study was in China for the period 1985–2011.
18
The study was in India for the period 1980–2006.
19
The study was in Taiwan for the period 1961–2006.
20
The study dealt with eight Eastern Asian countries, namely India, Indonesia,
Japan, Malaysia, the Philippines, Singapore, South Korea, and Thailand, for the period
1979–2008.
21
The study included nine OECD countries, namely Australia, Austria, Canada,
France, Italy, Japan, Switzerland, the United Kingdom, and the United States, for
the period 1961–1996.
24 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25

Table A1
Results of panel granger causality test (by replacing BRM with PSC).

Dependent variable Independent variables (possible ECT−1 coefficient (for possible long run
sources of causation) causality)

Case 1: GDP, PSC, LID

GDP PSC LID ECT−1
GDP – 6.06* 5.20* −0.50*
[–] [0.00] [0.00] (−5.38)
PSC 6.01* – 3.65** 0.23
[0.00] [–] [0.05] (4.85)
LID 10.3* 6.81* – −0.23**
[0.00] [0.01] [–] (−3.19)

Case 2: GDP, PSC, NID

GDP PSC NID ECT−1
GDP – 3.47** 1.38 −067*
[–] [0.05] [0.85] (−6.07)
PSC 4.03** – 3.89** 0.19
[0.05] [–] [0.05] (3.30)
NID 6.19* 8.83* – −0.05
[0.00] [0.00] [–] (−0.84)

Case 3: GDP, PSC, TID

GDP PSC TID ECT-1
GDP – 3.31** 0.74 −0.59*
[–] [0.05] [0.95] (−5.34)
PSC 5.81* – 4.58** 0.19
[0.00] [–] [0.05] (3.42)
TID 8.29* 7.54* – −0.01
[0.00] [0.00] [–] (−0.16)

Case 4: GDP, PSC, LIP

GDP BRM LIP ECT−1
GDP – 5.39* 1.62 −0.51*
[–] [0.00] [0.80] (−5.79)
BRM 5.10* – 3.74** 0.18
[0.01] [–] [0.05] (4.11)
LIP 9.73* 8.44* – −0.12**
[0.00] [0.00] [–] (−3.36)

Case 5: GDP, PSC, NIP

GDP PSC NIP ECT-1
GDP – 5.25* 5.86* −0.60*
[–] [0.00] [0.00] (−6.29)
PSC 5.07* – 3.83** 0.22
[0.01] [–] [0.05] (4.63)
NIP 15.1* 23.5* – −0.06
[0.00] [0.00] [–] (−1.91)

Case 6: GDP, PSC, TIP

GDP PSC TIP ECT−1
GDP – 5.78* 2.35 −0.55*
[–] [0.00] [0.67] (−5.99)
PSC 5.11* – 3.71** 0.19
[0.01] [–] [0.05] (4.28)
TIP 5.25* 13.5* – −0.02
[0.00] [0.00] [–] (−0.36)

Note 1: GDP: per capita economic growth rate; BRM: broad money supply; PSC: private sector credit; LID: life insurance density; NID: non-life insurance density; TID: total
insurance density; LIP: life insurance premium; NIP: non-life insurance premium; TIP: total insurance premium.
Note 2: VECM: vector error correction model; ECT: error correction term.
Note 3: Values in squared brackets represent probabilities for F-statistics.
Note 4: Values in parentheses represent t-statistics.
Note 5: Basis for the determination of long run causality lies in the significance of the lagged ECT coefficient.
*
Rejection of the null hypothesis at the 0.01 level.
**
Rejection of the null hypothesis at the 0.05 level.

References Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo
evidence and an application to employment fluctuations. Review of Economic
Adams, M., Andersson, J., Andersson, L. F., & Lindmark, M. (2009). Commercial Studies, 58(2), 277–297.
banking, insurance and economic growth in Sweden between 1830 and 1938. Arena, M. (2008). Does insurance market activity promote economic growth? A
Accounting, Business and Financial History, 19(1), 21–38. cross-country study for industrialized and developing countries. Journal of Risk
Ahmed, S. M., & Ansari, M. I. (1998). Financial sector development and eco- and Insurance, 75(4), 921–946.
nomic growth: The south-Asian experience. Journal of Asian Economics, 9(3), Avram, K., Nguyen, Y., & Skully, M. (2010). Insurance and economic growth: A cross
503–517. country examination. Working paper. In Australian Centre for Financial Studies-
Allen, F., & Santomero, A. (2001). What do financial intermediaries do? Journal of Finsia Banking and Finance Conference Australia.
Banking and Finance, 25(2), 271–294. Barro, R. J., & Sala-i-Martin, X. (1995). Economic growth. New York: McGraw Hill.
Andersson, L. A., Eriksson, L., & Lindmark, M. (2010). Life insurance and income Beck, T., & Webb, I. (2003). Economic, demographic, and institutional determinants
growth: The case of Sweden 1830–1950. Scandinavian Economic History Review, of life insurance consumption across countries. The World Bank Economic Review,
58(3), 203–219. 17(1), 51–88.
 R.P. Pradhan et al. / Global Economics and Management Review 19 (2014) 16–25 25

Beenstock, M., Dickinson, G., & Khajuria, S. (1986). The determination of life Lee, C. C., Chang, C. P., & Chen, P. F. (2012). Further evidence on property–casualty
premiums: An international cross-section analysis 1970–1981. Insurance, Math- insurance premiums: Do multiple breaks and country characteristics matter?
ematics and Economics, 5(1), 616–634. Japan and the World Economy, 24(3), 215–226.
Billio, M., Getmansky, M., Lo, A. W., & Pelizzon, L. (2012). Econometric measures of Lee, C. C., & Chiu, Y. B. (2012). The impact of real income on insurance penetration:
connectedness and systemic risk in the finance and insurance sectors. Journal of Evidence from panel data. International Review of Economics and Finance, 21(1),
Financial Economics, 104(3), 535–559. 246–260.
Boon, T. K. (2005). Do commercial banks, stock market and insurance market pro- Lee, C. C., Lee, C. C., & Chiu, Y. B. (2013). The link between life insurance activities
mote economic growth? An analysis of the Singapore economy. Working paper of and economic growth: Some new evidence. Journal of International Money and
the School of Humanities and Social Studies. Singapore: Nanyang Technological Finance, 32(3), 405–427.
University. Lee, C. C., Tsong, C. C., Yang, S. J., & Chang, C. H. (2013). Investigating the stationarity
Chang, T., Cheng, S. C., Pan, G., & Wu, T. P. (2013). Does globalization affect the of insurance premiums: International evidence. The European Journal of Finance,
insurance markets? Bootstrap panel granger causality test. Economic Modelling, 19(4), 276–297.
33(1), 254–260. Lee, S. J., Kwon, S. I., & Chung, S. Y. (2010). Determinants of household demand for
Chang, T., Lee, C. C., & Chang, C. H. (2013). Does insurance activity promote insurance: The case of Korea. Geneva Papers on Risk and Insurance: Issues and
economic growth? Further evidence based on bootstrap panel granger causal- Practice, 35(1), 82–91.
ity test. The European Journal of Finance, 1–24. http://www.tandfonline.com/ Levin, A., Lin, C. F., & Chu, C. S. J. (2002). Unit root tests in panel data: Asymptotic
doi/pdf/10.1080/1351847X.2012.757555 and finite-sample properties. Journal of Econometrics, 108(1), 1–24.
Chen, P. F., Lee, C. C., & Lee, C. F. (2012). How does the development of life insur- Liu, G. C., & Lee, C. C. (2014). Insurance activities and banking credit causal nexus:
ance market affect economic growth? Some international evidence. Journal of Evidence from China. Applied Economics Letters, 21(9), 626–630.
International Development, 24(7), 865–893. Maddala, G. S., & Wu, S. (1999). A comparative study of unit root tests with panel data
Chen, S. S., Cheng, S. C., Pan, G., & Wu, T. P. (2013). The relationship between glob- and new sample test. Oxford Bulletin of Economics and Statistics, 61(S1), 631–652.
alization and insurance activities. A panel data analysis. Japan and the World Maeso-Fernandez, F., Osbat, C., & Schnatz, B. (2006). Towards the estimation of equi-
Economy, 28(3), 151–157. librium exchange rates for CEE acceding countries: Methodological issues and
Ching, K., Kogid, M., & Furuoka, F. (2010). Causal relation between life insurance a panel cointegration perspective. Journal of Computational Economics, 34(3),
funds and economic growth evidence from Malaysia. ASEAN Economic Bulletin, 499–517.
27(2), 185–199. Mark, C. N., & Sul, D. (2003). Cointegration vector estimation by panel DOLS and long-
Choi, I. (2001). Unit root tests for panel data. Journal of International Money and run money demand. Oxford Bulletin of Economics and Statistics, 65(5), 655–680.
Finance, 20(2), 249–272. Nektarios, M. (2010). A growth theory for the insurance industry. Risk Management
Craigwell, R., Downes, D., & Howard, M. (2001). The finance-growth nexus: A multi- and Insurance Review, 13(1), 45–60.
variate VAR analysis of a small open economy. Savings and Development, 25(2), Outreville, F. J. (1990). The economic significance of insurance markets in developing
209–223. countries. The Journal of Risk and Insurance, 57(3), 487–498.
Curak, M., Loncar, S., & Poposki, K. (2009). Insurance sector development and eco- Outreville, F. J. (1996). Life insurance markets in developing countries. The Journal
nomic growth in transition countries. International Research Journal of Finance of Risk and Insurance, 63(2), 263–278.
and Economics, 34(3), 29–41. Pagano, M. (1993). Financial markets and growth: An overview. European Economic
Enders, W. (2004). Applied econometric time series. Wiley Publications: New York. Review, 37(2–3), 613–622.
Engle, R. F., & Granger, C. W. J. (1987). Cointegration and error correction: Represen- Pedroni, P. (2000). Fully modified OLS for heterogeneous cointegrated panels.
tation, estimation and testing. Econometrica, 55(2), 251–276. Advances in Econometrics, 15(1), 93–130.
Enz, R. (2000). The S-curve relation between per-capita income and insurance pene- Pedroni, P. (2001). Purchasing power parity tests in cointegrated panels. Review of
tration. Geneva Papers on Risk and Insurance: Issues and Practice, 25(3), 396–406. Economics and Statistics, 83(4), 727–731.
Haiss, P. R., & Sumegi, K. (2008). The relationship between insurance and economic Pradhan, R. P., Arvin, M. B., Norman, N. R., & Hall, J. H. (2014). Causal nexus between
growth in Europe: A theoretical and empirical analysis. Empirica, 35(4), 405–431. economic growth, banking sector development, stock market development and
Han, L., Li, D., Moshirian, F., & Tian, L. (2010). Insurance development and economic other macroeconomic variables: The case of ASEAN countries. Review of Financial
growth. The Geneva Papers on Risk and Insurance, 35(2), 183–199. Economics, 23(4), 155–173.
Harris, R., & Sollis, R. (2006). Applied econometric time series modelling and forecasting. Pradhan, R. P., Arvin, M. B., Norman, N. R., & Nishigaki, Y. (2014). Does banking sector
Willey Publications: England. development affect economic growth and inflation? A panel cointegration and
Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1988). Estimating vector auto regressions causality approach. Applied Financial Economics, 24(7), 465–480.
with panel data. Econometrica, 56(6), 1371–1395. Sigma (various years) World Insurance, Zurich: Swiss Re.
Horng, M. S., Chang, Y. W., & Wu, T. Y. (2012). Does insurance demand or finan- Sümegi, K., & Haiss, P. (2008). The relationship between insurance and economic
cial development promote economic growth? Evidence from Taiwan. Applied growth: Review and agenda. The ICFAI Journal of Risk and Insurance, 5(2), 32–56.
Economics Letters, 19(2), 105–111. UNCTD. (1988). United Nations Conference on Trade and Development (UCTAD). The
Hussels, S., Ward, D., & Zurbruegg, R. (2005). Stimulating the demand for insurance. report on services in the world economy. Geneva: United Nations.
Risk Management and Insurance Review, 8(2), 257–278. Vadlamannati, K. C. (2008). Do insurance sector growth and reforms affect economic
Kao, C., & Chiang, M. H. (2000). On the estimation and inference of a cointegrated development? Empirical evidence from India. The Journal of Applied Economic
regression in panel data. Advances in Econometrics: Nonstationary Panels, Panel Research, 2(1), 43–86.
Cointegration and Dynamic Panels, 15(1), 179–222. Ward, D., & Zubruegg, R. (2000). Does insurance promote economic growth? Evi-
Koop, G., Pesaran, M. H., & Potter, S. M. (1996). Impulse response analysis in non- dence from OECD countries. Journal of Risk and Insurance, 67(4), 489–506.
linear multivariate models. Journal of Econometrics, 74(1), 119–147. Wasow, B., & Hill, R. D. (1986). Determinants of insurance penetration: A cross-
Kugler, M., & Ofoghi, R. (2005). Does Insurance Promote Economic Growth? Evidence country analysis. In B. Wasow, & R. D. Hill (Eds.), Insurance industry in economic
from the UK. In Paper presented at the Money Macro and Finance (MMF) Research development (pp. 160–176). New York University Press.
Group Conference United Kingdom. Webb, I. P., Grace, M. F., & Skipper, H. D. (2005). The effect of banking and insurance
Lee, C. C. (2011). Does insurance matter for growth: Empirical evidence from OECD on the growth of capital and output. SBS Revista De Termas Financieros, 2(2),
countries. The B.E. Journal of Macroeconomics, 11(1), 1–26. 1–32.
Lee, C. C. (2013). Insurance and real output: The key role of banking activities. Wolde-Rufael, Y. (2009). Re-examining the financial development and economic
Macroeconomics Dynamics, 17(2), 235–260. growth nexus in Kenya. Economic Modelling, 26(6), 1140–1146.