Sains Malaysiana 43(9)(2014): 1439–1450

Two-Stage DEA Method in Identifying the Exogenous Factors of Insurers’

Risk and Investment Management Efficiency
(Kaedah DEA Dua-Peringkat dalam Mengenal Pasti Faktor Luaran Terhadap Kecekapan
Pengurusan Risiko dan Pelaburan Penanggung Insurans)

RUBAYAH YAKOB*, ZULKORNAIN YUSOP, ALIAS RADAM & NORISZURA ISMAIL

ABSTRACT
The objective of this study was to identify the exogenous variables of risk and investment management efficiency by
using a two-stage data envelopment analysis (DEA) method. The first stage involves obtaining the efficiency scores of
risk and investment management via DEA that requires only the traditional inputs and outputs. In the second stage, the
Tobit regression analysis is conducted in which the efficiency score obtained from the first stage is treated as a dependent
variable, while the exogenous factors are considered to be independent variables. The exogenous factors consist of
operating systems, organizational form, consumer preference and size. The results showed that the mutual company as
well as the takaful system demonstrate better risk management performance than their stock and conventional system
counterparts. In addition, size is also a significant indicator for risk management efficiency in which the larger insurer/
takaful operator exhibits better risk management performance than the smaller one. However, consumer preference
is found to be insignificantly correlated with the efficiency of risk management. In contrast, with risk management,
organizational form, operating system and size are not indicators of the investment management efficiency, but consumer
preference is significantly and positively associated with investment management efficiency.

Keywords: Efficiency; exogenous factors; risk and investment management; two-stage DEA

ABSTRAK
Objektif kajian ini adalah untuk mengenal pasti pemboleh ubah luaran terhadap kecekapan pengurusan risiko dan
pelaburan dengan menggunakan kaedah analisis penyampulan data (DEA) dua-peringkat. Peringkat pertama melibatkan
perolehan skor kecekapan pengurusan risiko dan pelaburan melalui DEA yang hanya memerlukan input dan output
tradisi. Pada peringkat kedua, analisis regresi Tobit dijalankan dengan skor kecekapan yang diperoleh dari peringkat
pertama dilayan sebagai pemboleh ubah bersandar, manakala faktor luaran dipertimbangkan sebagai pemboleh ubah
bebas. Faktor luaran terdiri daripada sistem operasi, bentuk organisasi, keutamaan pengguna dan saiz. Keputusan
menunjukkan bahawa syarikat bersama, begitu juga sistem takaful mempunyai prestasi pengurusan risiko yang lebih baik
berbanding rakan-rakan syarikat stok dan sistem konvensional mereka. Di samping itu, saiz juga merupakan penunjuk
yang signifikan bagi kecekapan pengurusan risiko syarikat insurans/pengendali takaful yang lebih besar mempamerkan
prestasi pengurusan risiko yang lebih baik daripada syarikat yang lebih kecil. Walau bagaimanapun, keutamaan pengguna
didapati tidak berkorelasi secara signifikan dengan kecekapan pengurusan risiko. Berbeza dengan pengurusan risiko,
bentuk organisasi, sistem operasi dan saiz bukan merupakan penunjuk kepada kecekapan pengurusan pelaburan, tetapi
keutamaan pengguna mempunyai hubungan yang signifikan dan positif dengan kecekapan pengurusan pelaburan.

Kata kunci: APD dua-peringkat; faktor luaran; kecekapan; pengurusan risiko dan pelaburan

INTRODUCTION investment management. First, is the increasing number

Underwriting, pricing and claims handling are the technical of cases of insolvency among insurers. Insolvency can
elements of the insurance production process, which is happen even to large insurers that have been involved in
referred to as manufacturing by Black and Skipper (2000). the business for a long time. On average, the insolvency
During the process, there is a need for the insurance company problem is caused by the inefficiency of the risk
to make decisions relevant to risk, capital and investment. management function. Second, is the uncertainty of
The mismanagement of these elements can affect the whole financial markets and fluctuation of interest rates. Both of
system both within and outside the insurer/takaful operator, these factors affect the investment portfolio of insurers,
thus making the risk and investment management function which is highly important in considering the appropriate
very important to insurance/takaful business. matching between their assets and liabilities. Third,
This study has identified four conditions to motivate globalization has intensified competition. Unfortunately
insurers/takaful operators in enhancing their risk and for insurers, competition is keen among themselves as well
1440

as from other financial institutions, such as banks, mutual Humphrey (1997) and Cummins (1999) suggested that
fund organizations, finance companies and securities firms. frontier efficiency methodologies as a better alternative.
Last but not least, consumer preferences have changed to They clarified that the frontier efficiency methodologies
a more complex product with a small margin but higher seemed very important and this new benchmarking
risk. Recently, protection-based products seem to have techniques measured the firm performance relative to
been overtaken by investment-based products. Although best practice frontiers derived from firms in the industry
the risk is transferred to policyholders in investment-based or branches within financial firms. The advantage of
products, the attractive investment element makes it more such measures, as compared to financial ratio analysis,
interesting. Clearly, based on these four reasons, insurers/ is their ability to summarize firm performance in a single
takaful operators must respond with far greater efficiency statistic that controls for differences among firms using
in their risk and investment management. However, the a sophisticated multidimensional framework (Cummins
efficiency of risk and investment management depends 1999). Moreover, Cummins and Weiss (2000) commented
on various exogenous factors. These factors could be that all economic hypotheses related to insurers about
macroeconomics or/and firm-specific variables. such matters as economies of scope and scale, distribution
Thus, the main objective of this study was to identify systems, organizational forms and the effect of M&A will
the exogenous variables that affect the risk and investment not be convincing unless they applied the frontier-based
management efficiency of life insurers as well as takaful performance measures.
operators, since Malaysia has two different insurance A frontier efficiency methodology that has become
markets, namely, conventional life insurance and takaful. A increasingly important is the DEA which is first introduced
two-stage data development analysis (DEA) method is most by Charnes et al. (1978). The centre attention of DEA
suitable to perform this analysis. The first stage involves is largely on the technological aspects of production
obtaining the efficiency scores of risk and investment correspondences, thus it can be applied to calculate
management via the slack-based measure (SBM) - DEA that technical and scale efficiency without requiring estimates
requires only the traditional inputs and outputs. In the second of input and output prices. On the other hand, if the data
stage, the Tobit regression analysis is conducted in which on input prices are available, cost efficiency also can be
the efficiency score obtained from the first stage is treated measured by using DEA (Aly et al. 1990; Ferrier & Lovell
as a dependent variable, while the exogenous factors are 1990). Cummins and Weiss (2000) write, ‘Intuitively, the
treated as independent variables. The exogenous variables method involves searching for a convex combination of
that are considered in this study are limited to non-financial firms in the industry that dominate a given firm’. They
firm-specific variables that are not the traditional inputs and further explained that these firms form the given firm’s
assumed to not be under the control of managers (Coelli reference set and if the reference set comprises only of
et al. 2005). These variables include organizational form, the firm itself, it is said self-efficient and has efficiency
operating system, consumer preference and size. score equal to 1. Conversely, if other firms instituted the
This study contributes to the literature of efficiency dominant set, then the firm’s efficiency is less than 1 and
in terms of two elements. First, this study will investigate thus considered as inefficient.
exogenous factors that affect the efficiency of the primary Subsequently, they were extended to find the cause of
functions of an insurance company, known as risk and the difference in efficiency between decision making units
investment management functions. This is in contrast to (DMUs) by associating the inefficiency measurement with
many previous insurance efficiency studies, which mostly the exogenous factors. These exogenous environmental
focused on the causes that affect the insurance firms as a factors include the operating system, size, changes
whole. Finally, it is very constructive to engage takaful in consumer preference, labour relations, ownership
operators in this study because of the privileges of the differences, location characteristics, the legal system and
insurance industry in this country that have two different government regulations and organizational form (Fried et
operating systems, namely, conventional insurance al. 1999)
and takaful. Furthermore, very few studies have been Accordingly, many previous researchers had
undertaken on the efficiency of the risk and investment investigated the empirical relationship between insurance
management function among takaful operators. firm efficiency and organizational form and their findings
The paper unfolds as follows. The following section were mixed. Brockett et al. (2005, 2004), Cummins et al.
discusses the literature on previous studies and the (2009) and Hussels and Ward (2007) supported the expense
subsequent section describes the methodology and data. preference hypothesis by showing that the stock insurers
The next section discusses the experimental results and were more efficient than mutual insurers. In contrast, the
the final section concludes the study.
findings by Attiea et al. (2009), Carr (1997), Eckles (2003)
and Eling and Luhnen (2010) were not consistent with the
LITERATURE REVIEW expense preference behaviour hypothesis. Meanwhile,
In recent years, a considerable amount of literature has Cummins and Zi (1997), Fukuyama (1997), Gardner and
been published concerning the efficiency of insurance Grace (1993) and Greene and Segal (2004) found mutual
firms. In measuring the firm’s efficiency, Berger and and stock insurers to be equally efficient.
 1441

Abiding by the concept of scale and scope economies, easy to calculate and simple and enable one to perform
Yao et al. (2007) was convinced that larger insurers were the statistical test in determining the significant exogenous
more efficient than smaller insurers. The same results variables affecting efficiency.
were obtained by Cummins and Zi (1997), Diacon et al.
(2002), Eckles (2003), Gardner and Grace (1993), Hao and
DATA AND METHODS
Chou (2005), Klumpes (2007) and Meador et al. (1997).
Similarly, Hao (2008) proved that, on average, the large
firms experienced higher cost efficiency than the smaller DATA
firms. In addition, Carr et al. (1999) concluded that large For the purpose of this study, the selection of the firms is
insurers are more efficient because they have the advantage restricted to direct insurers (composite and life) and takaful
of distribution channels and market power. However, operators operating in Malaysia. Moreover, data for this
Cummins et al. (2009) formed a different conclusion study are limited to the life and family takaful business
in which the larger insurers indicate lower efficiency as well as investment-linked business. For the composite
changes compared with smaller insurers, while Yuengert insurers that offer general and life products, the data is
(1993) found that size and efficiency were significantly segregated between the two lines of business and can be
uncorrelated. obtained from the financial report of the companies. The
Globalization, technological change and shifting study also totally excluded the new entrants during the
consumer preferences have led to firms adopting a number study periods but maintained the firms involved in merger
of innovative business strategies (Gera 2003). Empirical and acquisition activities. Finally, this left a sample of 20
evidence provided by Meador et al. (1997) suggested firms, consisting of 7 life insurers, 9 composite insurers
that firms that diversify across multiple insurance as and 4 takaful operators that were consistently present
well as investment insurance products, can enhance their throughout the period 2003 to 2007. This sample represents
X-efficiency more than a focused production strategy firm. about 91% of the total players for the study period and
This result is consistent with the prediction of Khaled et accounts for approximately more than two-thirds of the
al. (2001) who studied the scope and scale economies total assets of the life insurance fund as well as the family
of the New Zealand insurance industry. In contrast, Hao takaful fund in the overall life insurance and takaful
(2007) indicated that product mix could not help the life industry, respectively. Data on the financial statement of
insurers in Taiwan to increase their level of efficiency. the firms is adopted from the Companies Commission of
Similarly, takaful operators with limited product lines were Malaysia. The firms under observation according to the
experiencing higher cost efficiency than takaful operators type of business are depicted in Table 1.
with a variety of product lines (Abdul Kader et al. 2010). The 5 year time span of 2003-2007 is considered as
It seems that only a few of the previous studies correlate this period is after the financial crisis of 2001 and 2002 and
the legal system with the efficiency of the insurance before the global credit crunch in 2008. As Zurich Financial
industry. This situation may be due to the fact that many Services (2007) reported that when the stock market
countries in the world apply the same legal system for all dropped substantially between 2000 and 2002 and the
insurance companies operating in the country. However, level of corporate bond weakens, the insurance company
in some countries in Asia, Europe and Africa, there are has suffered severe losses in their investment portfolios.
two insurance systems operating in the market, namely, The same thing happened during the credit crisis in 2008,
conventional insurance system and Islamic insurance
where insurers posted USD239 billion in write downs and
system (takaful). In this respect, Eling and Luhnen (2010)
credit losses worldwide from the global credit crunch in
examined the effects of civil, mixed and common law on
2008. These extreme cases have to be excluded in order
the efficiency of the insurance industry internationally.
to avoid biased results. In addition, over the years, various
Their study proved that the efficiency of the insurer was not
insurance companies had been coming and going out of
affected by the type of legal system practiced in the country.
the Malaysian insurance industry. There are also quite a
It is possible to adjust these exogenous variables
number of merger and acquisition (M&A) activities within
accordingly to compare their relationship with the
the industry. This has posted a challenge to get a most
efficiency of DMUs. There are four common techniques
consistent set of data representing the highest percent of
in which these variables can be accommodated in DEA
the players in the industry. The larger the number of years
(Coelli et al. 2005). The first technique was introduced
would imply more challenge to achieve that. Basically, the
by Banker and Morey (1986). The second approach is
study excluded the new entrants during the study periods
the so-called frontier separation approach, established by
but maintained the firms involved in M&A activities.
Charnes et al. (1981). The third method is known as the
all-in-one approach (Fried et al. 1999), while the fourth
is commonly known as the two-stage approach. The two- TWO-STAGE DATA ENVELOPMENT
stage approach has, so far, been the most recommended ANALYSIS (DEA) METHOD
as it likely addresses the problems encountered in the This study will implement the two-stage method in order
above methods (Coelli et al. 2005). The other advantages to identify the exogenous factors that affect the risk and
suggested by Coelli et al. (2005) are that this method is investment management efficiency of insurers/takaful
1442

TABLE 1. The list of insurer/takaful operator under observation 2003-2007

No. Name of Firm Type of Business

1 Allianz Life Insurance Malaysia Berhad (A) Life
2 Uni. Asia Life Assurance Berhad (B) Life
3 Manulife Insurance (Malaysia) Berhad(C) Life
4 Asia Life (M) Berhad (D) Life
5 Mayban Life Assurance Bhd (E) Life
6 Great Eastern Life Assurance (Malaysia) Berhad (F) Life
7 Commerce Life Assurance Berhad (G) Life
8 Tahan Insurance Malaysia Berhad (H) Composite
9 Hong Leong Assurance Berhad (I) Composite
10 AmAssuranceBerhad (J) Composite
11 MCIS Zurich Insurance Berhad (K) Composite
12 Malaysian National Insurance Berhad (L) Composite
13 Malaysian Assurance Alliance Berhad (M) Composite
14 Takaful NasionalSdn. Bhd. (N) Composite
15 Takaful Ikhlas Malaysia Sdn. Bhd. (O) Composite
16 Syarikat Takaful Malaysia Berhad (P) Composite
17 MaybanTakafulBerhad (Q) Composite
18 Prudential Assurance Malaysia Berhad (R) Composite
19 ING Insurance Berhad (S) Composite
20 American International Assurance Company, Ltd (T) Composite

operators. According to Coelli et al. (2005), the first stage objective. The slacks give the estimate of input excess and
involves obtaining the efficiency scores via DEA, which output shortfalls that could be improved without worsening
only requires the traditional inputs and outputs. In the any other input and output. According to Tone (2001), for
second stage, the regression analysis is conducted in which each DMUj ( and input matric used by DMUj and amount
the efficiency score obtained from the first stage is treated of output matric yielded by DMUj, with the assumption,
as a dependent variable, while the exogenous factors the data set is positive and , the production possibility set
are independent variables. The second-stage regression for SBM is defined by:
analysis is used to determine separately the effect of
exogenous variables on efficiency. They also explained that P = {(x, y) ǀx ≥ Xλ,y ≤ Yλ, λ ≥ 0} (1)
the exogenous variables include all the factors that cannot
be treated as traditional inputs and are not assumed to be where λ is a nonnegative vector in Rn. In an attempt to
directly under the control of managers. estimate the efficiency of a DMU (xo, yo), the following
In the second stage, the Tobit regression analysis is fractional program (FP) is formulated:
used in order to obtain the exogenous factors that influence
the risk and investment management efficiency of the
insurers/takaful operators (Banker & Natarajan 2008; (2)
Coelli et al. 2005; Pasiouras 2008). The first stage had been
undertaken separately, which is slack-based measure – data
envelopment analysis (SBM-DEA). This study focuses only subject to
on the second-stage regression analysis. However, the first-
stage SBM-DEA will also be explained in order to facilitate xo = Xλj + s–
understanding in the second-stage.
yo = Yλj – s+
SBM-DEA
The SBM model is a variant of the additive DEA model, 0 ≤ λ, s–, s+
which was first presented by Tone (2001). As in the
additive model, the SBM differs from the Charnes, Cooper Then, (2) is replaced by the following linear program
and Rhodes (CCR) and Banker, Charnes and Cooper (BCC) (LP) in t, S–, S+ and Λ:
model as it combines both orientations in a single model,
(3)
i.e. input-oriented model and output-oriented model. SBM
focuses on maximizing the non-zero slacks in the optimal
 1443

subject to Carr (1997), Chilingerian (1995), Oum and Yu (1994) and

Bjurek et al. (1992) suggested applying the Tobit analysis
in the second-stage of the DEA approach. Given that the
DEA efficiency scores resemble corner solution variables
(Hoff 2007), this study will also employ a two-limit Tobit
txo = XΛ + S– regression to estimate the effect of the operating system,
organizational form, size and consumer preference on
tyo = YΛ – S+ the risk and investment management efficiency. The
relationship may be described by the model:
Λ, S–, S+ ≥ 0; t > 0
(4)
The constraint t > 0 make the transformation is
reversible, thus, the FP is equivalent to LP (Cooper et al. where μi ~ N (0, σ2). Yi* is a latent variable following
2007). If the optimal solution of SBMLP would be (τ*, t*, censored normal distribution with mean Xiβ and variance
Λ*, S–*, S+*), then, the optimal solution of SBMLP will be σ2. Xi is a k × 1 vector of observations on the constant and k
defined by (p* = τ*, λ* = Λ*/t*, s–* = S–*/t*, s+* = S+*/t*). – 1 efficiency factor explanatory variables; β a k × 1 vector
Therefore, based on this definition, a DMU (xo, yo) can be of unknown coefficients. The data generating process (DGP)
decided as SBM-efficient if and only if ρ* = 1. This condition – (equation 1) postulates that Yi is the observed SBM-DEA
is achieved when s–* = 0 and s+* = 0, i.e. the value of all efficiency score and the censored values of with censoring
slack variables is equal to zero. below 0 and above 1 (McDonald 2009). is defined by the
Both DMU s in this study - risk and investment following measurement equation:
management functions for each insurer and takaful
operator - have different inputs and outputs. For risk
management, the inputs are investment risk, underwriting (5)
risk and leverage, while the output is the amount of benefits
paid plus reserves. On the other hand, for investment
management, the inputs consist of net actuarial reserves Ordinary least squares (OLS) appear to be less accurate
and total investment assets and its outputs are the solvency in estimating censored regression models (Gujarati 2003;
score and investment return. The efficiency of risk and Wooldridge 2002). Gujarati (2003) has shown empirically
investment management for each insurer/takaful operator that the OLS estimator for the censored regression model is
is now can be calculated using the SBM-DEA and it is done biased as well as inconsistent – meaning that the estimated
separately. parameter is not going to converge with its real value, no
matter how large a sample size is observed. He further
Tobit Regression Analysis In this study, the dependent explains that this is because the conditional mean of the
variable is the efficiency score, which is obtained from error term, μi in the censored regression model is nonzero
the first stage analysis (SBM-DEA), while the independent and it is also correlated with the dependent variable; it is
variables or exogenous factors that have been identified known that both these conditions violate any assumptions
that may have an influence on the efficiency comprise under OLS. In accordance with the explanation by Carr
the operating system, organizational form, consumer (1997) that the normality assumption underlying OLS
preference and size. It is noted that the efficiency score regression cannot be defended because the DEA efficiency
lies within the range of 0 to 1. Thus, it is very important score as the dependent variable lies within the range of
to ensure that the analysis used must be in accord with the 0 to 1. Therefore, the Tobit model is usually estimated
habits of the dependent variable that only takes the values​​ using the maximum likelihood (ML). For a data set with
in the range of 0 to 1. N observations, the ML function is:
The regression analysis that can take into account
the dependent variables with such a limited value is the
(6)
censored regression model, also known as the Tobit model
(Gujarati 2011; Greene 2003; Wooldridge 2002). The Tobit
analysis, which was proposed by Tobin (1958), assumes In general, the Tobit analysis is preferred over the other
that the dependent variable is clustered or censored at a regression techniques because it will take into account
limiting value, which is usually 0. Hoff (2007) summarized all observations to estimate the regression line, including
what was stated by Wooldridge (2002) in that Tobit analysis those at the limit and those above it, while, for the other
is appropriate when the dependent variable is bounded by techniques, the estimation of the regression line is based on
the lower or upper limit or both, ‘with positive probability observations above the limit (McDonald & Moffit 1980).
pileup at the interval ends, either by being censored or by As stated earlier, four exogenous variables – operating
being corner solutions’. In respect of the DEA efficiency system, organization form, consumer preferences to non-
score as a dependent variable, Pasiouras (2008), Hoff traditional product (specifically, towards investment-linked
(2007), Coelli et al. (2005), Ruggiero and Vitaliano (1999), products) and size – are considered in this study. Table 2
1444

summarizes the independent and dependent variables, as As shown in Table 3, the efficiency of risk and
well as the measurements that are used in the Tobit analysis. investment management is relatively moderate. The
All independent variables described above are average efficiency of risk and investment management is
regressed with the dependent variable, which is the SBM- 0.675 (67.5%) and 0.609 (60.9%), respectively, signifying
DEA risk and investment management efficiency score of that the average insurer/takaful operator could further
the insurers/takaful operators using Tobit analysis. The improve by 32.5 and 39.1%, respectively, in order to
Tobit analysis will be carried out separately, one is for be on the efficient frontier. This would imply that there
the efficiency of risk management and the other is for are considerable opportunities for the insurance and
investment management efficiency. takaful industry to improve the performance of risk and
investment management. The efficiency dispersion, which
is represented by the value of standard deviation for
FINDINGS
risk and investment management is 0.3092 and 0.2546,
respectively. The increase in the average efficiency and
SUMMARY STATISTICS small dispersion in efficiency is a good sign because
Generally, the efficiency of risk and investment insurers/takaful operators are competing with each other
management is achieved by different insurers/takaful to improve their performance in both risk and investment
operators for each year from 2003-2007 (Appendix 1). management (Cummins 1999).
According to Appendix 1, for risk management, there are Furthermore, the average size of the insurer/takaful
5, 8, 5, 8 and 9 efficient insurers/takaful operators in the operator for the industry is 4.41 billion with a standard
year 2003-2007, respectively. In contrast, 4 insurers (H, deviation of 6.07 billion. The average total investment-
J, M and O) have been identified as inefficient throughout link asset in billion, which represents the preference of
the year 2003-2007. Likewise, the distribution of insurers/ consumers towards the investment-linked product is 0.36
takaful operators that are efficient in terms of investment billion with a standard deviation of 0.67 billion. This is
management efficiency is also not the same throughout most likely because the offering of investment-linked
the years (Appendix 2). There are 3, 3, 6, 3 and 5 insurers/ products is still at its early stage. In addition, there are still
takaful operators having efficient investment management many people in this country who do not have an insurance
in 2003-2007, respectively, while 10 insurers/takaful policy, not even a basic policy that only provides protection
operators experience inefficient investment management (whole life/endowment/term policy). Perhaps, the insurers
throughout the observed years. These results also confirm may not yet feel confident to offer a complex product, such
that on average, insurers that are inefficient in terms of as an investment-linked policy, which, of course, is only
investment management are much more than insurers those required by a small section of the society.
are inefficient in terms of risk management.

TABLE 2. Dependent and independent variables used in Tobit analysis

Dependent Variable Measurement

Risk management efficiency SBM-DEA risk management efficiency score
Investment management efficiency SBM-DEA investment management efficiency score
(Both scores lies in a range 0 -1)
Independent Variable Measurement
Operating system 0 – takaful operator; 1 – conventional insurer
Organizational form 0 – mutual; 1 - stock
Consumer preference Total investment-linked asset/Total life asset
Size Natural logarithm of total asset

TABLE 3. Summary statisticsa

Variable Min Max Mean SDb

Efficiency (RM)c 0.031 1 0.675 0.309
Efficiency (IM)d 0.158 1 0.609 0.255
Size*
0.007 32.87 4.41 6.04
Consumer preference
*
0 3.49 0.36 0.67
a
Organizational form is not listed in the table because it is a dummy variable taking the value 0 or 1; bSD–standard deviation;
c
RM–risk management; dIM–investment management; *size and consumer preference in billion
 1445

RESULTS OF TOBIT REGRESSION ANALYSIS Table 4 also exhibits that size is found to be significant
Although the operating system (conventional vs. takaful) at the 1% critical level and has a positive effect on risk
was originally to be included as an exogenous factor, management efficiency. Thus, this means that from the
unfortunately it had to be dropped from the analysis perspective of risk management, large insurers/takaful
because of the high correlation between the operating operators tend to be more efficient. The findings of this
system and organizational form (stock vs. mutual). This study also can be justified by Cummins et al. (2009) who
is because all the conventional insurers are also stock claimed that smaller insurers experience higher shadow
companies except for the insurer MCIS-Zurich and all prices for risk management and thus in the long-run, may
takaful operators, which are also mutual companies. have difficulty competing with larger insurers. They added
In Table 4, the organizational forms are significant that this was due to resource constraints and economies
at the 1% critical level. The results imply that the mutual of scale in risk management activities and systems.
insurers are more efficient than stock insurers in terms Furthermore, large insurers can take advantage of the
of risk management. The results obtained in this study economies of scale and scope (Yao et al. 2007). Compared
are consistent with the managerial discretion hypothesis, to the organizational form and size, it thus seems that
incentive conflict (Mayers & Smith 1981) and the theory consumer preference is not an important determinant of
of adverse selection (Smith & Stutzer 1990). Furthermore, risk management efficiency (Table 4). This implies that the
since the conflict between the owners and policyholders offering of products based on consumer preference does
is lower, the mutual company is said to have more stable not affect the efficiency of insurer/takaful operator risk
prices and provide better services (Cummins & Weiss management. Hence, it can be said that the investment-link
2004). From an alternative point of view, differences products do not significantly change the risk activities of
in exposure to the risk profile experienced by mutual insurers/takaful operators. It may be better for an industry
insurers are most likely attributable to the differences to maintain its traditional products, such as whole life,
in the operating system between takaful operators endowment and term, rather than being involved with
and conventional insurers. This relation is justified multiple products including investment-linked, which are
inasmuch as all the takaful operators are mutual insurers. acknowledged as being more complicated (Carr et al. 1999;
Obviously, takaful operators have unique characteristics Cummins 1999; Eling & Luhnen 2010).
of underwriting and pricing practices when compared with In contrast to risk management efficiency, it is
the conventional insurance system, as described by Ali found that the organizational form is not a significant
(1989) and Kwon (2007). In addition, there are several predictor to investment management efficiency (Table 5).
clauses in conventional insurance as suicidal clauses and This result is in line with Spiller (1972) and Hansmann
policy loans are not applicable and modified in the takaful (1985), who claimed that different organizational forms do
system. This makes takaful policies less complicated not influence investment performance, which is measured
than conventional insurance. In addition, the Takaful Act by the rates of return on invested assets. Hansmann (1985)
1984, which provides guidance on the practice of takaful commented that, in principle, the owner of the insurer,
operations, particularly in terms of investment practices, regardless of organizational form, must ensure that the
reduces problems in the takaful operation. investment of equity capital and policy premium, at any

TABLE 4. Tobit regression results; (dependent variable = efficiency score)

Risk management efficiency

Independent variables Coefficients z-Statistic
Constant -2.341 -3.123**
Size 0.159 4.253**
Consumer preference 2.60E-05 0.350
Organizational form -0.375 -2.880**
*
significant at 5%; **significant at 1%

TABLE 5. Tobit regression results; (dependent variable = efficiency score)

Investment management efficiency

Independent variables Coefficients z-Statistic
Constant 0.619 1.259
Size -0.002 -0.075
Consumer preference 0.0001 2.568*
Organizational
form 0.012 0.159
significant at 5%; **significant at 1%
*
1446

time, must be able to pay all claims submitted, even if the that better risk management performance experienced
mortality experience turns out to be higher than expected. by mutual insurers are most likely attributable to the
Similarly, size does not provide a significant effect differences in the operating system between takaful
on the efficiency of investment management of insurers/ operators and conventional insurers. It is also evidenced
takaful operators, which is shown in Table 5. This that larger insurers/takaful operators are likely to exhibit
situation is likely due to the strict regulations governing better performance of risk management than smaller
the insurance and takaful industry, particularly in matters ones. The most prevailing reason is that large insurers
relating to investment activities. Investment regulations can take advantage of the economies of scale and scope,
and capital requirements come with limitations (Lee 1997). low levels of insolvency risk and capability in improving
In its effort to control the investment activities among flexibility to position the best combination of their inputs
insurers/takaful operators, the Government of Malaysia and outputs. However, consumer preference towards
introduced the Authorized Malaysian Assets, which investment-linked products does not affect the efficiency
comprise a range of assets typically held against insurance of the risk management function of an insurer. The finding
funds (Lee 1997). The minimum percentage of total implies that the existence of investment-linked policies in
assets in insurance funds to be maintained in Authorized the Malaysian market seems unlikely to significantly alter
Malaysian Assets is 80%. This requirement has caused the the risk profile of the insurer.
insurance funds of insurer/takaful operators to be largely In comparing with the risk management efficiency,
held in fixed-income investments, such as the Malaysian organizational form and size are not significant indicators
Government Securities and corporate securities. Thus, of investment management performance. The owner of the
regardless of the size of the insurer/takaful operator, the insurer, regardless of organizational form, must ensure that
investment portfolio diversification is restricted to some the investment of equity capital and policy premium, at any
degree to the percentage and choice of portfolio. More time must be able to pay all claims submitted, even if the
importantly, the safety, yield and liquidity of investment mortality experience turns out to be higher than expected.
activities must be considered. The same reason can also be used to justify why the size
However, consumer preference is found to be of the insurer does not affect the efficiency of investment
positively and significantly related to the performance management. Regardless of the size of the insurer/takaful
of investment management at the 5% critical level operator, the investment portfolio diversification is
(Table 5). The result implies that the insurers/takaful restricted to some degree of the percentage and portfolio
operators with more products that meets the consumer choices and must adhere to the guidelines of the Malaysian
demand are demonstrating better investment management Authorized Assets. However, the investment management
performance. According to Adams (1996), the investment efficiency of insurers/takaful operators does affected by
choices for insurers are highly dependent on the nature consumer preference towards investment-linked policy.
of policies in force. Further, he explained that consumer The investment-linked products, with increasing frequency,
preference for a product, such as investment-linked will increase the number of these policies in force, and,
policies, the risks and investment returns, are normally thus, change the rate of new business of insurers/takaful
borne by the policyholders. It is true that this will eliminate operators. This situation will urge insurers/takaful
the need to accurately match the investment earnings with operators to provide a high rate of return in order to meet
outstanding liabilities; however, the growth of investment- any maturity or death benefit promised in the policy as
linked policies will increase the number of policies in force well as the rate of return promised to policyholders.
in the market and this will affect the rate of acquisition From these results, insurer/takaful operators are
of new business. Therefore, the higher the amount of expected to enhance their risk management practices
investment-linked products offered by insurers, the higher and investment strategies to ensure that their role as risk
the investment earnings they are likely to achieve in order bearer, financial service provider and intermediary can be
to meet any maturity or death benefit promised in the policy preserved as well as accomplish the requirements of their
in the event of inadequate reserves, as well as promised stakeholders. In response, the government should provide
return to the policyholders. Accordingly, at a favourable the infrastructure to support the improvement of risk and
stage of investment earnings, the solvency of insurers will investment management efficiency among insurers/takaful
be preserved. operators, mainly with respect to the license for takaful
operators, merger and acquisition and development of
CONCLUSION consumer-based products.

This study confirmed that the mutual insurers are

demonstrating better risk management efficiency than
REFERENCES
their stock counterparts. Mutual insurers are likely
contending with minimum incentive conflicts, low-risk Abdul Kader, H., Adams, M. & Hardwick, P. 2010. The cost
insurance prospects, lower chances of making mistakes efficiency of takaful insurance companies. The Geneva
due to the standard and less complex policies and less Papers on Risk and Insurance-Issues and Practice 35:
161-181.
risky future cash flows. It is also interesting to highlight
 1447

Adams, M. 1996. Investment earnings and the characteristics References and DEA-Solver Software. US: Springer
of life insurance firms: New Zealand evidence. Australian Science+Business Media LLC.
Journal of Management 21: 41-55. Cummins, J.D. 1999. Efficiency in the US life insurance industry:
Ali, Kasi Md. Mortuza, 1989. Principles and practices of Are insurers minimizing costs and maximizing revenues?
insurance under Islamic framework. Insurance Journal: In Changes in the Life Insurance Industry: Efficiency,
29-38. Technology and Risk Management, edited by Cummins, J.D.
Aly, H.Y., Grabowsky, R., Pasurka, C. & Rangan, N. 1990. & Santomero, A.M. Boston: Kluwer Academy Publishers.
Technical, scale and allocative efficiencies in US banking: An pp. 75-115.
empirical investigation. Review of Economics and Statistics Cummins, J.D. & Weiss, M.A. 2004. Consolidation in the
72: 211-218. European Insurance Industry: Do Mergers and Acquisitions
Attiea, M., Rao, A. & Kashani, H. 2009. Cost efficiency and Create Value for Shareholders? Working Paper. The Wharton
value driver analysis of insurers in an emerging economy. Financial Institution Centre, University of Pennsylvania.
Managerial and Decision Economics 30: 265-280. Cummins, J.D. & Weiss, M.A. 2000. Analyzing firm performance
Banker, R.D. & Natarajan, R. 2008. Evaluating contextual in the insurance industry using frontier efficiency methods.
variables affecting productivity using data envelopment In Handbook of Insurance, edited by Dionne, G. Boston:
analysis. Operations Research 56: 48-58. Kluwer Academic Publishers. pp. 767-829.
Banker, R.D. & Morey, R.C. 1986. Efficiency analysis for Cummins, J.D. & Zi, H. 1997. Measuring Cost Efficiency in the
exogenously fixed inputs and outputs. Operations Research US Life Insurance Industry: Econometric and Mathematical
34: 513-521. Programming Approaches. Working Paper, The Wharton
Berger, A.N. & Humphrey, D.B. 1997. Efficiency of financial Financial Institution Centre, University of Pennsylvania.
institutions: International survey and direction for future Cummins, J.D., Dionne, G., Gagne, R. & Nouira, A. 2009.
research. European Journal of Operational Research 9: Efficiency of insurance firms with endogenous risk
175-212. management and financial intermediation activities. Journal
Bjurek, H., Kjulin, U. & Gustafsson, B. 1992. Efficiency, of Productivity Analysis 32: 145-159.
productivity and determinants of inefficiency at public day Diacon, S., Starkey, K., O’Brien, C.O. & Odindo, C. 2002.
care centers in Sweden. Scandinavian Journal of Economics The Most Successful Insurers in the UK Long-term Market.
94: 173-187. Discussion Paper, Centre for Risk & Insurance Studies, The
Black, K. Jr. & Skipper, H. D. Jr. 2000. Life & Health Insurance. University of Nottingham.
New Jersey: Prentice Hall, Inc. Eckles, D.L. 2003. Economic Theory and Financial Issues
Brockett, P.L., Cooper, W.W., Golden, L.L., Rousseau, J.J. & in Insurance Markets. PhD Dissertation. University of
Wang, Y. 2005. Financial intermediary versus production Pensylvania (unpublished).
approach to efficiency of marketing distribution systems and Eling, M. & Luhnen, M. 2010. Efficiency in the international
organizational structure of insurance companies. The Journal insurance industry: A cross-country comparison. Journal
of Risk and Insurance 72: 393-412. Banking and Finance 34: 1497-1509.
Brockett, P.L., Cooper, W.W., Golden, L.L., Rousseau, J.J. Ferrier, G.D. & Lovell, C.A.K. 1990. Measuring cost efficiency
& Wang, Y. 2004. Evaluating solvency versus efficiency in banking: Econometric and linear programming evidence.
performance and different forms of organization and Journal of Econometrics 46: 229-245.
marketing in US property-liability insurance companies. Fried, H.O., Schidmt, S.S. & Yaisawarng, S. 1999. Incorporating
European Journal of Operation Research 154: 492-514. the operating environment into a nonparaetric measure of
Carr, R.M. 1997. Strategic Choices, Firm Efficiency and technical efficiency. Journal of Productivity Analysis 12:
Competitiveness in the US Life Insurance Industry. PhD 249-267.
Dissertation, University of Pennsylvania (unpublished). Fukuyama, H. 1997. Investing productive efficiency and
Carr, R.M., Cummins, J.D. & Regan, L. 1999. Efficiency and productivity changes of Japanese life insurance companies.
competitiveness in the US life insurance industry: Corporate, Pacific-Basin Finance Journal 5: 481-509.
product and distribution strategies. In Changes in The Gardner, L. & Grace, M. 1993. X-efficiency in the U.S. life
Life Insurance Industry: Efficiency, Technology and Risk insurance industry. Journal of Banking and Finance 17:
Management, edited by Cummins, J.D. & Santomero, A.M. 497-510.
Boston: Kluwer Academy Publishers. pp. 117-157. Gera, S. 2003. Innovative Business Strategies and Firm
Charnes, A., Cooper, W.W. & Rhodes, E. 1978. Measuring the Performance in the New Economy. Asia-Pacific Economic
efficiency of decision making units. European Journal of Cooperation, APEC Secretariat: Singapore.
Operational Research 2: 429-444. Greene, W.H. & Segal, D. 2004. Profitability and efficiency in the
Charnes, A., Cooper, W.W. & Rhodes, E. 1981. Evaluating US life insurance industry. Journal of Productivity Analysis
program and managerial efficiency: An application of 21: 229-247.
data envelopment analysis to program follow through. Greene, W. 2003. Econometric Analysis. New Jersey: Prentice
Management Science 27: 668-697. Hall Inc.
Chilingerian, J.A. 1995. Evaluating physician efficiency in Gujarati, D. 2011. Econometrics by Examples. UK: Palgrave
hospitals: A multivariate analysis of best practices. European McMillan.
Journal of Operational Research 80: 548-574. Hansmann, H. 1985. The organization of insurance companies:
Coelli, T.J., Rao, P., O’Donnell, C.J. & Battese, G.E. 2005. An Mutual versus stock. Journal of Law, Economics &
Introduction to Efficiency and Productivity Analysis. US: Organization 1: 125-153.
Springer Science+Business Media, LLC. Hao, J.C. 2008. Measuring cost efficiency in the Taiwan Life
Cooper, W.W., Seiford, L.M. & Tone, K. 2007. Data Envelopment insurance industry. International Journal of Management
Analysis: A Comprehensive Text with Models, Applications, 25: 279-286.
1448

Hao, J.C. 2007. Efficiency test on Taiwan’s life insurance Smith, B. & Stutzer, M. 1990. Adverse selection, aggregate
industry-using X-efficiency approach. Information and uncertainty and the role of mutual insurance contracts.
Management Sciences 18: 37-48. Journal of Business 63: 493-511.
Hao, J.C. & Chou, L.Y. 2005. The estimation of efficiency for Spiller, R. 1972. Ownership and performance stock and mutual
life insurance industry: The case in Taiwan. Journal of Asian life insurance companies. The Journal of Risk and Insurance
Economics 16: 847-860. 39: 17-25.
Hoff, A. 2007. Second stage DEA: Comparison of approaches for Tobin, J. 1958. Estimation of relationships for limited dependent
modelling the DEA score. European Journal of Operational variables. Econometrica 26: 24-36.
Research 181: 425-435. Tone, Kaoru. 2001. Theory and methodology: A slack-based
Hussels, S. & Ward, D.R. 2007. The Impact of Deregulation on measure of efficiency in data envelopment analysis. European
the German and UK Life Insurance Markets: An Analysis Journal of Operational Research 130: 498-509.
of Efficiency and Productivity Between 1991 and 2002. Wooldridge, J.M. 2002. Econometric Analysis with Cross Section
Working Paper, Cranfield Management Research Paper and Panel Data. Cambridge: MIT Press.
Series, Cranfield University. Yao, S., Han, Z. & Feng, G. 2007. On technical efficiency of
Khaled, M., Adams, M.B. & Pickford, M. 2001. Estimates of china’s insurance industry after WTO accession. China
scale and scope economies in the New Zealand life insurance Economic Review 18: 66-86.
industry. The Manchester School 69: 327-349. Yuengert, A.M. 1993. The measurement of efficiency in life
Klumpes, P.J.M. 2007. Consolidations and Efficiency in the insurance: Estimates of a mixed normal-gamma error model.
Major European Insurance Markets. Working Paper, Imperial Journal of Banking and Finance 17: 483-496.
College. Zurich Financial Services. 2007. Investment management - A
Kwon, W.J. 2007. Islamic principle and takaful insurance: Re- creator of value in an insurance company (November). pp.
evaluation. Journal of Insurance Regulation 26: 53-81. 1-15.
Lee, H.L. 1997. The Insurance Industry in Malaysia. Kuala
Lumpur: Oxford University Press.
Mayers, D. & Smith Jr., C.W. 1981. Contractual provisions, Rubayah Yakob*
organizational structure and conflict control in insurance School of Management
markets. Journal of Business 54: 407-434. Faculty of Economics and Management
McDonald, J. 2009. Using least squares and Tobit in second stage Universiti Kebangsaan Malaysia
DEA efficiency analyses. European Journal of Operational 43600 Bangi, Selangor
Research 197: 792-798. Malaysia
McDonald, J.F. & Moffitt, R.A. 1980. The uses of Tobit analysis.
The Review of Economics and Statistics 62: 318-321. Zulkornain Yusop & Alias Radam
Meador, J.W., Ryan, Jr. H.E. & Schellhorn, C.D. 1997. Product Faculty of Economics and Management
Focus versus Diversification: Estimates of X-Efficiency for Universiti Putra Malaysia
the US Life Insurance Industry. Working Paper. The Wharton 43400 Serdang, Selangor
Financial Institution Centre, University of Pennsylvania. Malaysia
Oum, T.H. & Yu, C. 1994. Economic efficiency of railways and
implications of public policy: A comparative study of the Noriszura Ismail
OECD countries’ railways. Journal of Transport Economics Faculty of Science and Technology
and Policy 28: 121-138. Universiti Kebangsaan Malaysia
Pasiouras, F. 2008. International evidence on the impact of 43600 Bangi, Selangor
regulations and supervision on bank’s technical efficiency: An Malaysia
application of two-stage data envelopment analysis. Review
Quantitative Finance and Accounting 30: 187-223. *Corresponding author; email: [email protected]
Ruggiero, J. & Vitaliano, D.F. 1999. Assessing the efficiency of
public schools using data envelopment analysis and frontier Received: 25 September 2013
regression. Contemporary Economic Policy 17: 321-331. Accepted: 10 January 2014


APPENDIX 1

SBM-DEA Results for Risk Management for Individual Insurer/Takaful operator

2003 2004 2005 2006 2007

DMU Efficiency DMU Efficiency DMU Efficiency DMU Efficiency DMU Efficiency
score score score score score
A 0.6496 A 1.0000 A 0.2960 A 0.6072 A 1.0000
B 0.4815 B 1.0000 D 0.4650 B 0.5993 B 1.0000
C 1.0000 C 0.9334 E 1.0000 C 0.4911 C 0.5040
D 0.5051 D 0.3747 F 1.0000 D 1.0000 D 1.0000
E 0.5842 E 0.3942 H 0.3895 E 1.0000 E 0.7466
F 0.4606 F 0.3727 I 0.8498 G 1.0000 F 0.6337
H 0.3765 G 1.0000 J 0.3824 H 0.1340 G 1.0000
I 1.0000 H 0.3175 K 1.0000 I 0.6005 I 1.0000
J 0.8292 I 0.4200 L 0.4299 J 0.3883 J 0.6015
K 1.0000 J 0.5967 M 0.2994 K 1.0000 K 1.0000
L 0.3739 K 1.0000 N 1.0000 L 1.0000 L 0.5634
M 0.2256 L 1.0000 O 0.1674 M 0.3472 M 0.5315
N 0.6349 M 0.2279 P 1.0000 N 1.0000 N 1.0000
O 0.0312 N 1.0000 Q 0.3416 O 0.2600 O 0.2379
P 1.0000 O 0.1617 S 0.8144 P 1.0000 Q 1.0000
Q 0.1926 P 1.0000 Q 0.2545 R 0.7621
R 1.0000 Q 0.3569 R 1.0000 S 1.0000
T 0.3911 T 1.0000 S 0.7008 T 0.4453
1449
 1450

APPENDIX 2

SBM-DEA Results for Investment Management for Individual Insurer/Takaful operator

2003 2004 2005

DMU Efficiency score DMU Efficiency score DMU Efficiency score DMU Efficiency DMU Efficiency score
score
A 0.5984 A 0.5771 A 0.5213 A 0.4997 A 0.5578
B 0.6718 B 0.3412 B 0.5232 B 0.3669 B 0.3670
C 0.3390 C 0.2733 C 0.6268 C 0.4978 C 0.7488
D 0.5018 D 0.4877 D 1.0000 D 0.4748 D 1.0000
E 0.2789 E 0.4005 E 0.7904 E 1.0000 E 1.0000
F 0.6312 F 0.6899 F 1.0000 F 0.9001 F 0.9064
G 0.3801 G 0.6762 G 0.5198 G 0.5632 G 0.6359
H 0.2205 H 0.3235 H 0.5202 H 1.0000 H 0.2804
I 1.0000 I 1.0000 I 0.6645 I 0.4528 I 0.6469
J 0.4614 J 0.3672 J 0.6329 J 0.5144 J 0.5730
K 0.2615 K 0.2704 K 0.5007 K 0.5570 K 1.0000
L 1.0000 L 0.2919 L 1.0000 L 0.4079 L 0.4967
M 0.8935 M 1.0000 M 1.0000 M 1.0000 M 1.0000
N 0.1576 N 0.1870 N 0.4188 N 0.3691 N 0.4668
O 1.0000 O 1.0000 O 1.0000 O 0.3349 O 0.4820
P 0.1724 P 0.3377 P 0.5781 P 0.4420 P 0.6796
Q 0.2532 Q 0.6091 Q 1.0000 Q 0.8806 Q 1.0000
R 0.5116 R 0.5500 R 0.7408 R 0.6892 R 0.7758
S 0.3490 S 0.3939 S 0.6019 S 0.4692 S 0.5509
T 0.2945 T 0.5160 T 0.7429 T 0.8098 T 0.8046