VIP Call Girls Pune Kirti 8617697112 Independent Escort Service Pune
Consumer Marketing of Life Insurance
1. MBA Project Report
"ConsLlmer Marketing of Life lnsurance Products: a
case study of lndian Consumers"
By
Dinabandhu Bag
Proyect Submitted in Partial Fulfillments for the Award of the
Degree of Master of Business Administratlon (MBA)
Under the guidance of
Prof. V. Narayanan
HOD
R.C. College of Commerce, Bangalore, 560 001
December 2002
2. b*4
(Prof. V Narayrf,nan)
Bangafore, 560 OO1
Certificate of Approval
This is to certify that the lcNou MBA project tiiled " consumer Marketing of Life
Insurance Products: a case study of Indian consumers ,,,
the proposar of which
was approved vide pp No. 28!152 r 2002, submitted by sri Dinabandhu Bag is a
bonafide work carried out by him under my guidance. This has neither been
published nor been submitted to any other institution/university for award of any
other degree.
Professor & Head of the Depaftment of commerce & Management,
R C College of Commerce.
I
3. a
CONTENTS
TITLE
CHAPTER 1: INTRODITT|ON
CHAPTER 2: DATA MINING & INSURANCE MARKETING
CHAPTER 3: BUSINESS STRATEGY
CHAPTER 4: SAMPLE SEGMENTS & RESULTS
CHAPTER 5: CONCLUSIONS
APPENDIX
PAGE
3-16
r7-28
29-36
3t-49
50-53
54-59
4. PREFACE
The liberalisatim of the insurance sector has thrown up the domestic lndian
insurance rnarket to ttre private and forelgn players. Marketing of insurance
F,rodlcts to hdat cocsr.rrers is a recent phenonenon and poses unique
ctlafenge to fE irclrers. Th€re is a need to identify and develop demographic
custqrEr segfrEnts thal coufd help minimilng marketing costs and make
efiecrive rnarketirg. Further there exist a need to suggest an appropriate
business strategy across the segments for better positioning of life insurance
products. Modem data mining tools could provide business intelligence solutions
used for marketing. Data base segmentation is the most common application of
data mining solution. This study attempts to highlight the significance of data
based segmentation to achieve business advantages using modern data mining
techniques.
5. )
t
)
)
)
)
)
)
,
a
CHAPTER 1
INTRODUCTION
The recent liberalisation of the insurance sector has thrown up the
domestic Indian insurance market to private and the foreign players. since then'
a number of private companies associated with the foreign entities have entered
into the Indian market as joint players, The opening of the sector to the private
players witnessed the introduction of a number of new products deserving the
attention of the customers and possibly the Indian insurance market is one of the
fastest growing markets. However, marketing of insurance products to Indian
consumers is a recent phenomenon and poses unique challenge to the insurers'
Life insurance is a device of spreading a possible financial loss over a
large number of persons over the lives of the insured' lt seeks to reduce the
financial unceftainties arising out of natural calamities or contingencies of old
age and death. Life insurance is a promise to perform in future. Life insurance
product continues to exist over a long period of time, and for marking its service
available, the insured person has to go on paying the purchase price throughout
the term of the policy. Finally, the seller has not only to sell its product but also to
keep the contract in force by continuous and efficient servicing. In other words
sales and service in case of a life product go together'
6. Do we sell life insurance to everybody? What is the scope
insurance marketing in India? These are the questions that needs
answered appropriately.
1.1 Life insurance in lndia
Life insurance products in India became popular with the formation of the
Life Insurance corporation of India vide the Nationalisation Bill in 1956. Since
then the LIC has grown with no leaps and bounds.
Table 1.1 and Table 1.2 depict the stock and growth of life insurance in
India. As shown in Table 1.1, the total sum assured of all LIC policies has grown
from Rs 5,36,450 crores in 2000 to Rs 6,48,000 crores in 2001. Simlarly, the
number of policies sold has also gone up from 1 ,1 crores in 2000 to 1 .5 crore in
2001. The total number of policies sold at 1,5 crores are miniscule for our
country of a population over 100 crores.
Table 1.1: Life lnsurance in lndia
of life
to be
Year 2001 2000
Sum assured (Rs. crores) Assurances 645041 536450
Annuities 526 489
Pension plan 2434 1644
No of Policies (000's) Assurances 1 ,131 ,1 10 1,013,890
Annuities 4,840 4,530
Pension Plan 8,040 5,060
Source: Annual Report, Life lnsurance Corporation of lndia 2001
7. Year # of policies sold
(000's)
Total Sum assured
(Rs. crorest
1999-2000 16.989.000 91,490
2000-2001 1,96,65,000 1,24,950
2001-2002 2,32,50,079 1,92,575
2002-2003 2,45,29,946 1,79,693
Table 1.2: Growth of Life lnsurance in lndia
Source: Annual Reporis, Life lnsurance Corporation of tndia
The numbers of pension policies sold are too low at 8040 conpared to
the number of total assurances over 1 crores. The penetration of insurance
assessed as a ratio of the insurance premium to the Gross Domestic Savirqs
(GDS) was 9%1 in India in 1999. Therefore there exists wide potential for lrfe
insurers to not only increase their penetration in the Indian market but also to
widen their offering given the size of the market.
1.2 Marketing of Life insurance in India
Life insurance contracts are typically sold to customers through agents
rather than bought. The concept of buying an ihsurance product to compensate
for an accidental event is not popular, since life insurance policies are compared
with bank fixed deposits or mutual funds, where the returns are usually higher.
Therefore insurance as a product is typically sold to the customer rather than
nnual report Insurance Regulatory Development Authority 2000
8. nan(etrng approach in selling its policies. With a shift from the product and
se#tog concepts to the marketing concept, the insurance companies appear to
have begun analysing market opporlunities and to develop newer target
rnarkets. Therefore, there is a need to examine the possibilities of demographic
segrrentation in the indian market, Fufther, there is also a need to evaluate the
possibilities of product positioning across a viable business strategy.
1.3 Objectives of the study
The objectives of the study includes;
. To identify and develop customer segments based on key
demographic attributes for target marketing of life insurance product for a given
region in lndia, using modern data mining techniques
. To suggest an appropriate business strategy across the segments
for better positioning of life insurance products and increasing profitability.
1.4 OperationalDefinition
o ff life insurance policy is a policy that promises to cover the
proposer (or its nominee) against the event of the occurrence of death, where
such a promise arises out of the contractual agreement between the proposer
and the insurer.
o I common and popular life insurance policy means the most widely
known (awareness about that product) policy of a most widely known insurer
devised on easier terms and more common demographic criteria.
10
9. . Demographic attributes means usual attributes of a census
population such as age, gender, location or it can also mean wider attributes
suci as income, health information, education, etc.
o Responders to a marketing campaign means an observation
existing in a marketing cluster that has met the policy criteria and hence has a
higher likelihood of being a customer,
. Profits from the business segments are measured by a nreasure of
the responders and their profiles.
1.5 Design of research
A research design is a framework for conducting the marketing researcft
project. tt details procedures necessary for obtaining the information needed to
structure or solve marketing research problems. lt may be mentioned here that
research design followed here is conclusive, which is more formal and
structured. lt is based on large representative samples and the data obtained
are subjected to quantitative analysis, The findings from this research are
considered to be conclusive in nature that they are could be used as input into
managerial decision making. Conclusive research designs could be either
descriptive or causal. We follow the descriptive research here to describe tfp
potential marketing characteristics attributes.
Why use descriptive research?
Descriptive research is conducted for the various reasons such as rnarket
studies, market share studies, sales studies or brand image studies. Market
studies could describe the size of the market, buying power of the consumers.
u
10. availability of distributors and consumer profiles. Market share studies coulc
determine the proporlion of total sales received by a company and rts
corrpetitors, or sales analysis studies, which describe sales by geographic
region. product line, type and size of the account. lmage studies, which
determine consumer perception of the firm. Such research are also done to
describe the characteristics of relevant groups, such as consumers or market
areas. For example, we could develop a profile of the ,,likely
consumers,, of a
company' lt could be for estimating the percentage of units in a specified
population exhibiting a ceftain behavior. For example the percentage of
consumers of a company going for a particular product. Sometimes it is used for
determining the perceptions of product characteristics, determining the degree to
which marketing variables are associated. For example, to what extent buying
product A is related to buying product B? Descriptive research could also help
us make specific predictions. For example, what will be the retail marketing
potential for selling product in a given regton, etc.
These example show that descriptive research assumes that the
researcher has much prior knowledge about the problem or situation.
Descriptive research designs could be either cross sectional or
longitudinal' Research process is formal and structured. Cross sectional study
is the most frequently used descriptive design in marketing research. Cross
sectional design involve collection of information from any given sample of
population elements only once. Similarly, cross section design could be either
single or multiple cross section.
1:
11. 1.6 Scope of the Study
The purview of non-life insurance products are out of the scope of the
study. Given the lack of wider differentiation of life insurance products in the
Indian market, this study attempts to cover a few popular life insurance policies.
The project intends to focus on the demographics of a given geographical
region. Socio-economics of the same region may be out of the scope.
Other issues in marketing such as pricing methods, Vpes and means of
product development, channels of marketing, methods of promotion. etc are out
of the scope of the study.
1.7 Limitations of the Study
As mentioned in the scope of the project covers identifying marketable
segments for a given geographical region based on demographic information.
using modern data mining techniques. The validation of the findings on
population characteristics is done in sample and not out sample, since survey
data for the same responders may not be easily available for a region in two
time periods. The applicability of a marketing solution is not only limited to the
availability of such information for a region, but is also limited to applying for a
different region. The usual sampling limitations in applying secondary data also
exist.
1.8 Data
Demographic and socio-economic data available to a marketer includes
NSSO data, Census data, Credit Bureau data, etc which are published sources
of information. Similarly such data is also collected by market research agencies
13
12. -l- 3 f,''3^ regjon' However. the block design approach to sampling of NSSO
-:-= 3i€s obvious advantages _to be used in our anarysis. Fufther, NSS9
=rciEts annual surueys that cover various FSEs including the unit level data.
Smady' NSSO suryey also comprises the expenditure information of a
-=c'onder that courd be used as a proxy for income in the anarysis.
The secondary data will be obtained from annual household survey data
cr the National sample survey organisation. The National sample Survey
organisation, an apex agency under the ministry of statistics and programe
implementation of the Govt' of India conducts annual household surveys of both
urban and rural areas' The customer data which would form the benchmark for a
marketing approach would be collected from Llc. The Life lnsurance
corporation of India is the largest life insurance company in the country that
covers the widest spectrum of customer demographics profiles in view of its
widest variety of life policies offered. There could be limitations to the collection
of both the NSSo data and the customer data for the country as a whole. Hence
this study intends to appry the demographics of Karnataka urban onry,
The rest of the project reporl is organised as follows; Chapter 2 deals
with the need to identify customer segments based on key demographic
attributes and the significance of data mining.approach to insurance marketing.
It also describes the advantages of a segmentation approach to a marketer and
the literature on the techniques of applied data mining. chapter 3 enlightens us
the possibilities of product differentiation and the need for an appropriate
positioning across the segments as a viable business strategy taking the
It
13. exanple of the products of LlC. Chapter 4 discusses the results of the analysis
ard tre business modalities based on the segmentation outcome in scope.
Fnally. Ghapter 5 concludes the story drawing attention to business
inplications, highlighting the limitations of this study and also points to future
directions for research.
15
14. CHAPTER 2
DATA MINING AND INSURANCE MARKETING
The insurer takes a decision as to sell a particular policy to a person on
the basis of the information disclosed by the buyer himself in the proposal form.
Therefore the insurer uses some criteria to evaluate the worlhiness of the
proposer to become a customer. Hence any proposer courd not become a
customer by itself as some element of classification always exists at the end of
the marketing deparlment.
The first objective of the analylis is to identify and develop customer
segments based on key demographic attributes for target marketing of life
insurance product for a given region in India using modern data mining
techniques.
Why customer segments?
lnsurance companies need to focus on customer segments to fulfill the
following objectives;
' To precisely identify the customers from among the marketable
universe of responders' Everyone meeting or not meeting the criteria is not
going to actually purchase a policy and the key question in marketing becomes
identifying marketabre seEments of genuine buyers for a given geographical
locality.
16
15. * ;,#:'.:il:,Jffi" :_,,":,::Tl,:,": ;", j::
rE- needs better' For example, an insurance product for aterm insurance
prodrrct may target an age group of 30-50, where as a pension product mav
target any one at a higher age group only.
' To gain segment leadership. lt is true that in modern day business
fierce competition exists in the consumer markets and hence no company can
easify aspire to become a market leader in ail products. However companies can
do aspire to become product readers in the same market. For exampre,
endowment products are very popu rar for some companies where as the
pension poficies are more popular in some other companies. These refer to
different segments of the same market where two different companies do better
by effectively targeting them.
o Increasing sares. Traditionar approaches to marketing of rife
insurance product invorve attempts to increase the customer base by simpry
expanding the efforts of the sares department at higher budgets. However
segmentation could help increase the sales by soriciting prospects who would
respond to a marketing campaign. This could effectively utilize the marketing
budget' segmenting the universe of potentiar customers to focus on specific
groups can make marketing campaigns more efficient and furlher increase the
return per unit of marketing effort. since the budget for marketing is limited the
promotion efforls cannot seek to target everyone in a region. Hence this cars for
targeting people who are meeting certain criteria going in for an insurance
t7
16. !€l -r. Focus on marketing communication. Marketing communication serves
st€ nrcst effective method of soliciting prospects as to be customers and
tlerefore it is essential to have the right creative for the right customer segment.
. Increasing profitability. Business planning involves selling the
products to the customer at a given price. However, the same product gives
different value to different customers. Customers would be willing to pay more
for a product depending up on the value they get from the use of a product.
Customer segmentation helps in better pricing based on the values so that the
company's profitability could be increased. The benefits of objective
segmentations are manifold. Insurers can increase profitability by identifying the
most lucrative customer segments and then prioritize marketing campaigns
accordingly. Problems with profitability occur if firms do not offer the "right" policy
or the "right" rate to the "right" customer segment at the "right" time.
. Gaining market knowledge. Database segmentation help the
insurer develop a process around which the strategy followed in one segment
could be copied to other segment. Knowing why and how to treat each customer
segment requires understanding of the secrets in marketing data. The most
valuable data for marketers is buried in their own database. The problem is that
marketers may not know how to look at their own data to unlock the secrets that
will help them communicate more effectively with their customers and prospects,
thereby developing its knowledge and skill in understanding the market better,
What are the advantages of segmentation?
18
17. It costs 8 to 10 times as much to make the first sale to a prospect as it
does to make another sale to a customer. Spending 20 times as much to sell a
prospect as to resell a customer is not uncommon. Typically, buyers who just
rnade a purchas e are twice as likely to respond to a new offer as buyers from
one year ago. One-year-ago buyers are twice as likely to respond as two-year-
ago buyers and so on, This explains why a smafi marketer will contact new
customers more often. Most marketers know that 50o,," to 60oo of their customers
are one-time buyers. Half of them have bought only one of their leading items.
This tends to be a huge group-one-quafter to one-third of the database. They
are easily identified, and they are much harder to resell than one-time buyers
who initially purchased more than one item. Having a strategy to deal with them
is crucial to gaining more two-time buyers. Buyers who spend a small amount
on their initial purchase are likely to continue to spend less than average on
future purchases. The amount (per order) customers have spent in the past is
usually an excellent indicator of how much they will spend (per order) in the
future. past average order size is often a better predictor than overall monetary
value. Better to communicate now, while the first sale is fresh, than to wait until
the relationship with the customer is stale. Hence the advantage of
segmentatic,r not only lies in reaching a prospect for a sell but also lies in
making a cross sell provided the customer is correctly identified in a segment
and hence it brings in the advantage of customer retention also'
What are the criteria for segmentation?
I9
18. Segmentation criteria could involve demographic, psycho-graphic. social
economic and geographic profiling. This study involves profiling of the customers
by dernographic and socio economic variables only. Geographic and psycho
graphic profiling are out of the scope.
Demographic segmentation involves profiling the prospects by common
attributes like age, sex and rural or urban, etc. socio economic. profiling involves
segmenting the customer by occupation viz; farmer, salaried, labourer, self
employed or others, etc. Economic profiling involves dividing the prospects
based on their annual income.
Why segment a region in India?
It would be ideal to conduct a demographic segmentation exercise for
both the urban and rural lndia as a whole. However the data limitations and the
limitations of computation would make it difficult to conduct a segmentation
exercise for the country as a whole. Further even if it is true that we could never
design a product for one marketing region only, it is true which every marketer
understands that certain products could be better sold in one region than in
otl'er region' Therefore the understanding of the market a given region could
throw beit€f understating of the market as a whole. Hence this study perlains to
the demographics of Karnataka urban only.
of late, there has been a dramatic surge in the level of interest in data
mining, with business users wanting to take advantage of the technolog y for a
competitive edge' The growing interest in data mining has also resulted in the
introduction of a myriad of commercial products, each described with a set cf
19. terms that sound similar, but in fact refer to very different functionality and based
on distinct technical approaches, The advantages of using data mining
technique in the modern day could not be undermined.
How does modern data mining technique helps build segmentation?
Data mining can be defined as the process of seleoting exclonng and
modeling large amount of data to uncover previously unknc,'.r paii=:rs. !'.iodern
techniques that are guided by more quantitative data minin,c asc'secies cai
lead to more focused and better results of marketing campaigns.
Hence modern data mining tools seek to provide business tnte -:er:e
solutions used for marketing. Data base segmentation is the most con:,'nor'
application of data mining solution. Modern day companies spend large bud.oe:s
for collecting and recording information. Judicious application of data mining
could help save a lot to these companies.
What are the techniques of data mining ?
Data mining algorithms are designed with specific kinds of predictions
and specific types of input data in mind. lf the goal of a data mining approach ts
to estimate some kind of numeric quality, the data mining algorithm need to
produce a numeric output. Data mining tasks could be either directed or
undirected. In directed data mining the goal is to predict, estimate, classify, or
characterize the behavior of some pre-identified target variable. In undirected
data mining there is no target variable to be predicted. Instead, the goal is to
discover structure in the data set as a whole. Examples of directed data mining
include determing which pockets or group could be better targeted for a product
21
20. !
h
h
t
b
l
l
t
F
F
|,
l't
-a
b
-t
-D
l:*-33r!r-l !.''hen the target variable is the existence of the product. Examples of
--:'aoted data mining include determining what products should be grouped
:::3iner. finding groups of prospects with similar tastes, and discovering natural
:-slomer segments for market analysis. The literature describes three data
-rning techniques called
. Automatic cluster detection,
. Decision trees. and
o Neural networking.
Out of the 3 methods outlined here the most common being that of
automatic clustering and decision tree, which we are going to describe nere.
Autom atic c I u ster d etect io n
The undirected data mining is the most commonly implemented algorithm
of automatic cluster detection. There are many mathematical approaches to
finding clusters in the data. Some methods called divisive methods, starl by
considering all records to be one parl of a big clusters, which are themselves
split until into two parls or smaller clusters, furlher themselves split into each
record level. At each step of the process, some measure of the value of the
splits is recorded so that the best set of clusters can be chosen at the end. Other
methods, called agglomerative methods, staft with each record occupying a
separate cluster, and iteratively combine clusters until there is one big one
containing all the records. There are also self oraganizing maps, a specliazed
form of neural network that can be used for cruster detection.
22
21. Among all the 3 automatic methods, the most commonly used technique
is the K means clustering algorithm. lt is used in a wide variety of commercial
data mining tools and is more easily explained than most. lt works best when the
input data is primarily numeric. The algorithm divides the data into a
predetermined number of clusters, called K. The statisticians calls an average
that refers to the average location of all of the members of a par-ticular cluster,
What does it mean to say that cluster members have a location r.,,hen they are
records from the database? This courd be explained by gecmetry. To form
clusters, each record is mapped to a point in record space. The space has many
dimensions, as there are fields in the records. The values of each fielis is
interpreted as a distance from the origin along the corresponding axis cj tre
space' ln order for this geometry interpretation to be useful, the fields must atl be
convefted to numbers and the numbers must be normalised so that a change in
one dimension is comparable to change in another. Records are assigned to
clusters through an iterative process that stafts with clusters centered at
essentially random locations in record space and moves the cluster centroids
around until each one is actually at the center of some cluster of records. In the
first step we select k data points to the seeds, more or less arbitrarily. Each of
the seeds is an embryonic cruster with onry one erement.
However there exists limitations to the approach of clustering also. The
choice of clustering as the data mining technique to apply to a problem has
consequences of the kinds of questions that can be addressed and for the kinC
of data preparation that will be required. Because automatic clustering detection
F
h
h
b
22. Classification trees label records and assign them to the proper class.
Classification tress can be provided the confidence that the classification is
correct. In this class, the classification tree repofts the class probability. which is
the confidence that a record in a given class.
Regression trees estimate the value of a target variable that takes on
numeric values. All of these trees have the same structure. When a tree rnodel
is applied to data, each record flows through the tree along a path determined by
a series of tests such as " is filed 3 greater Ihan 27"? Or " is filed 4 red. green or
blue"? Until the record reaches a leaf or terminal node of the tree. There it is
given a class label based on the class of the records that reached that node in
the training set, in the case of regression trees, assigned value based on the
mean of the values that reached that leaf in the training set.
Various decision tree algorithms such as CHAID, C4.5/C5.0, CART. and
many with less familiar acronyms, produce trees that differ from one another in
the number of splits allowed at each level of the tree. How are the split chosen
when the tree is built?
Decision trees are built by a process called recursive partitioning.
Recursive partitioning is an iterative process of splitting the data up into
parlitions and then splitting it up some mole. Initially all the records in the
training set-the pre classified record that are used to determine the structure of
the tree are together in the big box. The algorithm then tries breaking up the
data. The algorithm chooses the split that partitions the data into two parts that
are purer than the original. The process stafts with the training set consisting of
1i
23. cr+-classified records. Pre classified records means that the target field, or
oependent variable has a known class. The goal is to build a tree that
:istinguishes among the classes, For simplicity assume that there are only two
target classes and that each split is a binary parlitioning. The splitting criterion
easily generalizes to multiple classes and clearly any multiway paftitioning can
be achieved through repeated binary splits. We do not loose much information
by addressing the simpler case in order to make the explanation easier to follow.
The first task is to decide which of the independent variables make the
best split. The best split is defined as the one that does the best job of
separating the records into groups where a single class predominates. The
measure could evaluate a potential splitter is the reduction in diversity. The
process continues until more splits are found, For simplicity assume that there
are only two classes and that each split is a binary partitioning. The splitting
criterion easily generlises to multiple classes and clearly any multiway
parlitioning can be achieved through repeated binary splits. Decision trees
method are good choice when the data mining task is classification of records or
prediction of outcomes. lt is used when your goal is to assign each record to one
of the few broad categories. Decision trees are also a natural choice when your
goal is to generate rules that can be easily understood. Similarly it is also used
to prioritize independent variables. Another useful consequences of the way that
imporlant variables float to the top that it becomes very easy to spot input
variables that are doing too good a job of prediction because they encode
26
24. (^3,',:edge of the outcome that is available in the training data, but would not be
a.ailable in the filed.
Therefore the choice of the data mining would not only depend upon the
l-rpe and kind of database that the marketer has in hand, but it also depends
upon the objective in mind. lf the objective is to form natural demographic
segments that could be used for any marketing such as life marketing, non life
marketing, or even for any consumer product it would be ideal to go for
clustering exercise. Furlher clustering is an ideal choice when the marker has
absolutely no idea about the prospects or the product profiles. However when
the marketer in question has the sole objective of separating the marketable
segments from the non marketable segments by the objective of a customer-
product in mind, it is suggested that the marketer should go for a decisions tree
based approach so that as far as possible, the maximum number of customers
responders could be identified in minimum number of segments.
This answers to our question of deciding on a tree based approach rather
than a clustering approach to analyze our research problem.
2l
25. CHAPTER-3
BUSINESS STRATEGY
The purchase of one form of life insurance can differ in many ways from
the purchase of other savings or investment products or it can also differ from
the purchase of other forms of insurance, since it combines protection with
savings' There exists controversy with respect to the proponents of permanent
insurance as an investment and the advocates of pure protection. Insurance
investments are also criticised for low returns on the consumer,s money. Term
insurance is the insurance that covers the risk of death only. some believe that
the consumer would be better of by buying term insurance and investing
separately the difference in premiums between term and permanent insurance.
some amount of flexibility does exist when part of the money comes back to the
proposer at fixed intervals. Sometimes the insurers attempt to provide a choice
between term and permanent insurance in terms of protection than an
investment decision by pointing out that the term insurance can not be continued
beyond the age of 50. The product is so designed than the individual when
survives gets nothing out of the poricy at maturity? why is the upper age rimit of
entry is at 50? By age 50 the economic value of the individual has declined
substantially' Further, one has little need for income earning protection at this
age' Besides, rising infration may seriousry reduce the rate of retum m
insurance investments. Hence, if the individuar has no need for protectim
26. Fr
Iar
a
h
Cr
FD
cD
.r
*,r
AD
.,
Ao
TO
+o
+r
tr
rF
+t
+.
F
F
t-t
1-
r
"i
'F
FD
t
,f"
{,
h
")z-s:
3'otection of death, it is unlikely that life insurance will match other
- . :s:'l€nt alte rnatives.
Similarly, there also exists a savings argument for the purchase of life
^sLrrance for the purpose of investment. Since it is the compulsion of savings
:'rat it entails, Many people do not have the discipline of savings for the regular
contribution to a savings product. and buying an insurance policy brings in the
discipline of savings,
Given the above premises, the need for insurance is primarily because of
the need for protection against dea:h primarily, The need for investment does
exist to the extent that the product is flexible enough to generate some income
during the term of the policy.
Therefore, a business strategy for insurance marketing is indeed
challenging, given the need of the consumers.
The second objective of this project is to suggest an appropriate business
strategy across the segments for better positioning of life insurance products
and increasing profitability.
A buyer when takes a decision to buy an insurance product he could
evaluate the same in the light of, How much to buy? What kind of product should
it buy? Where from should he buy, etc?
What are the product offerings existing in the market?
The market offers a wide range of insurance product all of them are quite
popular and which have evolved over the years. lt could be whole life. terr:
assurance, endowment, money back, children plans, pension plans or a
27. thD
ID
h
a
a
Gr
A!
AD
Fr
*!
*;
*i
*rt
hD
*!
hD
-;.".'€ -a.i: Lle products in lndia
F:,.;-, Tlpe Policy
Coverage
Minm
Age
Maxm
Age
Minm
sum
assured
Maxm sum
assured
Maxm annual
Premium for
Rs 1 lacs
r'r'rc.e life Provides'
!.r.rctr' Protection
, exclusivelY for
I
I dePendents
15 60 30,000 No limit 7,936
Endor,vment
Policy
Provides
family
protection and
future
provision
12 65 30,cc0 No limit 22,331
Money back
policy
Money at the
end of every
five years +
risk cover
13 50 40,000 No limit 8,038
Combination
of whole life
ano money
back plan
Risk cover
through out
the life
13 65 2,00,000 No limit 31,395
Pension
Plan
1B 65 10,000
Term plan Risk cover
through out
the life
18 50 3 crores 300
Children
Plans
Money Back
for children
0 10 1,00,000 25,00,000 16,239
Source: Handbook of LIC Agents 2003
28. ,ID
,D
t
t
t
t
t
I
t
h
t-t
t
+.
r-h
+r
+t
,+
rfr
h
hD
ht
h
hD
qt
ft
aD
a
C
:c't"cination of all of the above, enumerated in Table 3.1 here. lt also lists the
e -:rcility terms and premiums for a standard sum assured of Rs 1 lakhs. Whole
t 'e policies are cover the risk through out the life of the porposer including a
payment of bonus along with the maturity sum, lt is a high risk cover plan that
also provides financial security. Endowment is the most popular plan for fulfilling
long and short term financial needs including cover for a term. The proposer
gets the sum assured including bonus at marurity. Money back gives money
back to the proposer at fixed intervals and cover for a term that includes bonus
at the end of the term and balance of the sum assured at maturity. Term
assurances are the low premium pure life products where the survivor gets
nothing at maturity for fixed term. Children's plan are the plan that give cover to
the children and the guardian proposer including money back to the child at fixed
interval during the term. Pension plans are old age annuity plans that gives a
sum assured at maturity, cover during the term and annuity till death.
The next question of how much to buy is an argument better debated with
the estimate of the net present value of the stream of income that a family needs
to survive after the death of the proposer or is an estimate of his economic life.
The science of determining how much to buy is under the purview of risk
management and hence it is out of the scope of'this study.
The cosumers may have a wide choice among the companies that sells
insurance products in the market. While choosing among the companies that the
buyer may consider many factors such as financial strength of the seller. Wpes
of policies offered and a policy that would meet his requirement that t:ra
29. conpany provides. The other imporlant question the consumer would ask is on
the price of the product? what is the premium for a policy? why are the
prerniums different across insurers? Some insurers may offer pafiicipating
policies to non participating policies. Parlicipating policies are those where the
proposer is assigned to a benefit that the insurer makes, commonly in the form
of dividends. The differences in premiums between insurers could be on account
of actuarial differences, differing rates at which the savings elements if the
contracts accumulate, to differences in dividends, to varying levels of expense
and insurer profit. commissions to sales agents also differ from companies to
companies resulting in wide differentials in the premium' New companies' in
pafticular which are attempting to grow rapidly often pay substantially higher
commissions to their agents and those commissions are passed on to
consumers.
These discussions outline the background thoughts that an insurer must
have while designing an effective business strategy' After the market has been
Separated into its segments, the marketer has to select segment as targets'
Resources and efforl will be targeted at the segment, called positioning. As
discussed above, the marketer could choose to target a single product offering
at a single segment or the marketer may target a single product at all segments
or may target multiple products at multiple segments' We will throw more light on
the above strategy given the results of our segmentation in the next chapter'
why positioning? Positioning your product is defining who you are in the
mind of your customer. The process of positioning involves a frame of reference
)2.
30. or category and the uniqueness of your product. The process of positioning also
helps you to become very clear about how your product or service is different
from your competitors, and how to communicate this uniqueness to your
customers.
How to position? The following questions need to be answered. Who
would find this an ideal product? What is the single biggest reason that
customers would buy this product? What is the difference between you and the
competitor? There are many different positioning methods. you can position by
attribute, which is typical in the automobile industry. BMW stresses handling and
engineering efficiency, Volvo has emphasized safety and durability. An insurer
can find a product attribute such as "term rider", "double cove/,, ,,triple
coved,,
"dual cove/', "accident ridef', "stock option", ,,mutual
fund option,,, ,,health
rider,,,
"housing loan cover", etc. you can position by price or quarity, which is how
consumers distinguish between a Maruti 800 and Fiat Padmini at the low end,
Maruti Zen and Honda city on the high end. Pure Term pollicies could be a good
example both at high and low end. Similarly, money back policies are typically
designed at the higher end of the market. You can position with respect to use or
application, such as Nescafe coffee being po.sitioned as the best coffee for ,
morning. For example a children's plan, a marriage plan, a critical health
insurance plan etc could be positioned as the use or application. position with
respect to a competitor as GE does since "it is number one in the market it does
not have to try harde/'. Other example of positioning could be age when an
insurer could go for a specific product at each age segment. For example apart
31. frqn children plans, the insurer could give plans that seeks to protect housing
ban payrrents to an age group. There could also exist gender based positioning
by devising female child plans, female health plans, female education plans or
fernale group plans. Does LIC do positioning? Yes. LIC has a variety of products
ainred at various segments of the market. The products are well differentiated in
terms of the needs of the customers and the offerings of the competitors. For
example LIC products include, pure term products, endowment products, money
back products, pension products, pure savings products and children savings
products, etc which are aimed at different segments of the market.
How it works? A target market defines the group of people most likely to
purchase your products. Your target market might be the male wage earner in
the age group of 25 to 54 and having income above 20,000 per annum. A frame
of reference includes all the options your target market has available to fulfill
their need for a product or service, including substitute products. Your point of
difference or uniqueness is the benefit that buyers get from your product that
they can't get anywhere else.
T{rerefore positioning is successful only when the positioning strategy
results in profits for the company. Effective positioning is achieved when the
segmental needs of a customers, are appropriately positioned with the
marketers product. Hence a profitable scenario is the one when the you the
targeted product is actually purchased by the segments.
_1-t
32. The business decision for appropriate positioning could be taken only
*sr fie profiling of the segments are complete. Hence we discuss our strategy
h tE resr.rfts section explained in Chapter 4.
_3_5
33. CHAPTER 4
SAMPLE SEGMENTS AND RESULTS
Data
The secondary data used in this analysis was obtained from annual
household survey data of the National Sample survey organisation. The
National Sampre survey organisation, an apex agency under the ministry of
statistics and programrne imptementation of the Govt. of lndia conducts annual
household surveys of both urban and rural areas. This study covers NSS9
schedule 1:O:3 survey data for both rural and urban Karnataka for 51"1, 52nd. sgrd
and 54th round during the period 1990_2OOO.
Schedule 1'o of NSSO FSE rounds comprise house hold level data for
the following imporlant demographic attributes which are very imporlant for a
marketing campaign; size of household, occupation code, type of household, per
capita expendiiure, type of dwelling, dge, sex, marital status, education, etc.
These data are collected at unit levels or at the ievel of households.
Life Insurance Corporation of lndia is the largest life insurance company
in the country that covers the widest spectrum of customer demographics
profiles in view of its widest variety of life policies offered. The Bangalore I
division comprises the areas of Bangalore and urban therefore, the Bangalore I
division could serve as a good representative division for urban customers. The
4.1
36
34. LJC data comprises customer informalion such as proposal Number, proposa)
codle. po{icy code. policy type, income of the policy holder, age of the policy
nolder. sex. first second poricy, etc. Appendix 4 &% give a sampre print of the
NSSO data and the LIC data used in the analysis.
4.2 Sample preparation
The LIC sample was a random pull of 10,000 customers existing on their
books as on November 2002. Since the LIC sample comprised the data for
urban customers we combined the schedule 1.0 NSSO data for urban Karnataka
comprising 18205 records with the LIC sample of 10000 records to obtain a
response rate of ss%. (See Appendix lll for the data flow diagram). Appendix
lV & V give a sample print out.
4.3 Distribution of sample
Tables 4'1, 4'2,4.3 and 4.4 here describe here the income distribution,
age distribution, distribution of occupation in the sample, The Andersen Darling
Normality test confirmed that the data was normal at p value of 0.05.
Table 4.1: lncome distribution in the sample (Rs. OOOI
TYPE #
sample
size
MEAN MINIMUM MAXIMUM
CUSTOMERS 10,000 83.45 1.0 999.0
PROSPECTS 18,205 8.O2 1.0 82.0
37
35. Fr
'D
h
h
c,
^D
.D
.o
a{i
.ti
*tt
h.
*.
*,
+rf
.h
*i
*
-D
h
.l-
fD
h
h
t
h
Fr
t,
h
h
h
Tne average income of Llc customers seems to be much higher than that
3'n€ NSSO data. Similarly the average age of sample in the LIC customer data
s lclrer than that of NSSO data.
Table 4.2: Age distribution in the sample
LIC customer belonged to various occupational groups. However for the
purpose of a marketing segmentation we have broadly identified 4 groups called,
labourers, self-employed, regular wage and others. The Llc sample comprises
almost equal proportion of salaried and self_employed.
Table 4.3: Occupation distribution of LIC customers
ryPE # MEAN MINIMUM MAXIMUM
UUS I UMTHS 10,000 19,0 0 49
I-HUSPtCTS 18,205 26.0 0 99
OCCUPATION
ryPE
LABOURER
1+ OF OUSTOMERS % CUSTOMERS
76 0.8%
DtrLI- trIVIPLUYIIJ 3,664 36%
NtrUULAH WAGE 3,513 35%
UIHtrH 2,747 28%
IUIAL 10,000 100%
# OF CUSTOMERS % CUSTOMERS
Table 4.4: Distribution of LIC customers bv sex
38
36. -TD
c+,t
f_,
a
C-r
..
8D
Grt
C.
alt
.t
+r
.trr
+r
{-
t
.f
ht
4o
I
.hD
i
FD
F
{rt
e
tr
h
r,
fr
rt
rr
ff
i
Taole 4.5: Distribution of LIC customers bv Income
4.4 Segmentation:
The decision tree is depicted in Appendix ll.
The iechnique followed here is called CHAID, Chi-Square Automatic
llnteraction detector, which is the data, mining technique of objective
segmentation. The objective here is to identify the best tree, the branches of
which contain the maximum separable proporlion of LIC customers. The details
of applying the technique of decision trees have been explained in Chapter 2.
LIC customers are the responders in the analysis and are the target variable to
r.t:- = 7 527 73%
r.r.,ilE | 2.673 27%
10,000 100%
f
"come
Range
<1 000
1 000-1 0000
1 0000-20000
|| of customers
'/oage o
:ustomers % of Cum customers
250 t% )%
i43 3% )%
1 301 13% 22%
20000-50000 2288 23% 45%
50000-
i
t100000 l1 36 31% 767"
> 100000 2101 1"/" 007"
39
37. ht
h,
i-
i
q
h
fb
tf-
T"
rtt
ht
tht
f,
F
ht
]tr
tF-
t,
Ti
h
t-
h
t-
tr
h
fr
t"
F
b
)
h
b
be segregated. The pockets are then selected as segments when the Chi-
Square value of separation is significant at p value of 1oh.
The independent numeric variables used in the analysis are age and
income' Other categorical variables such as occupation, location, sex, marital
status, etc were also used in the analysis The per capita expenditure reported in
the NSSo data were used as the proxy for income of the household.
4.5 Sample Validation
The validation of the results is done in sample i.e. by breaking the sample
into two groups of 50% each called the development sample and the hold out
sample (See Appendix ltl). The validation is done trying to redevlop the tree on
the hold out sample.
4.6 Results
The results of the segmentation exercise identify a total of 8 segments
that could be targeted for a marl<eting campaign (See Chart 2). The union of the
8 segments would form the marketable segment of the universe. Each segment
is identified by three principal attributes available (occupation, house hold
income and age). For the purpose of easier identification the sample preparation
has grouped occupation into four categories only. Against each of the
occupation categories we have two groups of income called low income and
high income. The low income groups are not favorable because their response
rates are much lower compared to the high income group. Fufther, against each
of the high income group we have 3 segments that could be identified by age,
The teens, the young and the middle age. The teens and the young are the most
40
38. -t
-t
-r
t
I
t
t
D
hr
h
+.t
h
h
t-
F,
t--
IP
R
ll-
hr
P
F
Fr
l.
F
F
b
favorable segments for their higher response rates than the middle age
segment' Hence the total number of marketable segments are 8. However, the
tree does not end here. We could still grow the tree beyond the middle age
segment when we want to furlher churn higher income group among the middle
age segment.
Gain's charl (Chart 1) here is the plot of the performance of the decision
tree' The gain's charl explains the lift obtained in churning out the responders
while separating the responders from the non responders. called customers from
the sample. As shown here the decision tree captures over g5% of the
responders in over S% of the sample.
Gain's Chart
U)
L
E
c
o
o
L
c)
-
o
o
o,
(d
s
100%
80%
60%
40%
20%
0%
--
o% 11% 20% 34% 35% 3B% 64% B9% 100%
"/" age of the sample
Chartl:LiftChart
When we combine these B segments we get a response rate of g3o/" at
34"h of the sample. Hence the rift here is (gg%lg4% = ) 2.74 times. The
41
39. 3-s res's implications of a gain of 2.74 times implies that we could save mailinq
a.3s:s for a campaign by 2J4 times.
:.A Discussion
As shown here Chart 2, these segments are identified by the occupation,
Income and age of the responders. other attributes such as location, sex, house
hold size are not significant in separating the responders,
Given the definition of these segments we could combine these g
segments to form 4 principal segments defined by occupation, These 4 principal
segments would form the marketing base of an insurer in urban Karnataka.
Each of these occupation segments are defined by an income limit exceeding
Rs' 1 40001- and an upper age limit. The income limit for regular wage (salaried
class) is Rs 25,ooo/-' lt simply means that the likelihood of customers in the
range of Rs 14000/- to Rs25OOO/- are row. probabry emproyed peopre berow the
annual income of Rs 25oool- have some other form of insurance through their
employer or Government. similarly, the category of others define their segment
for an upper age limit of 45' others occupation category include small retailers,
barbers' shop owners or any other form of income where the potential to earn
could be higher at an increasing age.
Fufther, the upper age limit of 35 for all the segments and 45 tor others
does not necessarily mean that the insurer would stop under writing any body
who has exceeded this age. lt means that there could be higher income pockets
beyond these age group who courd be targeted with an appropriate product.
+_
40. Similarly, the age group below 35 are the combination of two segments
called the teen for 19 and the young for 35. The teen and the young would have
differing product needs and hence both of them should be targeted differently.
4.7 Business Strategy
The next step in marketing would involve examining the possibilities of
targeting these segments appropriately looking at the benefits of doing so. The
product offerings existing in the market has already been discussed in detail in
Chapter 3, which identified term assurance, whole life, endowment, money
back, pension plan and children's plan, etc. These products are characterized
?S, pure insurance, pure insurance and investment and pure insurance,
investment and periodic returns.
The positioning could be as follows to result in gains for the marketer, An
insurance policy called term insurance could be solicited to any one in the data
base whose income is exceeding Rs. 14ooo/- (Rs 25000/- if he belongs to the
class of regular wage) and is younger than 45 (or 35 if he does not belong to the
class of regular wage). However, it wourd depend up on the income of the
individual and his occupation group to find out whether a whole life policy would
be more suitable for him than the term policy. Further a pension product again
would be eligible for any one in the database who falls into the young segment
and does qualify for an insurance based on annual income.
Could there be any one within the database who is eligible for not only a
pure insurance like the term assurance but an investment product combined with
life protection or vice versa. Are there any furlher segmentation opportunitbs
4_r
41. F,
F
!h|)
G,
A,t
Ih'
',
.hr
h!
F,
t
ht
h
h
h
Ft
h
GI
Fr
hD
rr
GD
qt
F'
h
q
F
fr
h
rt
a
rtr-lr;fl the results for better positioning of wider products? There could be. where
are the opportunities for more penetration? Looking at Appendix 2 the young
and the middle age segment could be drilled down for further opportunities.
Among each occupation segment, the age segments of 1g-35 and 35-gg (45_gg
for the regular wage class) could be grown further.
The results are as shown in Table 4.2. The further opporlunities of
positioning within the major occupation segments has been outlined. We find 4
Income segments (3 income segments for the regular wage) for the young
segment of the serf-emproyed and others category, in the range of 14-25,25_50,
50-82 and 82-999. simirarry, we find 2 income segments for the middre age
segment of the self-employed and others category,
ll
.fr
42. a
OCUPATION OF THE REPONDER
OTHEBS
il ls income > 25, 000 ls income > 14, 000
ls Age <35
SOLICIT FOR LIFE INSURANCE'PRODUCT
Chaft 2 : Customer Segments
45
44. Given these additionar pockets by income, there courd be hopes of
positioning appropriate products for higher income groups and lower or middle
Income groups separately. Hence endowment plans and money back plans are
the plans for relatively middle and higher income groups. Furlher, given the
segments of income, the distinctions in targeting could be between a ,with
profit
product' and an 'without profit product,, A pension pran courd not onry be a
product for the young, it could also become a product for the high income -
middle age groups.
Finally plans for the first segment by age which is the teen age segment,
There could also be further income pockets within the teen age segment.
However adequate differentiated offerings should exist to be able to target each
income pocket' children's and related plan coverthe age group of o-1g in most
of the cases' The children's plan is a insurance for the proposer, combined with
money back for the ward at fixed intervals. Variations to the basic plan could be
based on the requirements of this teen age group at various ages, All entrants
should be targeted with education plan offerings. The stream of money back
requirements for different income pockets within the teen age could also chanqe.
Female entrants courd be targeted with marriage offerings,
Hence, as discussed above, the marketer could choose to target a single
product offering at a single segment or the marketer may target a single product
at all segments or may target multiple products at multiple segments. Such
h
er
45. fr
Go
h
6ir
t
i
rD
-
E
cf
CD
i
t
i
t
t
rD
business decision are profitable only when we have a multi-product mutti-
segment strategy.
Hence' positioning is successfuf onfy whe the positioning strategy
resufts in profits for the company. Effective positioning is achieved when the
segmentar needs of a customer segments are appropriatery positioned with the
marketers' product. Hence a profitabre scenario is the one when the you the
targeted product is actually purchased by the segments.
+!
46. nr
'
,,
f-D
h,
Li
hr
tb
I
+r
It-rr
i.r
hD
h
h
hr
Fl
h
h
h
h
hD
h
h
Fr
CD
tu
dr
h
CHAPTER 5
CONCLUSIONS
5.1 : Opporlunities
The first objective of identifying and developing customer segments
based on the key demographic attributes of occupation, income and age for
urban Karnataka is successfully met here. The second objective of this study
involved two parts that comprised an appropriate business strategy for better
positioning of products and increasing profitability. We have met the first parl of
the second objective by assigning a bundle of product categories to the select
segments. However the question of profitability could not be answered in the
analysis.
Neverlheless, the opporlunities for data mining techniques for insurance
marketing are wide as outlined in this study. Data mining techniques would not
only help identify the segmental needs better. but also help the insurer reduce
marketing costs including the gains due to a better response against its
marketing campaigns.
The opporlunities of implementing this solution are also wide. Aparl from
the NSSO data that the Govt.vide NSSo/cpD dated (copy enclosed in
reference), insurer could also use the data base of the Census of india, provided
they could collect the income information also. Similarly, the recent formation of
the Credit Information bureau (ClB) that covers customer information for the
49
47. ll- r
Gr
ct,
eir
A,
+.
{-
1;tr,
trt
h,
fr
q
af''
a
Gr
fr
i-
AD
i
h
Cf
*D
ir
aD
i
lf
t
tD
financial services companies, could also give a marketable database within lr: =
for target marketing' The other avenue of database marketing could inciu:=
paftnership marketing, which could mean using the customer database of a
different product market to market your own product. For exampre, you courd
select customers within a credit card holders database. Further, insurers could
also collect survey data from various market research agencies and apply these
solution' for soliciting prospects. Therefore, the opportunities are unlimited.
5.2 : Limitations
The validation of the decision tree on population characteristics is done in
sample and not out sample, since survey data for the same responders may not
be easiiy available for a region in two time periods. out os sample validation
could statistically give more robust resutts.
Using a sample has its own limitations which could be because of
statistical bias such as aggregation bias, responder bias, collector bias, etc. For
example, random sampling error could occur because the particular sample
selected is an imperfect representation of the population of interest. lt is
reflected in the variation between true mean value for the population and the
true mean value of the original sample. Nevertheless this project does not
attempt to substantiate that the sample characteristics are similar to the
population characteristics.
The methodology and the scope of this project intends to suggest a viable
and strategic solution to marketing of life insurance products. The resutts
outlined in the sccpe of the study does icjentify marketable segments for urbar
a
-l
48. '-
l_
'-
tt
I'
t
h
t.b
I
4-
ttt
J
Ihrt
J
tht
JhD
Jb
ab
cb
h
h,
h
h
rt
i
i
?.f
t
t
'-
Karnataka and hence it does not give a generic solution for the nation as a
whole' Fufther, since the solution suggest using income as a cut off criterion a.
question could be asked on the very issue of implementation of a segmentation
based on income. lt could be termed as a means of discriminating agarnst
ceftain classes. The applicability is not only limited to the availability of such
information for a region, but is also limited to applying for a different region,
Fufther, the regulatory provisions relating to the marketing of life
insurance could also hamper the applicability of the suggested business
strategy. For example IRDA provisions imply the insurers to meet cerlain time
boltnd targeted number of policies for various categories of the customers such
as rural, weaker, female, etc.
Similarly, a marketing strategy could be successful only when the
channels of acquisition are also successfully planned. A direct marketing,
telemarketing or e-marketing or sales contact, etc could greatly affect the
efficacy of a selection. Hence, some channels are effective for some tarqeted
segments and others are effective for some other segments.
There exists obvious limitations to the computing power of modern data
mining techniques in terms of the size of the sample, which may limit the
application of this solution.
Further, these segments are dynamic in nature and due to a change in
the demographic patterns of the population the nodes may change and hence
the marketing need to revise the segmentation criterion. The question of
profitability could not be answered using the product pricing database of LlC.
49. e,
e,
Gl
e.
A,
elD
e
Returns information for LIC products are not easily available for all the
products. Secondly, due to reasons of business interest and confidentiality, such
information is not disclosed to public. Therefore, the value of profitability for each
of these segments could not be projected here.
5.2: Further Research
The data mining approach to insurance marketing is not an end in itself.
In fact a well designed and formidable marketing approach should not only
answer questions on targeting responders and converting them, it should also
answer questions on their cross sell potential and retention. lt can be mentioned
that insurance companies make a lot of money through cross sell, however a
malor poftion of recently acquired customer attire within the first two years.
Hence both the questions of attrition and cross sell are important.
The bigger challenge for the insurers is also in the area of pricing, which
is on account of growing competition in the market. A pricing question could be
better answered only when you have the data on product profitability and
product demands.
Finally, it is product development that helps a marketer survive in the
market through constant innovation of its product chain meeting the life cycle
needs of its customers.
Hence we need to look at the question of a marketing approach also from
point of view of product profitability. Furlher, simply acquiring customers is
viable unless you have acquired the creamy customers to retain them. A
a
e
ch
h
a
a
fr
a
-
h
rb
b
h
rt
i
i
i
i
t
t
t
the
not
51. c-
e-
C=)
ct
Git
c
Chr
G-
BIBILIOGRAPHY
Annual Reporl 2001, Insurance Regulatory Development Authority, Delhi
Annual Reporl 2002, Life lnsurance corporation of India, Mumbai
Aft weinstein et a|1993, Market segmentation,. Mc Graw Hill, London
Berry M and Linhoff G, 1997, Data Mining Techniques, John willey
Bhatia S, 2003, Handbook for LtC agents, Mashbra
Dorfman Mark s, 1992, Life lnsurance: a financial planning approach,
Deabrone Trade
Drumond Graeme & Evison, John strategic Marketing The charlered
Institute of Marketing B & H 2OO1
Ennew christine, 1996, Marketing of Financial seruices, Butterworth-
Heinemann
Gayle stanford, 2000, Data Mining in the rnsurance lndustry, sAS Inc
Harrison Tina, 2000, Financial seruices marketing, prentice Hall, London
Hedrick rerry E, 1993, Applied Research Design: A practical guide, sage
Ramachandran, 2001, Present and Future Scenario of lnsurance in lndia,
sahoo s c & Sinha P K, 1 gg1, Emerging Trends in tndian marketing in the
90s, Academic Foundation
Sahoo S C, 2OO0 , Seruice Marketing: Tert and Readings
Chr
€h,
*
€h,
1.
2.
3.
4.
5.
o.
7.
8.
L
10.
11.
12.
13.
14.
-
€b
rh,
fr
Cb
fr
h
h
i
.b
t
(t
i
a
i
F
i
O
O
52. Ui
(IJ
'd
-
o
E
G
c)
(5
_l
U'1
L(
o
()(
Er
(C
o:
o:
o,
CU,
ob
.s
E3
o
aln
'Fo
!)=
g;
^L
(/)ol
P
-l
(,F.l
# Ll
c(6t
ool
'i al
cb Eli
L
o
o
o
L
b-
U;
L
o
o_
E
O)
o
c
q)
U)
le
la)
IJ
lL
to
l=
lc)
ls6
l(u (/)
l=e
t(u
IL
lFg
o o>l
-l
E€I
'= ol
ool
O .E-l
I([
Itr
la
lo
l6
la
I
I
lo
IF
lc
to
Io)
l(d
o
E
(5
L
o
q)
L
I s,,
l6
,lo-
'lo
lq)
lr
I
lo-
to
tco
Io 6
t-
| ^a
lg b
hP
-,rt €
691
>61
lu;
lL
C)
E
(u
U;
p
(u
E
lu;
lL
lq)
lo_
lo
lo
-Y
o
a
f
o
s
lPs
lc a)
to
to o
IBE
l* O
tc
li €r
l6 sl
tF >l
lo Ll
te il
ool
c t-l
,d 3l
bEl
o >'1.
I DI.
lui
lL
lq)
lc)
l=
lo
l'-
ui
L
q)
c
E
o-
lui
lL
lo
IP
lc
lo
o-
L
o
o
U;
L
s
(5
',1 ui
lL
lG)
lc
'lc r:
ls6
t^
lp E
l6 _g
Irg
iP
OF
--
=E
IU;
lL
la)
{
o
=
E
L
(U
a
E
o
o
(u
ui
L
o
(u
=
O
E
c
(u
(d
.c)
C)
o
F
Eo
-L
.0e
(DE
oo
,p=
9o
o-- g
(d
r e-l
E
c
(u
o
o
q)
x
oo
L
niO
x{
Eo
S=
aa
'F Ol
coil
9(g1
<cl
GI
oi El
to
lL
lo
IY
lL
to
l=
IE
lo
Its
ICU
t6
l-
loO
E
o
L
c)
O
-
to
IL
lo
IJ
lL
lo
r=
a
o
(U
a
cri
IA
lL
lc)
IJ
lL
lo
l=
o.
o
L
o
U)
c'j
lq)
lar- g
lF o
l=Y
IJ L
lco
l=
lcEr
IF Ei
lr- Ol
lo Ll
l(- Ll
I.r2- P I
t€l
lol
,t dl
p ,El
FO'
E8
o
L
o
=
O
E
c
(s
o_
e
-O)
.. o _
rc(u
= o'=
h'd o
Xcno)
co(g
C)Y(u
g)o- b
o
E
o
()
L
(u
=g
t-h J
o.9
E6
lCl
c
o
E
O)
o
U)
o
E
o
=
o
U)
t
(d
p!
-J
O=
cc)
o)'=
oo,
a<
los
tr
t<-
l*.
lo
IO
t
l-rl
tl
tol
IOI
al
8l
ol
ol
sl
a)
o
r
su
(t)
.
ei
-ll
ol
FI
EI
(l)
t
s
i
el
al
NI
a
q
a
Ihr
54. APPENDIX III
DATA FLOW DIAGRAM
Fr
G]
a
Fr
Ft
.Qr
ehr
fir
a
at
F
*.
h
hl
ft
h
h
h
h
i
(
b
h
of
.r
t
ft
{r
i
t
VALIDATION SAMPLE
(50%)
(r4,103)
LIC
CUSTOMERS
(10,000)
NSSO HOUSEHOLD SURVEY
( 18,205)
RESPONDERS=LlC CUSTOMERS= 1
(20.0%)
(29,205)
DEVELOPMENT SAMPLE
(50Vo)
(14,102)
55. APPENDIX IV
C{r
Iar
Fr
t- Propo_No plan_code age_code income-code occupation- sex
a --=
^r 333?Z
car 31313
c= 3?t?7,
caa 3ll3g
01443
h3l3??
{3l3li
rrD 3l3l3
h3l3l;
ft31333
dr 3:?'^:^
dr
t'
qb
Jr
cr
013
713
711
009
110
610
All
214
110
-71 A
t t-
313
713
212
113
913
914
811
112
712
212
810
213
512
511
101
801
trno
512
212
812
709
812
411
909
807
501
101
zl
20
22
41
52
zz
92
92
20
20
21
20
10
zz
ZI
zl
zz
20
zz
+z
41
32
82
82
82
12
82
IJ
11
52
41
+J
001
518
001
501
000
901
512
300
750
513
520
520
OJU
524
020
505
009
305
820
250
912
000
820
001
507
ane
501
oo2
003
002
002
002
003
507
UUO
502
501
644
033
022
022
522
022
033
133
022
033
033
000
022
022
000
ae?
000
000
000
033
522
000
622
233
088
577
544
U.J.'
444
' 544
n?e
899
077
533
522
FSTSECND
M
F
M
M
M
M
F
F
M
F
F
F
M
M
M
M
F
F
M
M
F
M
F
M
M
F
M
F
F
F
F
F
M
F
M
M
M
57. g
o
c
o_
X
o
I
.E p ;- g 9g @ s 1o !q @ a? ..4l o ro r oq a s9 g? e9 o F* @ ro @ N @ .r,
o_@ cD <o
= S i A o { N o do g)|.A @ (99o o co $ N (o @ cy)
(d r st r r ro trl cD r..- crt @ $ o o + [ g i 6 tr d5 6 + + d d nj
(_) s r co @ $ @ rJ) $t !F $t f._ cD ; cD bo bo + o d _ N i s F_ r o o
: c! (Y) ro cl N cr rr) 9? a !g $J s cu N i =
qt qO i qJ i_- N cv r N N r
-uoo o oo ooe aO 9; ddaaa q>q qo do oooo
rLoooooooo6o6odd booo ooooooooo
ea
3i
a
A
a.
fi
+
qr
+
*
4r
.t
i
+t
+
s
+t
I
T
t
+
+
t
t
I
I
?
t
I
O
g) o o o) o) o) o) o) o) o) o) c! o o) o) o) o) o) o) _ o) o) o o o) o) o) N
TJ
o
a
U)
o
a
o_
I
!
o cD N $ cl cu cl ql N = $r r t N c2 aq qO N N S (o c @ c)C! c!
Jo c! o o o o o o o o cj o d d o 6 o d d d d d o o o oo
L
I
o-
-t
E
Y
o
a
=
f
- - - $ r
- O) r O) r N O) st
- - N Al - N r C! $ N S $ N C)
o
a
3s op
=
o
=
tr r- 9 9 I I g o cD,: Q Q Q o cD o o o o oo o
J.@ $ o r O co N g? E O O 'b = + q !! a O O + d6 F c,: r _ ce r_
..rr.- s $ @ + @ co g] ca qt g al co cir V Sg q rO @ ct (o (o (o (o (o (o o)
(J(o r u: o o cD
- N - O O; - cd e e e O 6 N d E o o o o o
= N |f)
'r)
o ro N o S)
= St O o 5 N 4 6 e 6 6 N d d o o o o cr
z. N (o (o o (o o N 1.._ co o o o, 6 cr.l dc o o @ o N o o o o o o cr)
o
.N
a
-
b
-
o
U)
I q S N 9 o rrl $ q) N 9 C! r o $ c'r re 1. q? 5r <o @ t_ u: @ o o ro
roOOOr o oocjoOoid ; adod Edb oorre
o
I
6 s s s s s.{- rrl q) s s s i- $ s t s s s s $ $ $ s $ * sf r
lo) o) o) o) o) o) o) Q Q q) o o, d cn q, a Q q) o) o o) o) or or o) o) o)
o oo o o o> o orRR R pFA A e e d d d 6 66o oooo
ii5 5 .l: f I 9 o
-
v v vuu u (J v r r r F r r '_ r r rr
x r-- r- co co N N o 9q gC tr go N F tr 6 e f. <o o br b o o o o) s lr)
rJ N N N cI c! c N cJ ot cv crt N i t dJ cv c.t N N N r r N N r N N