PRODUCT DIVERSIFICATION AND FIRM PERFORMANCE IN POLISH NON-LIFE INSURANCE SECTOR
DOI:10.30825/5.EJPAU.177.2019.22.3

Kazimierz Ortyński
Faculty of Economic and Legal Sciences, Kazimierz Pulaski University of Technology and Humanities in Radom, Poland

ABSTRACT

The article discusses the key characteristics of the non-life insurance market and the impact of selected product strategies on the financial results of the group of 7 leading insurance companies in Poland in 2012–2017. The main objective of the study is to consider the choice of an insurance strategy from the point of view of the financial results of insurance companies.  The article attempts to verify empirically which strategy, i.e. diversification or product specialization, was more effective for the group of 7 leading insurers on the Polish non-life insurance market in 2012–2017.  The study showed that the product diversification strategy implemented in the group of 7 leading insurers was statistically positively correlated with the insurers' profitability rates in the non-life insurance sector.  The product specialization in this group was negatively significantly correlated with the profitability rates. In addition, the study showed that the variables representing firm size and investment income positively and statistically significantly affect the performance of non-life insurers in Poland.

Key words: insurance, diversification, performance, insurance companies, specialization JEL Classifications: G22, L25, O50.

INTRODUCTION

Modern Polish insurance market, including the non-life insurance market, is relatively young. The formation of modern insurance market in the free market economy began in 1991, after the adoption of the Insurance Activity Act in 1990 [35]. This Act enabled to define the organisational forms and operating principles for insurers as well as to create the essential institutional structure of the market. Legal grounds for demonopolization of the insurance market were also created and a compulsory insurance system based on the insurance contracts was also introduced [34]. The regulations introduced in the following years eliminated the barriers to the Polish market for foreign investors in 1999 [36], then made it possible to incorporate the Polish market into the single European insurance market [37–40].  The latter regulations adapted the Polish insurance law to the EU Council Directives. In 2004 Poland became a member of the EU. In the first phase of Polish insurance market development after 1991, the process of its demonopolization and privatization took place, which resulted in the emergence of new private insurance companies. Since 1999, especially after Poland's accession to the EU, in the light of rapid development forecasts for the insurance market and profitability, the growth of direct foreign investments in this market has been taking place. Direct foreign investments on Polish insurance market caused the emergence of new companies, recapitalization of existing insurance companies, greater share of foreign capital in the core capital of insurance companies. These investments also contributed to an increase of importance of certain segments of the insurance market and an improvement of the overall insurance sector activity. As a consequence of these changes, there was a significant improvement in financial results of the sector, as well as profitability of insurance and investment activities and an increase in the development dynamics of the insurance market. This resulted in a qualitative change of the market, including an improvement of quality standards and availability of insurance services. In addition, foreign owners of domestic insurance companies made a significant contribution to the modernisation of the insurance market and its adaptation to international standards. The new circumstances fostered greater market competitiveness and faster development of insurance services. Since 2004, a process of consolidation of insurance companies has started in Poland (mergers and acquisitions), mostly through international insurance capital groups operating on the Polish market. An example is the merger of two insurance companies, i.e. Zurich Towarzystwo Ubezpieczeń SA and Generali Towarzystwo Ubezpieczeń SA in January 2004. Another characteristic feature of Polish non-life insurance market is excessive amount of compulsory insurance. It is estimated at about 160 insurance products [15]. This phenomenon is the subject of particular criticism in Polish literature. It has been rightly pointed out that the inflation of the insurance obligation leads to atrophy of insurance awareness. This criticism is particularly important considering insurance awareness in Poland has been at a very low level for years. The subject of this research are the key characteristics of the non-life insurance market and the impact of selected product strategies on financial results of the group of 7 leading insurance companies between 2012–2017. However, the main objective of the study is to consider the choice of an insurance strategy from the point of view of the insurers’ financial results. The article attempts to verify empirically which strategy, i.e. diversification or product specialization, was more effective for a group of 7 leading insurers on Polish non-life insurance market in 2012–2017. The insurance market research used indicator methods, that are commonly used in insurance analyses. Whereas, for the verification of the research hypothesis, the methods of panel econometric modelling were used. The remainder of the article is organized as follows. The second part of the article describes the state and changes in Polish non-life insurance sector in comparison to selected EU countries between 2012–2017.  The third part analyses strategy choices by insurers in the light of the empirical research results presented in selected literature. The fourth part covers theoretical assumptions of the empirical research. The following part presents the statistical data and panel econometric models used. The results of the empirical research and conclusions are presented in the final part of the article.

KEY CHARACTERISTICS OF NON-LIFE INSURANCE SECTOR PERFORMANCE IN POLAND

This part of the study focuses on the following issues: the level of development of the non-life insurance market, the degree of market concentration, the level of foreign investments, insurance distribution channels, the product structure of the market and the technical and financial results of insurers.  The following measures of the insurance market development were used in the study: gross written premium in absolute and relative terms (penetration rate, density rate and the relation of the insurers' investment portfolio to GDP). The level of market concentration was measured by the share of the three and five largest insurers (CR3 and CR5) in the gross written premiums. The ownership structure of the market was determined by the foreign capital share in the share capital of domestic insurers and the share of insurers with foreign capital in the gross written premium. The structure of insurance distribution channels was determined according to their share in gross written premium, and the product structure was also determined according to the share of the groups of products identified in the gross written premium. The research was based on statistical data published by the following institutions: OECD, Insurance Europe, the Polish Financial Supervision Authority and the Polish Insurance Association. In 2012–2017 in Poland, the non-life insurance market GWP growth rate increased by over 20%, while at the same time in other analysed countries it declined or the increase was insignificant (with the exception of France and Hungary) [22].

Such an increase in the volumes of gross written premium, i.e. an annual average of over 3.5%, should be evaluated positively. However, a significant increase in gross written premium did not take place until 2015–2017, mostly due to an increase in car insurance premiums (an increase in premium rates and an increase in the number of motor vehicles). The results of the comparative analysis of the penetration rate, density rate and share of the investment portfolio in GDP in 2017 indicate a significant gap in the development of non-life insurance markets in Western Europe as compared to Poland and other Central European countries [22].

The concentration level of Polish non-life insurance sector in the analyzed period was relatively high (CR5 concentration ratio at 66–69%, i.e. the share of the 5 largest insurers in the gross written premium of the non-life insurance market). In 2017 the CR 5 ratio reached the level of almost 70% [27]. This concentration level of the sector can be explained by mergers of insurance companies (e.g. in December 2012 HDI Asekuracja SA merged with TUiR Warta SA and in October 2014 MTU Moje TU SA merged with STU Ergo Hestia SA) and a more rapid organic growth of the largest insurance companies [26]. 

The effects of high concentration are not favourable for either the insurance sector or the insurance customers. High market concentration hinders fair competition and makes market stability dependent on the financial situation of dominant insurance companies [34]. Moreover, for insurance customers, high market concentration may result in unjustified price increases for insurance services. The increase of the number of insurance companies in 2016 resulted from the emergence of new domestic insurers in the form of public companies or mutual insurance companies with domestic capital. Foreign companies gradually decreased their share in the share capital of non-life insurance companies from 90% in 2012 to almost 68% in 2017. This was mostly related to the increase of domestic capital share in insurance companies.

During the period under consideration, the distribution structure of non-life insurance products did not change significantly.  

A characteristic feature of the distribution structure of insurance products was the dominant position of private insurance agents (market share at 61–62%) and important position of insurance brokers (market share at 18–19%). The next position was held by direct sales by organizational units of insurance companies [11].

The dynamics of changes in the structure of the insurance portfolio on the non-life insurance market in the audited period was insignificant. Motor insurance products maintained their dominant position at the level exceeding 50%, increasing to almost 60% in 2017. The high share of motor insurance resulted from the historical structure of insurance that was difficult to break through and high dynamics of the automotive industry in Poland. Property insurance remained second, but with a downward trend from 19.4% in 2012 to 16.4% in 2017. Other groups of insurance products that were analyzed did not demonstrate any clear development trends, maintaining their market share at 5–7% [27].

Technical results on insurance activity and revenues from financial investments were significant factors impacting the financial results of insurers [33]. It should be highlighted that the dynamics of non-life insurance development in Poland between 2012–2017 was also strongly influenced by the relatively high GDP growth with an average annual rate of about 3.3% [33]. A characteristic feature of the non-life insurance market in Poland in the audited period was a significant growth rate of insurance (the average annual gross written premium rate was over 3.5%). However, the development rate of this market is far from similar markets in Western Europe. The degree of market concentration is high with a large share of foreign capital in the share capital of domestic insurers. The dominant distribution channel are insurance intermediaries, in particular insurance agents. On the other hand, the product structure of the market is continuously dominated by motor insurance products.

LITERATURE REVIEW

Diversification is widely covered in literature, particularly in the field of management sciences, and the scientific achievements are significant but diverse. Ansoff [1] presented the basic issue of diversification from the theoretical point of view, designing the development of a company in the product-market relationship [13]. Ansoff distinguishes and defines possible company strategies, including product development and diversification strategies. The author emphasizes that the diversification strategy is the next stage of company’s development and requires a change in the technology of manufactured products and a departure from the current market structure. The issue of diversification was also addressed by Penros [25] and Chandler [3], among others. Penros discusses the issue of diversification on the basis of the theory of the company's development. The author points out that a company applies different strategies depending on the stage of its development. She sees the diversification strategy as the next stage of development after the stage of internal expansion. Chandler characterises the relationship between strategies and corporate structures and in this context distinguishes the diversification strategy from other strategies. Among empirical studies, one should point to the results obtained by Rumelt [30]. The author showed, among other things, that the companies adopting a diversification strategy in their development achieve better results than companies with a non-diversified strategy. Considerations about the impact of product diversification strategy on results of companies are widely discussed in the financial literature [31]. Each business strategy requires a cost-benefit analysis. The theory of diversification strategy indicates that companies operating in many segments gain through the economies of scale by spreading the costs across similar business lines and they can charge customers higher prices for product packages [14]. Another important reason for diversification is the need to spread the risk. The geographical diversification of property insurance companies enables insurers to reduce the company's underwriting risk through cross-regional risk distribution [4]. When concentrating on one product, there is a risk of the demand shift, emergence of a more active competitor, etc. The multi-product offer reduces this risk [7]. On the other hand, the main costs of diversification include agency costs and inefficiency of the internal capital market [42]. The existence of agency problems may further affect the investment decisions of companies and the designation and selection of relevant diversification strategies. The increase in agency costs may result from ineffective control mechanism over decisions made by managers of particular product lines (branch lines), and inefficiency of the internal capital market – from under-utilization of capital in face of new development opportunities. On the other hand, the theory of product specialization strategy indicates that companies focusing on specific products can maximize their value better, lowering agency costs and becoming more efficient. In the insurance literature, similar studies have examined non-life insurance industries [5, 9, 19]. Examining the effectiveness of type X for 321 insurance companies between 1990–1995, Meador et al. [21] proved that the strategy of product diversification impacted the increase of efficiency more than the strategy of specialization. Elango et al. [6] investigated the relationship between the product diversification strategy and insurers’ financial performance for more than 5600 property-casualty insurance companies in the US for the years 1994–2002. The authors showed a positive non-linear relationship between product diversification and a company's risk-adjusted ROA. The results of their research indicate the U-shape relationship between product diversification and a company’s risk adjusted ROA. Liebenberg and Sommer [19] researched the effects of corporate diversification using a sample of property-liability (P/L) insurers over the period 1995 to 2002. They found out that the undiversified insurers outperformed the diversified ones. Their findings provide strong support for the strategic focus hypothesis. Cummins, Weis and Zi [5] analysed the results of insurance companies (life, health and non-life insurance) in the US in the years 1993–1997, with particular emphasis on the positive effects of economies of scale, testing the hypothesis of diversification and product specialization. According to the authors, the results obtained support the views expressed in the literature, that the strategy of specialization is better than the strategy of product diversification. Fuentes et al. [8] showed, based on a technical analysis of effectiveness and efficiency of insurance companies in 1987–1997 in Spain, that companies combining two or three product lines achieve better results than companies operating with one product line. Pavič and Pervan [24] determined the relationship between product diversification and company performance in the non-life insurance sector in 2004–2007 in Croatia. The results of model analyses showed a negative and statistically significant effect of diversification as well as a positive and statistically significant correlation of the concentration index on the results of insurance companies. Berry-Stoelzle and Song [2] demonstrated the relationship between product diversification and the performance of insurance companies in the non-life insurance sector in the US in the years 1995–2004. The authors emphasize that the results of their analyses do not show any significant impact of product diversification on the results of the investigated companies. Krivokapic et al. [17] analysed the relationship between diversification and the results of insurance companies in the non-life insurance sector in Serbia between 2004–2014. The research results showed that these relations were significant and positive, meaning that insurers with a more diversified product structure achieved higher financial results than insurers using the product specialization strategy. In empirical insurance literature, there is no clear confirmation of which strategy is more beneficial for insurance companies. There is no empirical conclusive evidence of the impact of product diversification on the results of insurance companies.

HYPOTHESIS AND VARIABLE DEFINITIONS

Following the previous studies, the object of this article is to test which strategy (e.g. the product diversification or product specialization) is better for insurance companies in the non-life insurance sector from the financial results standpoint. With reference to the research results presented above, this article subjected to verification the hypothesis that the product diversification strategy is more effective for insurance companies in the non-life insurance sector from the financial results standpoint.  The measure of entropy, weighted index of product concentration, firm size and profitability ratio of investment activity were taken as independent variables. The performance of insurance companies does not only depend on diversification or concentration ratio but also on other firm characteristics. They were selected a posteriori from the set of key firm characteristics by eliminating the variable with the lowest t-Student statistic (alternatively with the highest p-value), i.e. by sequential elimination of irrelevant variables. The study has taken the risk-adjusted return on equity and risk-adjusted return on assets as dependent variables (explanatory) in order to avoid the impact of random effects on the annual return usually resulting from underwriting risk-taking and investing temporarily free funds [9]. Risk-adjusted return on equity and on asset is meant to compensate for this impact. The explanatory variables encompass diversification, concentration as well as other firm specific characteristics. The product diversification, measured by the entropy measure which was first introduced by Jacquemin and Berry [12] to incorporate different business segment information and to quantify unrelated product diversification.

The ratio in this analysis is determined as follows [17]

,

where  is the quotient of the gross written premium of the i-th insurer from j-th group of insurance (class of insurance) in the year t versus the sum of gross premiums written of the i-th insurer from the 18 group of insurance in the year t.   values were determined for each of the seven insurance companies in each of the six years (i=1,…, N; j=1,…,M; t=1,…,T; in the study N=7, M=18 and T=6). In other words, the entropy coefficient was calculated as the sum of 18 products of  values by the natural logarithm of the inverse  for each of the 7 insurance companies and for each of the 6 years. In the study the types and number of groups of insurance (class of insurance) is the same as in the annex to the "Act on Insurance and Reinsurance Activities of 2015" in section II [41].

The Herfindahl-Hirschman Index [8] is often used in the role of the concentration ratio, which is determined to the formula

,

where:   is the gross written premium of the i-th insurer from j-th group of insurance (class of insurance) in the year t, and   is the sum of the gross written premiums of insurance companies from j-th group of insurance (class of insurance) in year t; (i=1,…, N; j=1,…,M; t=1,…,T; in the study N=7, M=18 and T=6).

However, the study used a weighted ratio , i.e. according to the weight of the gross written premium of the insurance company "i" from the group of insurance "j" in the total sum of gross premiums written for the group of 7 insurance companies in the year "t".

The weights were determined according to the following formula [17]

 where:   is the gross written premium of the insurance company "i" from the group of insurance "j" in the year t, and   is the total sum of gross premiums written of the group of 7 insurance companies (from 18 classes of insurance) in the year "t"; (i=1,…, N; j=1,…,M; t=1,…,T; in the study N=7, M=18 and T=6).

Then, using the weights, a weighted concentration index was determined for each insurance company in each of the years considered, i.e.

.

The firm size (ln(GWP)) variable was defined as a natural logarithm of gross written premium, and Profinv variable was defined as investment income / the value of balance-sheet investments.

The list of characterised variables is included in Table 1.

	Variable	Definition
ROE	Profitability ratio of equity	net financial result / equity
ROA	Profitability ratio of assets	net financial result / assets
SDROE	adjusted for the risk of profitability ratio of equity	ROE / the standard deviation of observed returns on equity over the previous six years
SDROA	adjusted for the risk of profitability ratio of assets	ROA / the standard deviation of observed returns on assets over the previous six years
Entropy	product diversification indicator	=
WConc	weighted Herfindahl-Hirschman index (weighted index of product concentration)
ln(GWP)	Firm size- ln(gross written premium)	natural logarithm of (gross written premium)
PROFINV	Profitability ratio of investment activity	investment income / value of balance-sheet investments
Source: own calculations

DATA AND METHODOLOGY

The research covers data for 7 insurance companies, operating in Poland between 2012–2017. The following companies (joint stock companies) from the non-life insurance sector were selected for the study: TUiR Allianz Polska SA, Compensa TU SA, STU Ergo Hestia SA, Interrisk TU SA, PZU SA, Uniqa TU SA, TUiR Warta SA, whose turnover represented jointly about 75% of gross written premium in this sector. The study used statistical data of the Polish Chamber of Insurance (PIU) – published (“PIU Annual Report”) and unpublished data [27].

The research was performed using Gretl software and Microsoft Excel. Descriptive statistics for each variable depicting the firms’ profitability is presented in Table 2.

The study used econometric modelling for panel data.

The study was based on the following analytical form of the model

where:

 – is a vector of explanatory variables, i.e. the profitability measures of the i-th insurer (i=1,…,N), in the year t (t=1,…T); (N=7 and T=6),

 –  is a free expression (constant size),

 – is the vector k of the structural parameters of the model (k = 1, ..., K),

 – is a matrix of explanatory variables (characteristics of insurers' activities), of the i-th insurer (i = 1, ..., N) in the j-th insurance group (j = 1, ..., M), in the year t (t = 1, ..., T);

 – is a scalar disturbance term, i indexes the company in a cross-section, and t indexes time measured in years;  consists of the random effects part and the fixed effects part relating to a specific i-unit of the panel, .

ESTIMATION AND FINDING RESULTS

The estimation of econometric models was performed by means of the random effects generalized least squares method (GLS) using Nerlove’s transformation. The validity of the choice of this estimation method for econometric models (with random effects instead of fixed effects) was confirmed by Hausman's test result.

Results showing the effect of product diversification strategy on SD ROE with the values of statistical tests verifying the fulfilment of the assumptions of the adopted model estimation method have been presented in Table 3.

The Breusch-Pagan test confirmed the validity of the null hypothesis: the variance of error in the unit = 0, which leads to the conclusion that fixed effects do not occur, i.e. the GLS estimation method was reasonably assumed. The Hausman test has shown that there are no grounds to reject the null hypothesis (random effects are not correlated with the explanatory variables), that is, the GLS estimators obtained are consistent and effective. In the absence of grounds for rejection of the null hypothesis: the random component has a normal distribution, which means that the GLS estimation used is correct.  It should be emphasized that the use of Nerlove's transformation made it possible to obtain better estimators in random effects generalized least squares method (in terms of mean squared error) than estimators obtained using other methods [20]. Positive results of the above tests indicate the correctness of the estimated model. The estimated coefficients have correct signs and the explanatory variables are relevant. The statistical data used for the model are appropriate and the model meets the assumptions of the panel linear regression model. The parameter estimator for the variable representing the entropy coefficient (with the value of nearly 3.37) indicates that the change of the variable of the entropy coefficient by one unit causes the SD ROE variable to increase by the value of the estimator, under ceteris paribus condition. Whereas, the estimator for the variable representing the weighted concentration ratio (with the value equal to -78.92) emphasizes the significant and opposite direction of changes of the SD ROE variable for the change of the weighted concentration ratio by one unit.  The values of the estimators for the following variables, i.e. ln(GWP) (representing the insurer's size) and PROFINV (representing financial investment income) are positive and amount to 4.04 and 27.9, respectively. These estimators indicate that the increase of these variables is associated with the increase of SD ROE.

Table 4 presents the results of the model estimation on SD ROA along with the values of statistical tests verifying if the assumptions of the adopted GLS estimation method were met. The test values indicate that, similarly to Table 3, the use of the random effects GLS estimation method with Nerlove's transformation was justified, thanks to which the estimators obtained are compatible and the most effective and the basic assumptions of the econometric model were met. The parameter estimator for the variable representing the entropy coefficient (with the value of nearly 2.0) indicates that the increase of the entropy coefficient by one unit is associated with the increase of SD ROA by 2 units (under ceteris paribus condition). Whereas, the estimator for the variable defining the degree of concentration, with the value equal to -43.0, emphasizes the strong opposite direction of changes for the SD ROA variable (with the increase of this explanatory variable by one unit). The estimators for the other variables, i.e. ln(GWP) and PROFINV, are positive and indicate the same direction of change with the explanatory variable. The increase of the variable representing income from financial investments is associated with a significant increase in SD ROA.

CONCLUSIONS

This study revisits the diversification-performance (D-P) relationship in the non-life insurance industry in Poland, and contributes to the literature by attempting to solve this issue. The study showed that a product diversification strategy implemented in the group of 7 leading insurers was statistically positively correlated with the insurers' profitability rates in the non-life insurance sector. The product specialization of this group of insurers was negatively correlated with these rates. In addition, the study showed that the variables representing firm size and investment income positively and statistically significantly affect the performance of non-life insurers in Poland. The conclusions from the conducted research are particularly important for management boards of insurance companies in the non-life insurance sector, because they set business strategies, and the right product diversification plans can put them in a better market position. The study also provides some valuable insights into the contemporary Polish non-life insurance market. Polish non-life insurance market has great development potential due to, inter alia, considerable profitability of domestic insurers, a relatively large share of foreign investments, high economic growth of the country, as well as relatively low insurance penetration and density rates.  The market is characterised by a high degree of concentration and the dominant position of insurance agents in the structure of insurance distribution channels. When it comes to the product structure, motor insurance products have held the leading position for years.

Received: 20.04.2019
Reviewed: 17.07.2019
Accepted: 15.08.2019

Responses to this article, comments are invited and should be submitted within three months of the publication of the article. If accepted for publication, they will be published in the chapter headed 'Discussions' and hyperlinked to the article.