ICT Penetration and Insurance Sector Development: Evidence from the 10 New EU Member States

Abstract

The insurance sector provides protection to individuals and businesses against many types of risks and also promotes economic growth, being an important source of long-term capital. Analyzing factors that facilitate insurance sector development is important for both individuals and the entire economy. The purpose of this study is to investigate the relationship between information and communication technologies (ICT) represented by mobile cellular subscriptions per 100 people and individuals using the Internet (% of population) and insurance sector development represented by insurance company assets to GDP (%). Using data from 10 new member states of the European Union for the period 2000–2020, this study reveals a mutual interaction between ICT penetration indicators and insurance sector development. Furthermore, a regression analysis reveals that Internet penetration has a significant positive influence on insurance sector growth. Specifically, at the country level, the results indicate the existence of bidirectional causality between mobile cellular subscriptions and the insurance sector in Latvia, Poland, and Slovakia, and unidirectional causality between insurance and mobile cellular subscriptions in Estonia and Hungary.

1. Introduction

In a world full of uncertainties, even when people take measures to protect themselves from various risks, different types of loss (e.g., assets, life, health, business income) can occur. It is the role of insurance to reduce the financial impact of these risks and to compensate these types of losses by providing financial support to the people affected. However, insurance’s role is more complex because insurance companies are consider large investors, providing long-term finance for governments, banks, and other institutions. In addition, insurance stimulates the population to adopt responsible financial behavior and to increase financial literacy levels. Through these actions, the insurance sector, together with the other components of the financial markets, can exert a significant contribution to economic development. In modern economies, information and communication technologies (ICT) have been integrated into most activities and have generated significant changes in every domain, including financial markets. Consequently, the researchers started to explore the relationship between ICT and financial development [1,2,3,4,5,6,7,8,9]. Several studies argue that ICT improves financial inclusion by allowing access to different financial services to the poor or to marginalized people [1,9,10,11,12]. On the other hand, there are studies suggesting that financial inclusion generates improvements in ICT infrastructure [3,12] as a result of a higher demand for digital financial services generated by the initial access of the poor or marginalized people to such services that facilitated the improvement of the economic status of these people and the acquisition of more technology.

Previous studies that analysed the relationship between ICT and financial development have focused mainly on banking or financial market variables, without taking the insurance sector into account, despite the fact that insurance is a significant financial service in developed economies. One of the reasons for this situation is that, as Cappielo [13] argues, traditional insurance was resilient and adopted digitization very slowly, but once adopted this generated a profound transformation of the entire business [14].

Studies that investigate the determinants of insurance development generally found that macroeconomic (GDP growth, inflation, the unemployment rate, balance of payments as a percentage of GDP, trade openness, foreign direct investments, financial development), institutional (property rights, corruption control, government stability, law enforcement), demographic or social (educational attainment, tertiary education level, life expectancy, youth dependency ratio, urbanization) variables have various effects on insurance development [15,16,17,18,19,20].

Interaction between ICT and Insurance

The adoption of information and communication technologies by participants in the insurance market has a significant impact on the entire insurance value chain [21,22,23]. ICT generates changes in product design, offering the possibility of providing personalized instead of standard products and services and achieving a better risk assessment and, consequently, a more adapted insurance premium [14,21]. Additionally, ICT generates benefits for insurer marketing activities; the data obtained through ICT could be used for market and customer research and could generate better customer segmentation and a detailed overview of the customer needs and preferences [24]. Based on this information, insurers could develop a more suitable pricing strategy, design an appropriate communication and advertisement strategy, or could use the best suitable technologies such as websites, social media, and videos [25]. According to Cappielo [22], the most affected component of insurance by ICT development are sales and distribution. Online sales by using websites, apps or different social media platforms in the acquisition process are the significant implication of ICT in insurance distribution [14,25]. The use of ICT allows a vast amount of information to be shared with consumers that can provide a competitive advantage to insurance companies. Many companies operating in the insurance sector introduced chatbots or robo-advisors in their interaction with customers in order to collect data about clients’ needs and used them to customize their products and services, and to provide advice for choosing the best insurance policies that suit customers’ needs and financial situation. Video calls or phone conversations are used more and more by insurers to analyze emotions based on image and video processing and to generate new models to be used in product design [21]. The claim management process is also affected by the use of ICT. Filing an insurance claim using digital technologies improves not only company performance but also customer experience. By using Big Data and predictive analytics [14], insurers can improve the claim process and prevent insurance fraud.

By adopting digitalization, insurers can increase speed and efficiency by automating processes and decisions, generating simpler, more efficient and faster workflows. They can reduce operating costs and by issuing and delivering insurance policies in digital format, by automating the claim process, and by using digital archives and reducing the cost of storage, they can become more sustainable [26] and inclusive. Digitalization affects not only the core processes, but also the support services such as management, human resources, IT, and legal departments [25], inducing the transformation of insurers.

The crisis caused by COVID-19 forced insurers to implement digital transformation. Based on a study that involved the top 30 global insurers, Lanfranchi and Grassi [27] noted that during the COVID-19 pandemic, these companies adopted specific initiatives. There were cases of product adaptation that require no ICT involvement (inclusion of COVID-19 into the insurance coverage), but numerous initiatives were related to ICT use (improving existing technologies or adopting new technologies). Innovative initiatives such as realizing visual inspection using AI to deliver useful information for insurance underwriting, evaluating property or vehicle damages with AI, improving customer service communication by using call centers, using chatbots and specific applications for COVID-19 symptom checkers, using mobile apps for health care assessment, and providing access at no cost to telemedicine for all insured people have been implemented in the insurance sector recently.

Shevchuc et al. [28] investigated the digital transformations due to the coronavirus pandemic in the insurance sector of Ukraine and noticed that the most important processes became digital, here referring to the adoption of the electronic document management systems and significant transformations in distribution which consist in a dynamic shift from traditional to digital distribution, etc.

As Eling and Lehman [25] and Cappielo [22] argue, the digitalization generated by improving the ICT infrastructure and increasing the use of this infrastructure significantly impact the evolution of insurance activity in different ways: by providing innovative goods or services or new ways to deliver them, by generating lower costs or higher levels of efficiency and productivity, and also by helping insurers and their contracts to be more accessible to different categories of customers.

In a world dominated by technological innovations, customers, especially individual ones, press the insurers to provide simpler and more flexible products and buying opportunities and to develop new channels of communication [24]. With all of the changes mentioned above, insurers moved to a more customer-centric business in a tentative way in order to meet client needs adequately.

The customers’ answer to the insurers’ innovations depends on various factors. Different studies argue that ICT use reduces acquisition costs and insurance premiums [29], which will impact insurance customers’ satisfaction [30], which is the main determinant of purchasing intention [31]. Eckert et al. [30] indicated that customer-focused digital strategies can improve customer satisfaction, and Dash and Chakraborty [32] suggested that search engine optimization and search engine marketing practices display marketing and electronic customer relationship management practices that have a major influence on consumer satisfaction and purchase intention.

However, on the other side, Mau et al. [33] emphasized that there are some characteristics of the customer that influence research-shopper behavior, referring to the fact that there are still many customers that use a channel to search for a product or a service and another one for purchasing. Bryzgalov and Tsyganov [34] noted that digital sales are exclusively related to insurance products regulated by the state.

Time and financial savings create unforgettable and meaningful experiences for customers [35], which can contribute to consumer satisfaction and can lead to customer loyalty. New technologies erase boundaries and improve customer relations, therefore insurance companies along with other financial institutions must understand that the transition from a hot trend to a must-have technology needs to be done quickly.

As can be seen, the use of ICT infrastructure could generate benefits for insurance activity, both tangible (adequate information storage facilities, timeliness of insurance operations through rapid communication, the reduced effort required for individual tasks) and intangible (customer satisfaction and a strong corporate image) [36,37].

The insurance industry, like other industries, has experienced a major transformation due to the development of ICT. However, there is not only a unidirectional interaction, as the digitalized insurance industry can also affect ICT development, as argued by Pradhan et al. [3] and Pradhan and Sahoo [12]. Therefore, a bidirectional interaction between ICT indicators and insurance sector development is theoretically expected.

Our paper examines the relationship between the development of ICT infrastructure and the insurance market in the 10 new EU member states. We focus mainly on this specific group of countries that have certain characteristics in common. Due to European integration, selected countries changed their “political, market and economic conditions” [38] (p. 74), and now share common visions and goals. To the best of our knowledge, for this group of countries, this is the first paper that attempts to analyse the link between ICT penetration indicators and the growth of the insurance sector.

2. Literature Review

In one of the first papers dealing with the implications of the Internet on insurance markets and institutions, Garven [39] identified some trends and highlighted that one of the main effects of Internet use in the insurance area will be lower insurance premiums. This effect is a consequence of reducing administrative costs and of the higher competition between insurers that is a result of the amount of information available to consumers. Another important aspect of using the Internet in the insurance sector involves the capacity of insurers to offer flexible coverage options by offering their clients products adapted to their unique situations. Based on these transformations, marketing strategies must evolve, and insurers need to rethink their business strategies and adopt a customer-centric approach.

The effect of ICT on the level of health insurance premiums paid by consumers was investigated by Pauly et al. [40] in an attempt to see which category would benefit most from the use of ICT. As a consequence of using the Internet, a decrease in search costs was noted, but the results of Pauly et al. [40] show that by using new technologies, not all categories of customers will pay lower premiums for private health insurance, only the younger ones.

More empirical papers have been developed in the last period that endeavored to investigate the ICT–insurance nexus. These could be divided into four categories, keeping their approach in mind.

The first category of studies includes papers that try to investigate the impact of ICT penetration on people’s participation on the insurance market. These studies start from the hypothesis that information is important in determining higher participation in the financial markets [41,42,43]. Considering that the Internet is an important information channel for people in their attempt to minimize premium costs and obtain the best coverage through insurance, Liu et al. [44], Chen et al. [45], Hu et al. [46] and Lin et al. [47] argue that insurance market participation is positively influenced by the ICT infrastructure and its use, but differs for various types of insurance.

The second category of studies analyzes the reactions of people working in the insurance services towards using information and communications technology, trying to see their reactions related to the use of ICT. In this sense, Lee et al. [48] examine the suitability of mobile commerce systems based on personal digital assistant technology in the insurance sector from the perspective of the task-technology fit theory, and the authors argue that mobile commerce using PDA technology "is suitable for insurance tasks” [48] (p. 108) and that "the PDA technology provides different degrees of assistance to different types of insurance tasks” [48] (p. 108). Odoyo and Nyangosi [49] analyzed the opinions of employees and insurance agents regarding the perceived benefits of implementing ICT in insurance companies and Naicker and Van Der Merwe [50] analyzed the opinions of IT managers from South-African insurance companies in an attempt to identify factors that influence the adoption of mobile technology in the life insurance industry. In an attempt to see the impact of COVID-19 on the use of ICT in insurance, Eckert, Eckert and Zitzmann [51] analyzed the factors that influence the use of digital technologies in the sale of insurance for different intermediaries: exclusive agents, independent agents, and independent brokers. Their study used a questionnaire applied to persons acting in the German insurance market after the first wave of COVID-19 and concluded that the use of digital technologies in insurance sales is rather underdeveloped. The results of the survey underline that about 50% to 60% of the sales units use technology to interact with customers in the sales process; messenger services are the main used digital technologies, followed by video meetings.

In addition to this, technology is more frequently used in the underwriting process or for claims management, as more than 75% of the business transactions related to these aspects used digitalization. This situation confirms the idea that at least one cause of this conservative approach regarding digital technology comes from the consumers, as some of them still prefer face-to-face interactions. Another conclusion of the study highlights the fact that exclusive agents are more open to using digital technologies than independent agents or brokers, and also that younger people are using more digital technology than older ones.

The third category of studies evaluates the impact of ICT infrastructure on insurance companies’ transformations. Neirotti and Paolucci [52] used study cases and econometric techniques to highlight the importance of IT management practices in order to determine better financial results for insurance companies. Lyskawa et al. [53] analyzed the effect of ICT investments determined by digitalization in four major European insurance groups on the results of their activity. They suggested that the impact of ICT investments is diverse for the four insurers, suggesting that just investing in more developed technology does not by itself lead to better financial results. Eckert and Osterrieder [54] analyzed the previous literature to describe the major digital technologies that have significant importance for insurers’ transformation. Based on the benefits and opportunities of such technologies for the insurance activities, the authors suggest that by using these technologies, insurers could provide more customer-centric products and services, and in order for this to be achieved, it is necessary to integrate the digital transformation of the company into a strategic plan.

The fourth strand of studies tried to quantify the effect of ICT penetration on the development of the entire insurance market or on specific types of insurance, and is closely related to our approach.

Using a sample of average income countries from 2002 to 2011, Salatin et al. [37] examined the relationship between ICT use (seen as the number of mobile users) and the insurance industry, and indicated that the number of mobile users positively and significantly influences the insurance industry. Their explanations are based on the fact that sales/claims adjusting in the insurance industry will grow because of ICT use and the activities will be more specialized and, as a result, a higher speed and a better quality of services will be offered to the clients.

More studies were developed for African countries, and different results were revealed. Asongu and Odhiambo [55] noticed that an enhancement of mobile phone penetration and fixed broadband subscriptions generates a positive net effect on life insurance consumption, while a positive net effect on non-life insurance is obtained through an enhancement of fixed broadband subscriptions. Akinlo [8] established that the classic telephone positively influences non-life and life insurance but, on the other hand, mobile phones negatively affect the evolution of all segments of the insurance market. The Internet has a slightly different impact on non-life insurance (which is significantly positively influenced) than on life insurance, where the impact is also positive, but insignificant. Sibindi [56] found that all types of ICT analyzed (mobile, fixed phones, broadband or Internet) have a positive and significant impact on the life insurance market.

Benlagha and Hemrit [57] empirically analyzed the impact of Internet usage on insurance demand using a panel of 24 OECD countries for the 2007–2017 period. They noticed different results between life and general insurance: a positive effect of Internet use on non-life insurance and no effect on the demand for life insurance. The results can be explained by the fact that non-life policies have generally short-term coverage and are more adaptable to the distribution channels agreed by insured people.

Pradhan et al. [6] investigated the short-run link between ICT infrastructure use and the insurance market for a sample of high and medium-income countries, and their results indicate a significant number of cases of insurance market-led ICT infrastructure hypotheses and also relatively similar cases of ICT infrastructure–led insurance market hypotheses, but are also found some cases when the insurance market and ICT infrastructure influence each other or when there are no links between them. They concluded that the ICT infrastructure use in insurance activities is desired in order to supply better or more targeted products to different categories of people by the insurers, and also government policies toward more access to ICT infrastructure are necessary to increase the addressability of insurance products. As suggested by the authors, ICT infrastructure for insurance purposes could also be used in volatile market conditions and periods of high uncertainty, generating reduced risk for clients.

There are studies that analyze just one type of insurance, such as that of Xu et al. [58], which investigated the effect of Internet use on commercial health insurance purchases in China based on the data provided by a survey from 2017. The results show that even the use of Internet significantly influenced commercial health insurance purchases for all residents, but the effect was higher in the case of rural residents.

3. Data and Methods

This paper investigates the interaction between ICT penetration indicators and insurance sector growth in the 10 new member states of the European Union (EU) by a causality test and regression analysis. In econometric analyses, insurance sector development (INSURANCE) is represented by insurance company assets to GDP (%). On the other hand, ICT penetration is proxied by indicators of mobile cellular subscriptions (per 100 people) (MOBILE), and individuals using the Internet (% of population) (INTERNET). The ICT indicators are obtained from the World Bank database [59,60], and the insurance sector data is procured from the World Bank Global Financial Development Database by Mare et al. [61]. On the basis of limited available data for the insurance sector for countries selected, the study covers the period 2000–2020.

The panel comprised the 10 new EU member states: Bulgaria, Croatia, Czechia, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, and Slovenia. Romania is not included in the analysis due to the absence of its insurance data. Econometric tests are performed using the statistical programmes E-Views 12.0, Stata 16.0, and Gauss 11.0.

The main characteristics of the series are displayed in

Table 1. Dataset Characteristics.

Characteristics		INSURANCE	MOBILE	INTERNET
Mean		7.068	104.707	56.427
Std. Dev.		3.766	34.200	23.519
Maximum		17.500	163.131	90.229
Minimum		1.380	9.114	5.371
Bulgaria	Mean	3.651	102.619	40.716
	Std. Dev.	1.058	43.151	21.523
	Maximum	5.750	143.967	70.162
	Minimum	2.190	9.114	5.371
Croatia	Mean	8.521	91.878	50.044
	Std. Dev.	2.521	27.717	22.803
	Maximum	12.560	117.539	79.079
	Minimum	4.770	22.711	6.645
Czechia	Mean	9.238	113.385	57.885
	Std. Dev.	1.815	23.081	23.291
	Maximum	11.570	132.775	81.339
	Minimum	4.690	42.463	9.781
Estonia	Mean	5.841	118.423	69.524
	Std. Dev.	2.468	34.693	19.459
	Maximum	9.130	151.189	90.228
	Minimum	1.900	39.875	28.576
Hungary	Mean	7.215	98.133	55.803
	Std. Dev.	1.064	24.264	24.765
	Maximum	9.030	121.955	84.771
	Minimum	5.530	30.154	6.999
Latvia	Mean	2.307	95.815	59.534
	Std. Dev.	1.020	34.645	25.513
	Maximum	5.050	134.295	88.898
	Minimum	1.380	16.772	6.319
Lithuania	Mean	3.132	122.376	54.035
	Std. Dev.	0.928	45.611	24.458
	Maximum	4.700	163.131	83.056
	Minimum	1.530	14.557	6.427
Poland	Mean	9.122	103.283	53.062
	Std. Dev.	1.548	42.199	22.876
	Maximum	10.730	147.569	83.185
	Minimum	5.070	17.523	7.285
Slovak Republic	Mean	8.213	100.414	64.039
	Std. Dev.	1.557	31.779	22.211
	Maximum	10.070	135.674	89.921
	Minimum	4.860	23.132	9.427
Slovenia	Mean	13.439	100.738	59.631
	Std. Dev.	3.736	15.752	20.317
	Maximum	17.500	120.459	86.601
	Minimum	6.020	61.259	15.110

Source: Author’s calculations.

. The mean value of INSURANCE, MOBILE, and INTERNET are 7.068, 104.707, and 56.427, respectively. However, the variables of MOBILE and INTERNET particularly displayed a significant variation among the analysed countries. Furthermore, the size of the insurance sector differs considerably among selected countries, as we can see from the data presented in Table 1. On the other hand, the mean of mobile cellular subscriptions (per 100 people) is very high in all countries, and the mean of Internet penetration as a percentage of the population is above 50%, except for Bulgaria. The correlation matrix in

Table 2. Correlation matrix.

	LNINSURANCE	LNINTERNET	LNMOBILE
LNINSURANCE	-	0.196	0.136
LNINTERNET	0.196	-	0.386
LNMOBILE	0.136	0.386	-

Source: Author’s calculations.

also reveals a positive correlation between insurance sector growth and both ICT indicators and also supports the absence of multicollinearity problem.

The causal interaction between ICT penetration indicators and insurance sector development is examined with Emirmahmutoglu and Kose’s [62] causality test in view of the cross-sectional and heterogeneity presence among the variables. Emirmahmutoglu and Kose’s [62] causality test performs the causality analysis for each cross-section by applying the bootstrap method to Fisher statistics. The stationarity of the variable $\left(d_{ma{x}_i}\right)$ and optimal lag length $\left(p_i\right)$ are specified before the causality analysis. The error terms for each cross-section are then obtained by the following regression:

$$LNINSURANC{E}_{i,t}={\alpha }_{i,t}+{\displaystyle \sum }_{j=1}^{p_i+d_{ma{x}_i}}{\beta }_{ij}LNINSURANC{E}_{i,t-j}+{\displaystyle \sum }_{j=1}^{p_i+d_{ma{x}_i}}{\gamma }_{ij}LNIC{T}_{i,t-j}+{\epsilon }_{it}$$

(1)

$$LNIC{T}_{i,t}={\alpha }_{i,t}+{\displaystyle \sum }_{j=1}^{p_i+d_{ma{x}_i}}{\beta }_{ij}LNIC{T}_{i,t-j}+{\displaystyle \sum }_{j=1}^{p_i+d_{ma{x}_i}}{\gamma }_{ij}LNINSURANC{E}_{i,t-j}+{\epsilon }_{it}$$

(2)

The null hypothesis of the test suggests the absence of causality between two variables.

Furthermore, the influence of ICT indicators on insurance sector growth is analyzed through the following regression analysis (i (i = 1, …, 10) indicates the countries, and t (t = 2000, …, 2020) indicates the yearly time period):

$$LNINSURANC{E}_{it}={\alpha }_i+{\beta }_{1}LNINTERNE{T}_{it}+{\beta }_{2}LNMOBIL{E}_{it}+{\epsilon }_{it}$$

(3)

4. Empirical Analyses

The causal interaction between ICT indicators and insurance sector development for the selected countries is investigated by a causality test with cross-sectional dependence. In the econometric analysis, the presence of cross-sectional dependence and heterogeneity is firstly tested with tests of LM, LM CD and LMadj, and the results are reported in

Table 3. Cross-Sectional Dependence and Homogeneity Test Results.

Test	Statistic	p Value
LM test [64]	164.5	0.0000
LM CD * [65]	2.484	0.0130
LM adjusted test * [66]	28.22	0.0000
$\tilde{\Delta }$ test	13.982	0.000
$\tilde{\Delta }$adj test	15.540	0.000

* two-sided test.

. The null hypothesis of cross-sectional independence was denied at a 1% significance level, and the test results reveal the subsistence of cross-sectional dependence. In other words, ICT indicators and the insurance sector in one country of the panel can influence the other countries of the panel due to close economic and social relations. Homogeneity is then examined with the delta tilde test of Pesaran and Yamagata [63], and the test results disclosed in Table 3 reveal the presence of heterogeneity. In other words, there exists a country-specific heterogeneity.

The integration levels of the variables should be determined before the implementation of causality and regression analyses, because it is an input for causality analysis and is also necessary in order to avoid a spurious regression. The stationarity analysis of LNINSURANCE, LNMOBILE, and LNINTERNET are examined with the unit root test of Pesaran [67] CIPS (Cross-sectionally augmented IPS [68]) with cross-sectional dependence, and the unit root test results are displayed in

Table 4. CIPS Panel Unit Root Test Results.

Variables	Constant	Constant + Trend
LNINSURANCE	1.353	0.707
d(LNINSURANCE)	−2.927 ***	−3.189 ***
LNMOBILE	1.369	1.611
d(LNMOBILE)	−2.586 ***	−4.664 ***
LNINTERNET	−1.124	−0.584
d(LNINTERNET)	−3.214 ***	−5.646

*** it is significant at 1%.

. All variables are nonstationary for their level values, but they become stationary for their first-differenced values.

The Granger causality test investigates whether one variable is useful for forecasting another variable, and the null hypothesis suggests that there exists no causality from one variable to another [62]. The interaction between mobile cellular subscriptions and insurance sector development is investigated through Emirmahmutoglu and Kose’s [62] causality test, and the results of the causality analysis are displayed in

Table 5. Panel Causality Test Results.

Countries	LNMOBILE ↛ LNINSURANCE		LNINSURANCE↛ LNMOBILE
	Test Statistic	p Value	Test Statistic	p Value
Bulgaria	3.349	0.187	0.526	0.769
Croatia	0.054	0.816	0.047	0.828
Czechia	0.021	0.886	0.001	0.976
Estonia	1.109	0.574	5.203	0.074
Hungary	0.163	0.686	6.707	0.01
Latvia	9.814	0.007	7.135	0.028
Lithuania	0.006	0.94	0.043	0.837
Poland	10.845	0.004	10.731	0.005
Slovakia	11.597	0.003	5.941	0.051
Slovenia	0.079	0.779	0	0.99
Panel	38.739	0.007	39.632	0.006

Source: Author’s calculations.

. The panel-level causality analysis reveals a bidirectional causality between two variables. In another words, both variables are useful in explaining the other. On the other hand, the country-level causality analysis uncovers a bidirectional causality between LNMOBILE and LNINSURANCE in Latvia, Poland, and Slovakia, and a unidirectional causality from LNINSURANCE to LNMOBILE in Estonia and Hungary.

The bidirectional causality between insurance and mobile penetration is sustained by Pradhan et al. [6], but, on the other hand, Asongu and Odhiambo [55] and Sibindi [56] noticed a positive impact of mobile penetration on life insurance, and Akinlo [8] indicated that mobile has a negative effect on insurance development.

The interaction between Internet access and insurance sector development is investigated through the Emirmahmutoglu and Kose [62] causality test, and the results of the causality analysis are displayed in

Table 6. Panel Causality Test Results.

Countries	LNINTERNET ↛ LNINSURANCE		LNINSURANCE↛ LNINTERNET
	Test Statistic	p Value	Test Statistic	p Value
Bulgaria	1.072	0.301	0.002	0.962
Croatia	11.527	0.009	1.135	0.769
Czechia	6.198	0.102	11.089	0.011
Estonia	0.482	0.786	0.067	0.967
Hungary	3.088	0.079	0.973	0.324
Latvia	0.647	0.886	3.041	0.385
Lithuania	2.116	0.347	3.432	0.180
Poland	1.601	0.206	0.277	0.599
Slovakia	0.040	0.998	5.555	0.135
Slovenia	1.000	0.801	28.271	0.000
Panel	37.871	0.012	47.576	0.000

Source: own processing.

. The panel level causality analysis reveals a bidirectional causality between these two variables. On the other hand, the country-level causality analysis uncovers a unidirectional causality from LNINTERNET to LNINSURANCE in Croatia and Hungary, and unidirectional causality from LNINSURANCE to LNINTERNET in Czechia and Slovenia.

A bidirectional causality between insurance and Internet use was reported by Pradhan et al. [6], for life insurance (when penetration was used) and non-life insurance (when the density is used); but generally, a positive impact of the Internet on insurance was found for life insurance [6,8,56] and for the case of non-life insurance [57]. In addition, an impact of insurance on the Internet was found by Pradhan et al. [6] for the case of life insurance when density is used.

Lastly, the influence of ICT indicators on insurance sector growth is investigated through regression analysis. In this context, the Chow (F) test [69] and the Breusch and Pagan LM test [64] are employed for regression model selection, and their results are displayed in

Table 7. Results of Panel Regression Model Selection Pretests.

Test	p Value	Decision
Chow (F) test	0.0000	Alternative hypothesis is accepted. (Fixed effects model is appropriate.)
BP $\left({\chi }^2\right)$ test	0.0000	Alternative hypothesis is accepted. (Random effects model is appropriate.)
Hausman test	0.8807	Null hypothesis is accepted and random effects model is appropriate.

Source: Author’s calculations.

. The null hypothesis of the Chow test [69] indicates that pooled regression is appropriate and the alternative hypothesis suggests that the fixed effects model is appropriate. The alternative hypothesis is accepted because the p value is found to be lower than 5%. On the other hand, the Breusch and Pagan LM test [64] is conducted to make a selection between pooled regression and the random effects model, and the p value is found to be lower than 5%, and in turn an alternative hypothesis suggesting that the random effects model is appropriate is accepted. At the last stage, the Hausman test [70] is implemented to make a selection between the fixed effects model and the random effects model, and the p value is found to be higher than 5%. Therefore, the null hypothesis suggesting that the random effects model is appropriate is accepted.

The influence of ICT indicators on insurance sector growth is estimated through the random effects model, and the coefficients in

Table 8. Results of Panel Regression Estimation.

Variables	Coefficient	p Value
D(LNMOBILE)	0.018	0.820
D(LNINTERNET)	0.346	0.000
C	0.362	0.000
R-squared	0.543
Adjusted R-squared	0.538
F-statistics	122.910 (0.0000)
Woolridge endogeneity test	2.345 (0.1834)
Greene heterockedasticity test	205.13 (0.3254)

Source: own processing.

indicate that both ICT indicators have a positive influence on insurance sector growth, but the impact of Internet penetration on the insurance sector is revealed to be significant. In other words, Internet penetration has a positive influence on insurance sector growth. Endogeneity is also checked considering the presence of panel level bilateral causality between the series, but no endogeneity is revealed. Furthermore, autocorrelation and heteroskedasticity problems are respectively questioned by Wooldridge autocorrelation test [71] and the Greene test [72], and the results of both tests indicate that there exist no problems of autocorrelation or heteroskedasticity.

The causal analysis revealed at the panel level, a bidirectional relationship between the ICT penetration indicators (mobile cellular subscriptions and individuals using the Internet) and the insurance sector development.. These results suggest that the ICT penetration and insurance sector development influence each other, so the measures that are taken to develop one side (for example, more regions covered by mobile networks or by Internet providers, for ICT indicators or the introduction of deductible revenues for specific types of insurance or new mandatory insurance, or for insurance sector development) will generate the same effect on the other side. However, at the country level, the situations are various as a result of the different structures of the insurance markets under analysis, but also because of different evolutions of ICT penetration. The Internet is usually more important in countries where MTPL, travel policies and not-so-complex property policies are more significant because these types of policies are the first offered through electronic channels. Mobile is generally used for contacting insured persons or prospects, for reminders about deadlines, for payments, so it is necessary not only for basic products but also for more complex ones. As Akinlo [8] suggests, the high intermediation of insurance products could also limit the influence of mobile on insurance, because intermediaries have significant personal interactions with the insured persons.

Considering the impact of both mobile cellular subscriptions and individuals using the Internet on insurance sector development, we discovered that only Internet penetration has a significant positive effect on insurance. This result suggests that for the countries analyzed, the measures that determine the growth of the number of individuals using the Internet are more important for the development of the insurance sector.

5. Conclusions

We have tested the causal relationship between various types of ICT penetration and insurance development for the 10 new member states of the European Union using a panel data set covering the period between 2000–2020, and the results revealed a bidirectional causality between ICT penetration indicators and insurance sector development. These results suggest that by creating a constructive environment for increasing Internet usage and mobile phone subscriptions, we can stimulate insurance sector development.

In the financial and insurance services industry, technological transformation is not just something inevitable, but is a major force behind development and innovation. Managers in the insurance sector must understand that they need to take appropriate measures to modernize and expand their technological infrastructure. This requires more resources and significant efforts for all involved in the transformation process, but will reshape operations, reduce costs, and increase profitability.

Policymakers should understand the key role of the Internet for many sectors and ensure an optimal environment for Internet development. Policies and programmes specifically targeting companies with the purpose of improving ICT adoption can help insurers to build resilient innovative strategies and to accelerate the adoption of both hardware and software technologies. Demand-oriented innovation policies can create a supportive environment that encourages innovation in the insurance industry. Insurers can automate information submission, quoting, underwriting and renewal processes, and can improve risk management systems, etc.

The measures adopted to develop the insurance market (such as the improvement of insurers’ websites, the use of robo-advisors or chatbots, laws that favor the distribution of insurance through the Internet, and campaigns of awareness of the insurance role for people and businesses, etc.) will contribute to more mobile subscriptions and Internet users and will enhance ICT penetration. For various countries, the causal relations are different, suggesting that country-specific factors (macroeconomic, demographic or institutional ones) can also have a significant impact on insurance market development. A significant limitation of the study, because of the lack of some data, is that it does only cover the first period of the pandemic, when the use of ICT was more intense, and therefore future studies will be focused on a more extended period of time.