Performance determinants of non-life insurance firms: a systematic review of the literature

Abstract

The performance of non-life insurers is essential to the economy because of their role in mitigating the risks firms and households face. This study provides a comprehensive overview of studies examining factors affecting non-life insurers’ performance. Based on 235 studies published between 1990 and 2021, the review demonstrates that firm-level factors such as size, organisational form, diversification, capital structure, risk, reinsurance, corporate governance, distribution system, and group affiliation, and external factors such as market structure, macroeconomic, financial, and institutional development are the major determinants of non-life insurers’ performance. Although the empirical evidence on the effect of these factors is generally mixed, firm size, capitalisation, risk, macroeconomic conditions, and, to some extent, corporate governance and market structure issues show a clear relationship with insurer performance. One of the implications of this study is that there may be a need for increased solvency surveillance, especially for smaller insurers, which appear to have a higher risk of insolvency than their larger counterparts.

IMPACT STATEMENT

The performance of non-life insurers is important to the economy because of the role they play in mitigating the risks firms and households face. This study provides a comprehensive overview of studies published between 1990 and 2021 that examined factors affecting the performance of non-life insurers. Based on 235 studies the review demonstrates that firm-level factors such as size, organisational form, diversification, capital structure, risk, reinsurance, corporate governance, distribution system, and group affiliation and external factors such as market structure, macroeconomic, financial, and institutional development are the major determinants of non-life insurers’ performance. One of the implications of this survey is that there may be need for increased solvency surveillance, especially of smaller insurers which appear to have a higher risk of insolvency than their larger counterparts. Proper risk management could play a significant role in the operations of non-life insurers.

Keywords:

• Determinants
• performance
• efficiency
• non-life insurance
• systematic literature review

Reviewing Editor:

• David McMillan, University of Stirling, United Kingdom

Subjects:

• Economics
• Finance
• Business, Management and Accounting

1. Introduction

A robust insurance industry plays a significant role in the economy of any nation. Through insurance, economic agents can efficiently manage insurable risks threatening their viability. Insurance firms help reduce the cost of goods and services by indemnifying the insured. Apart from that, they facilitate trade and commerce, substitute and complement government security programs, encourage loss mitigation, and promote investment in critical sectors of the economy. However, the extent to which insurance companies perform these functions could depend on their productivity, efficiency, solvency, and profitability. Hence, it is vital to understand the factors that influence the performance of insurance firms.

Many empirical studies have examined these aspects, including economies of scale and scope, regulatory change, market structure, mergers and acquisitions, organisational form, distribution systems, corporate governance, macroeconomic conditions, and risk. It is worth noting that one stream of the literature that has experienced phenomenal growth focused on the application of frontier efficiency methodologies. For instance, Cummins and Weiss (2000) analysed 21 studies published between 1983 and 1999 that utilised frontier efficiency methodologies. Their study revealed six aspects (i.e., economies of scale and scope, organisation form, distribution systems, mergers and acquisitions, regulatory change, and management strategies) influencing insurer efficiency. Cummins and Weiss (2013) examined 53 additional studies published from 2000 to 2011 and identified two extra factors, market structure and corporate governance. Amel et al., (2004) focused mainly on studies that analysed the effects of mergers and acquisitions on the performance of firms in the financial sector in industrialised countries, including insurance companies.

In addition to discussing most aspects identified in Cummins and Weiss (2000, 2013), Eling and Luhnen (2010b) analysed three more factors (i.e., financial intermediation, risk management, and capital utilisation) explored in the literature on the efficiency of the insurance firms. Their review was based on 95 studies spanning 1993–2008. Kaffash et al., (2020) surveyed 132 studies published between 1993 to 2018. Their review focused exclusively on insurance studies that applied data envelopment analysis (DEA). They identified seven additional factors: business environment, failure, capacity, contingent commissions, customer loyalty, innovation, and intellectual capital. Zinyoro and Aziakpono (2023) conducted a comprehensive systematic analysis, examining 129 studies published between 1991 and 2021 to investigate the drivers of life insurance performance. Notably, among these studies, 71 utilised frontier efficiency methodologies. Their analysis identified seven key firm-level factors (i.e., size, organizational structure, capital structure, diversification, distribution systems, risk management, and reinsurance strategies) and three external factors (i.e., deregulation, competition, and macroeconomic conditions) that influence the performance of life insurers.

This review synthesizes studies exploring factors that may explain the variation in performance across non-life insurers and markets. To our knowledge, this study represents the first and most comprehensive systematization of the literature on drivers of performance of non-life insurers using five main categories of performance measures: efficiency, accounting, market, insolvency prediction, and rating). Previous reviews exclusively focused on studies that utilized frontier efficiency methodologies, except Zinyoro and Aziakpono (2023), who focused on performance drivers for the life insurance segment. Other measures of performance that indicate, for example, an insurer’s solvency, claims-paying ability, and shareholder value are equally important because of the fiduciary nature of the relations in insurance markets (Brockett et al., 1998, Brockett et al., 2004). The present review aligns with Zinyoro and Aziakpono (2023) study in scope but distinguishes itself significantly. While Zinyoro and Aziakpono focused on the performance of life insurance firms, our study concentrates on non-life insurance firms, thus contributing to a comprehensive understanding of the broader insurance industry.

Our study covers the period from 1990 to 2021 and, thus, provides the most recent empirical evidence on factors that influence non-life insurer performance. The study also discusses additional firm-level factors (i.e., capital structure, reinsurance, risk, and group affiliation) and country-level factors (i.e., macroeconomic conditions, financial sector, regulatory and institutional development indicators) that have been identified in the literature as essential drivers of non-life insurer performance.

The rest of the study is organised as follows. Section 2 describes our data collection method, data sources, and the studies we reviewed. Section 3 discusses the determinants of the performance of non-life insurers frequently examined in the literature. Section 4 summarizes the findings, conclusions, policy implications, and areas for future research.

2. Methodology and data description

We conducted a systematic literature review to assess the determinants of non-life insurer performance following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹ Figure 1 shows the main steps to identify the articles for inclusion in the final sample.

Figure 1. Search methods, strategies and sample selection.

2.1. Search strategies

The first step in our search strategy was to define a list of relevant keywords based on BarNiv and McDonald (1992), Cummins and Weiss (2000, 2013), Eling and Luhnen (2010a, 2010b), and Kaffash et al. (2020). We read the literature on organisational performance (particularly Richard et al., 2009) to capture the different performance dimensions. We ultimately used three groups of keywords. The first group comprised performance-related keywords: productivity, efficiency, profitability, cash holdings, solvency, insolvency (failure/financial distress), financial stability, rating, and performance. The second group comprised insurers, property-liability (casualty)/non-life/general,² and insurance (all related to the insurance industry). The last group was related to specific aspects influencing insurer performance, as identified in previous reviews. Examples included organizational form, consolidation, (de)regulation, diversification, market structure, and corporate governance.

We used these keywords to search databases such as the Web of Science, EBSCO, Science Direct Scopus, and Google Scholar. Additional articles were identified through the reference sections of relevant articles and other related reviews. Finally, specific insurance, economics, finance, and management-related journals were manually searched to identify more articles.

2.2. Inclusion and exclusion criteria

For this study, we considered only published papers in peer-reviewed journals, working papers, and book chapters between 1 January 1990 and 31 December 2021 for two main reasons. First, research applying frontier efficiency methodologies in the insurance industry started in the early 1990s (Berger et al., 1993), and have since witnessed a surge in number after that (see Cummins & Weiss, 2000, 2013). Second, most developments in the non-life insurance industry, including regulatory changes, advances in computing and communications, and consolidations, happened during this period (Cummins & Xie, 2008; Eling & Luhnen, 2010b).

We scanned the title, abstract, introduction, and, in some instances, the entire article to determine the eligibility of an article.³ Only studies that examined the factors or aspects influencing the performance outcomes of property-liability insurers were considered. We further excluded studies not written in English and those published in poor language (these papers are available from the authors upon request).

Of the 348 articles identified through database search, 137 were eligible for inclusion in the initial sample. We also identified 55 studies by reviewing the reference sections of existing surveys and other relevant articles. Finally, 43 more studies were added through manual searches of specific journals, bringing the total number of studies in the final sample to 235.

2.3. Bibliometric analysis

This section presents a bibliometric analysis of the studies included in this review.

2.3.1. Distribution of studies by publication source

Table 1 lists the 17 sources with the highest number of studies that explored the drivers of the performance of non-life insurers. Most of the studies (202, 86.0%) in our dataset were published in peer-reviewed journals, while 13 studies (5.5%) were published in working papers and four studies (1.7%) as book chapters. The most common journals include the Journal of Risk and Insurance (42 studies, 20.8%), The Geneva Papers on Risk and Insurance-Issues and Practice (19 studies, 9.4%), Journal of Banking and Finance (11 studies, 5.4%), Journal of Insurance Issues (11 studies, 5.4%), European Journal of Operational Research (6 studies, 3.0%), Risk Management and Insurance Review (5 studies, 2.5%) and Journal of Productivity Analysis (5 studies, 2.5%).⁴

Table 1. Top 17 study sources.

Study source	No. of studies
Journal of Risk and Insurance	42
The Geneva Papers on Risk and Insurance-Issues and Practice	18
Working Papers	13
Journal of Banking and Finance	11
Journal of Insurance Issues	11
European Journal of Operational Research	6
Risk Management and Insurance Review	5
Journal of Insurance Regulation	4
Journal of Productivity Analysis	4
Journal of Risk Finance	4
Journal of Financial Services Research	3
Managerial Finance	3
Book chapters	3
International Journal of Emerging Markets	2
International Journal of Financial Studies	2
Journal of Risk and Uncertainty	2
North American Actuarial Journal	2
Total	136

Source: Authors’ compilation.

2.3.2. Distribution of studies by year

Figure 2 presents the Distribution of studies by year. As evident from the figure, the number of studies exhibited significant fluctuations throughout the study period. 1991, 1994, 2000, and 2006 marked the lowest count of studies (1) recorded, while the highest count occurred in 2021. Notably, the fluctuation in the number of studies between 2008 and 2021 was less pronounced than in the earlier period (1990–2007), yet it consistently remained above 6.

Figure 2. Distribution of studies by year.

2.3.3. Distribution of studies by number of authors

Figure 3 illustrates the Distribution of studies according to the number of authors. Four hundred twelve distinct authors were identified, with an average of approximately two authors per publication. Single authors wrote approximately 16% of the studies, while 43% were collaborative efforts of two authors. Publications attributed to three authors constituted 29%, whereas those involving four authors accounted for 9%. A minority, roughly 4%, were contributed by five authors, representing the lowest proportion.

Figure 3. Distribution of number of studies by authors.

2.3.4. Distribution of studies by country and performance measure

Table 2 shows the Distribution of studies by country and performance measure. As can be observed, most country-specific studies (161, 78.5%) concentrated on developed insurance markets, mainly the US, which accounts for 55.6% (114 studies). The 30 studies in our sample (both regional and global) that explored insurer performance determinants across countries are skewed in favour of European (13 studies, 43.3%) and global insurance (12 studies, 40.0%) markets (see Table 3). Regarding the performance measures, studies in our database utilised five classes of indicators: accounting-, efficiency-, market-, insolvency prediction- and rating-based measures. Accounting-based measures comprise traditional financial ratios such as ROA, ROE, profit margin, loss ratios, combined ratios, and expense ratios. These indicators (dominated by ROA, ROE, and profit margin) were our reviewed studies’ most frequently used performance measures (40.4%, 95 studies). However, traditional financial ratios have been criticised for providing only a partial assessment of performance (Doumpos et al., 2012). The major problem with partial measures is that a firm may perform well based on one indicator but badly based on another, thereby giving an inconclusive picture of a firm’s overall performance.

Table 2. Distribution of studies by country and performance measure.

Country	Performance measure
Developed	Efficiency	Accounting	Market	Insolvency prediction	Rating	Number of studies
Australia	1			1		2
Austria	1					1
Belgium	1					1
Canada		2		1		3
France	3					3
Germany	5	4				9
Italy	1					1
Japan	3	2	1			6
Netherlands	1	1		1		3
Portugal	1					1
South Korea	1	1				1
Spain				5		5
Sweden		1				1
Switzerland	1	1				2
UK		7		1		8
US	29	50	14	14	7	114
Sub-total	48	69	15	23	7	161
Developing
Argentina	1					1
Bangladesh		2				2
China	4	2				6
Ecuador		1				1
Ghana		3				3
India	9					9
Iran	2					2
Malaysia	2	1				3
Pakistan		2				2
South Africa	2	1				3
Taiwan	4	4				8
Thailand	2					2
Turkey	1	1				2
Sub-total	27	17				44
Total	75	86	15	23	7	205

Source: Authors’ compilation.

Table 3. Distribution of multi-country studies by performance measure.

Focus	Performance measures					No. of studies
	Efficiency	Accounting	Market	Insolvency	Rating
Europe	9	2		1	1	13
Eastern Europe		1				1
Asia	1	1				2
SADC		1				1
BRIC	1					1
Global	2	8	1		1	12
Total	13	13	1	1	2	30

Notes: SADC = Southern Africa Development Community.

BRIC = Brazil, Russia, India and China.

Source: Authors’ compilation.

Efficiency measures were the second most popular performance indicators in our sample. These measures dominate traditional performance measures, particularly those that rely on book values instead of market values, such as accounting ratios (Cummins & Weiss, 2013). Efficiency indicators were utilised in 88 studies (37.4%) in our dataset. In this class, technical, scale, allocative, and cost efficiency were the common efficiency measures. Table A1 (Appendix 1) shows our sample’s main frontier efficiency estimation techniques, including sample sizes, inputs, outputs, efficiency types, and average yearly efficiency. Among the two main approaches (DEA and SFA) for measuring efficiency⁵, most frontier efficiency studies (61, 69.3%) applied DEA. DEA dominates in the literature because it possesses several desirable characteristics (Cummins & Weiss, 2013). First, DEA is non-parametric, meaning it avoids misspecification of the functional form (such as cost, revenue, or profit) or distributional assumptions of the error terms. Second, it is firm-specific, allowing for the decomposition of efficiency for each firm. Third, it enables the decomposition of cost and revenue efficiency into pure technical, scale, and allocative components. Fourth, the DEA can operate with a small number of firms. Fifth, it corresponds to maximum likelihood estimation. Sixth, its estimators are consistent and converge much faster than other frontier techniques’ estimators. Seventh, its estimators are also unbiased. Finally, DEA provides reliable estimates of the effect of contextual factors in a two-stage framework.

What is also noticeable is that 82.0% (50) of the DEA-based studies utilized classic models, i.e., the CCR (Charnes et al., 1978) and BCC (Banker et al., 1984) models. Non-classic models include range adjusted measure (RAM) DEA (Brockett et al., 1998, 2004, 2005; Jeng & Lai, 2005), robust DEA (Naini & Nouralizadeh, 2012), SBM DEA (Kweh et al., 2014), network DEA (Hwang & Kao, 2006; Kao & Hwang, 2008; Sinha, 2021), dynamic network DEA (Kuo et al., 2017), multi-stage metafrontier SBM (Shieh et al., 2020) and bootstrapped DEA combined with metafrontier analysis (Doumpos et al., 2018). SFA was applied in 23 studies (26.1%), Varian’s Weak Axiom of Profit Maximisation (VWAPM) in two studies (Emm, 2014; Garven & Grace, 2001), and DFA (Berger et al., 1997) and TFA (Bikker & Gorter, 2011) in one study.

As can be seen from Table A1 in Appendix 1, there appears to be agreement in frontier efficiency studies as to what constitutes inputs and outputs for insurance firms (see Cummins & Weiss, 2013). Labour, business services, debt, and equity capital, represent the commonly used inputs, while losses (claims) incurred and invested assets represent the outputs frequently utilized in our sample. However, due to data limitations, some studies combined labour and business services (e.g., Altuntas et al., 2019; Luhnen, 2009; Wende et al., 2008). Others (especially those outside the US) utilised losses incurred (e.g., Huang et al., 2011; Park & Park, 2015) instead of the present value of losses incurred (e.g., Cummins & Xie, 2008; Weiss & Choi, 2008; Xie, 2010). Using actual values of inputs and outputs in the studies reviewed is also standard practice.

Sixteen (6.8%) studies used market-based measures. Stock returns and Tobin’s Q were the most often utilised in this category. Market-based measures are forward-looking and less prone to manipulation than performance measures that rely on accounting data (book values). Nonetheless, market-based indicators can only be used for publicly traded insurers. For this reason, only a few studies have applied these measures in the insurance industry since most insurers, especially in developing countries, are private companies. Twenty-four studies (10.2%) in our sample used insolvency prediction performance indicators. This stream of the literature relies on samples of insolvent and solvent insurers. The main drawback of this class of performance measures is that data on insolvent insurers is generally unavailable in most jurisdictions. Among the five classes of performance measures, rating-based measures were the least popular; they constituted only 3.0% (7) of the studies we reviewed since ratings are primarily standard among large insurers. All studies based on this class concentrated on the US non-life insurance market.

3. Determinants of non-life insurer performance

The objective of this section is to discuss the performance determinants of non-life insurers. It discusses nine firm-level (i.e., size, organisational form, diversification, capital structure, risk, reinsurance, corporate governance, distribution system, and group affiliation) and three external determinants of non-life insurer performance (such as market structure, macroeconomic and financial sector, and institutional development) that have been commonly investigated in the literature (see Table A2).

3.1. Firm level factors

3.1.1. Size and non-life insurer performance

Size is one of the critical determinants of insurer performance examined in the literature. Large insurance companies can benefit from economies of scale and scope. In addition, they can better weather market instability and retain skilled workers than their smaller counterparts. Consequently, they have a better probability of success than their smaller peers. However, when an insurer becomes too large, complexity may cause coordination challenges, managerial conflicts, and agency concerns, leading to poor performance compared to smaller insurers (Cummins & Weiss, 2013).

The results show that the effect of size on insurer performance varies with the type of performance measured used. Frontier efficiency studies show mixed results across country-specific and cross-country studies. For example, Fenn et al. (2008), Ilyas and Rajasekaran (2019), and Park and Park (2015) concluded that large non-life insurers are less efficient than small insurers. However, Huang and Eling (2013), Luhnen (2009), and Eling and Luhnen (2010b) found the opposite. On the other hand, Kweh et al. (2014) and Naini and Nouralizadeh (2012) found that size does not matter, while Cummins and Xie (2013), Diacon (2001), Dionne et al. (2007) and Worthington and Hurley (2002) found a non-linear (u-shaped) relationship between size and efficiency. Sample sizes, efficiency measurements, periods, methodology, and nations studied may explain these inconsistent results (see Tables A1 and A2 in the Appendix). Most accounting research found that size increased profitability (Altuntas & Gößmann, 2016; Born, 2001; Cummins et al., 2012; Cummins & Nini, 2002; Foong & Idris, 2012; Liebenberg & Lin, 2019). Insolvency prediction studies revealed that large non-life insurers are less likely to go bankrupt (BarNiv & McDonald, 1992; Cheng & Weiss, 2012) than small insurers. Rating-based studies indicated that large non-life insurers perform better financially than small insurers (e.g., Eckles & Pottier, 2011; Gaver & Pottier, 2005; Park & Xie, 2014; Pottier & Sommer, 1999).

3.1.2. Organisational form and non-life insurer performance

The literature has explored the effect of different organisational forms on performance, including performance differences between stock and mutual insurers, independent and keiretsu insurers, international and domestic insurers, listed and unlisted insurers, and public and private insurers. However, most studies in our database focused on stock and mutual insurance firms. These studies tested the expense preference and managerial discretion hypotheses. The expense preference hypothesis states that stock insurers outperform mutual insurers because they have more robust owner-manager dispute resolution mechanisms. According to the managerial discretion theory, stock insurers are expected to dominate in business lines demanding substantial managerial autonomy, especially in pricing and underwriting decisions, such as commercial lines and wider operational scopes (Cummins et al., 1999; Cummins & Weiss, 2013). In contrast, mutuals are better suited for low-discretion insurance lines like personal lines, where the need for customised pricing and underwriting is reduced (Cummins et al., 1999).

There is needs for more consensus in the literature on whether stock or mutual non-life insurers perform better. For instance, in the US, in the non-life insurance sector, Cummins et al. (1999) found evidence for the managerial discretion and expense preference hypotheses regarding production technology and the cost frontier, respectively. Multiple studies, including Brockett et al. (1998, 2004, 2005), Diacon (2001), Doumpos et al. (2018), Eling and Luhnen (2010b), and Shi and Zhang (2011), have concluded that stock insurers are more efficient than mutual insurers. According to other studies, there is no discernible performance difference between stock and mutual non-life insurers (see Cummins & Xie, 2016; Emm, 2014; Garven & Grace, 2001; Weiss & Choi, 2008).

Overall, studies using accounting measures have found that stock insurers are more profitable than mutual insurers (see, for example, Altuntas & Rauch, 2017; Choi & Weiss, 2005; Lei, 2019; Liebenberg & Sommer, 2008) or that organization form has no significant influence on insurer performance (see, for example, Adams & Jiang, 2017; Liebenberg & Lin, 2019; Ma & Elango, 2008).

3.1.3. Diversification and non-life insurer performance

The diversification-organizational performance nexus has garnered considerable attention in the finance literature. Two main hypotheses, conglomeration, and strategic focus, have been tested to explore this link. According to the conglomeration hypothesis, diversified insurers gain from scope economies, tax advantages, a sizeable internal capital market, and risk reduction (Liebenberg & Sommer, 2008) and, as a result, outperform their undiversified counterparts. According to the strategic focus hypothesis, diversified insurers may suffer from overinvestment and capital misallocation, leading to a decline in performance. In our database, 73 studies explored the effect of diversification on the performance of non-life insurers (mainly evaluated using one minus the Herfindahl-Hirschman Index (HHI)).

These studies fall into two streams. The first stream ignores the conditions under which diversification may positively or negatively influence non-life insurer performance. In line with findings from other industries (Erdorf et al., 2013), this strand of the literature showed inconclusive results on the diversification-performance relationship, possibly reflecting the costs and benefits of diversification. In particular, some studies reported a positive effect (Shim, 2017a; Zanghieri, 2009; Zhang & Nielson, 2015), some a negative effect (e.g. Bikker & Gorter, 2011; Chang & Tsai, 2014; Shim, 2011a), and some an insignificant effect (e.g. Alhassan & Biekpe, 2016; Altuntas & Gößmann, 2016; Chang & Jeng, 2016; Davutyan & Klumpes, 2008; Ilyas & Rajasekaran, 2019). The mixed results from this stream of research suggest that contextual variables influence the effect of diversification in the non-life insurance industry.

The second stream investigates the conditions under which diversified insurers may outperform or underperform specialised insurers. Elango et al. (2008), for instance, examined the relationship between diversification and performance (accounting- and market-based measures) in the US insurance industry and found that insurers with higher product and geographic diversification underperform those with lower levels of both. Elango (2009) found that insurers with several product lines are less profitable in hard and soft markets than insurers with a single product line. Ai et al. (2016) reported that diversified non-life insurers with high-quality enterprise risk management (ERM) frameworks perform better in terms of profitability and value (Tobin’s Q) than those with low-quality ERM frameworks. Altuntas et al. (2019) showed that ERM moderates the relationship between an insurer’s diversification level and revenue scope efficiency in the German non-life insurance industry. Liebenberg and Lin (2019) concluded that diversified mutual non-life insurers perform better than their concentrated counterparts during normal economic conditions, but their performance was very dismal during the financial crisis. Doumpos et al. (2018) established that premium diversification had no impact on the efficiency of non-life insurers during crises such as the 2008–2009 global financial crisis and the European debt crisis (2008–2011).

In conclusion, the studies suggest that diversification may result in a discount or a premium, depending on the situation. Nonetheless, additional research is required because the available evidence is limited.

3.1.4. Capital structure and non-life insurer performance

The finance literature has extensively studied the relationship between capital structure (mixture of debt and equity financing) and firm performance. Modigliani and Miller ((1958) conducted the pioneering work on capital structure. Modigliani and Miller (1958) argues that a firm’s value is independent of its capital structure in perfect capital markets. Since the work of Modigliani and Miller (1958), several schools of thought that emphasize different elements, including agency costs, bankruptcy costs, taxes, and information signalling, emerged. These schools of thought suggest an optimal level of debt and equity, thus implying that as leverage (capitalization) increases (decreases), a firm’s value increases but only up to a certain point, beyond which further increases will lead to a decline in performance.

Ninety-four studies in our database investigated the effect of capital structure on non-life insurer performance. The impact of leverage on non-life insurer performance shows conflicting evidence. While some efficiency studies found a positive effect (e.g., Cummins & Xie, 2016; Lei & Schmit, 2010; Luhnen, 2009), others have found either a negative one (e.g., Colquitt et al., 1999; Lee & Lee, 2012; Lei, 2019) or no correlation at all (e.g., Hsu et al., 2015; Ilyas & Rajasekaran, 2019; Zhang et al., 2019). These contradictory findings may be due to different efficiency measures, sample sizes, methodologies, and periods. Nevertheless, non-efficiency studies showed that well-capitalized non-life insurers outperform poorly capitalized insurers in terms of profitability (e.g., Lei & Schmit, 2010; Liebenberg & Lin, 2019; Liebenberg & Sommer, 2008), investment performance (Camino-Mogro & Bermúdez-Barrezueta, 2019; Che et al., 2017) and financial strength performance (e.g., Eckles & Pottier, 2011; Gaver & Pottier, 2005; Pottier & Sommer, 1999).

3.1.5. Risk and non-life insurer performance

The relationship between risk and return has received considerable attention in the finance, management, and economics literature. According to conventional financial wisdom, risk and return are positively correlated. The Capital Asset Pricing Model (CAPM) and the Arbitrage Pricing Theory (APT) support this viewpoint. These theories contend that investors are rewarded primarily for systematic risk, while idiosyncratic risk is eliminated through diversification. When this perspective is applied to the insurance industry, insurers that underwrite riskier lines of business and invest in riskier assets should achieve superior performance. However, excessive risk-taking may result in poor performance due to low prices and the possibility of losing clients (Wakker et al., 1997).

The studies we reviewed primarily focused on the effect of investment and underwriting risks on non-life insurer performance. Generally, these studies concur that investment risk is negatively associated with insurers’ profitability (e.g., Born, 2001; Ma & Elango, 2008; Shim, 2011b, 2017b), efficiency (Cummins et al., 2010, 2012; Cummins & Nini, 2002), cash holdings (Colquitt et al., 1999; Hsu et al., 2015) and financial strength performance (e.g., Gaver & Pottier, 2005; Park & Xie, 2014; Pottier & Sommer, 1999). Similarly, most studies revealed a negative association between underwriting risk and insurer performance (see for example, Altuntas & Rauch, 2017; Carroll, 1993; Chang & Tsai, 2014; Chidambaran et al., 1997; Delhausse et al., 1995).

To conclude, there is a consensus in the literature that investment and underwriting risks harm insurer performance, hence the need for proper risk management.

3.1.6. Reinsurance and non-life insurer performance

Reinsurance is one of the oldest ex-ante risk and capital management tools at primary insurers’ disposal. Reinsurance offers several benefits to ceding insurers. For example, it helps stabilize loss experience, reduce insolvency risk, spread risks, increase underwriting capacity, reduce capital costs, and protect against catastrophic losses (Park et al., 2021). However, ceding insurers may pay more than an actuarially fair price for reinsurance protection, especially when there is a shortage of reinsurance capacity in the market, which usually occurs following significant losses (Cummins et al., 2021). Given that reinsurance has both costs and benefits, its net effect on ceding insurers’ performance depends on whether costs exceed benefits or benefits exceed costs (Lei, 2019). Although the quantification of the costs and benefits of reinsurance is a developing science, studies that explored the effect of reinsurance on non-life insurer performance essentially utilized the ratio of ceded premiums to gross written premium to proxy for reinsurance costs and different measures to estimate the benefits of reinsurance (Lei, 2019). For instance, Scordis and Steinorth (2012) utilized the ratio of reinsurance recoverable to ceded premiums to measure reinsurance costs and excess return, total return, and price to book value to approximate the benefits of reinsurance to publicly traded firms. Lei (2019) utilized the ratio of reinsurance recoverable scaled by reinsurance premiums ceded as the net cost of reinsurance; meanwhile, the author used ‘return on investment, return on underwriting, their respective standard deviations, their correlation, and underwriting leverage as indirect benefits of reinsurance’ (p. 45). Cummins et al. (2021) applied the cost function to approximate insurer costs and the growth rate of the volatility of the loss ratio (loss ratio is defined as the ratio of the present value of incurred losses to premiums earned during the same year) to measure the benefits of reinsurance.

Fifty-eight studies in our dataset explored the effect of reinsurance (usually measured as premiums ceded to gross written premiums ratio) on non-life insurer performance. These studies show contradictory results. For instance, Cummins et al. (2012) found that reinsurance utilization is positively related to efficiency (cost, revenue, and profit) and profitability (ROA and ROE). However, their study also confirmed a performance penalty for insurers with a high business concentration in top counterparties. Similarly, Choi and Elyasiani (2011) and Choi and Weiss (2005) found that reinsurance utilization improves ceding insurers’ cost efficiency and underwriting performance. Lee and Lee (2012) reported that highly profitable insurers purchase less reinsurance cover. Their study also confirmed a reverse causality from reinsurance to insurer performance, meaning that firms with high reinsurance cover tend to be less profitable. Fecher et al. (1993, 1991) found that reinsurance is negatively associated with efficiency. Chen et al. (2010) and Lei (2019) also found that reinsurance hurts insurers’ profitability. Ma et al. (2013), Weiss and Choi (2008), and Altuntas and Rauch (2017) found that performance is independent of insurers’ reinsurance decisions.

One notable observation from the reviewed studies is that most of them are based on a premium-based reinsurance indicator (ratio of premiums ceded to direct premiums written plus reinsurance premiums assumed), which does not consider the claims side of a reinsurance transaction (i.e., recoverable). Thus, future studies could utilise a measure that considers reinsurance’s net cost/benefit to examine the reinsurance-performance relationship.

3.1.7. Corporate governance and non-life insurer performance

Studies analysing the relationship between corporate governance and non-life insurer performance are generally limited. In our sample, only fifteen studies explored this topic. The literature considers board features (composition, independence, tenure, age, and size), CEO attributes (power, age, tenure, duality, expertise, and experience), and board committee characteristics (e.g., composition and expertise). Agency and resource-based theories represent the two leading theories that explain the relationship between corporate governance and firm performance. The agency theory states that unmonitored managers may act in their self-interest rather than the principals’. Thus, strong corporate governance (e.g., diverse and independent boards, audit and risk management committees, financial expertise on audit committees) will improve firm performance. According to the resource-dependency theory, boards and board committees supply human (expertise) and social capital (legitimacy and external ties), both of which are critical resources in an organization (Udayasankar, 2008). This hypothesis predicts that boards and committees with broad knowledge positively impact insurer performance.

Wang et al. (2007) was one of the first studies to examine how corporate governance affects the performance of non-life insurers in Taiwan. Their study revealed that insider ownership, cash-flow rights, and outside directors are positively associated with insurers’ expenses and allocative efficiency. They also found a negative link between performance and ownership concentration, cash-flow rights, board size, and dual CEOs. Huang et al. (2011) found that board size, the proportion of independent directors on the audit committee, director tenure, the average number of directorships, the proportion of insiders on the board, and auditor dependence improve non-life insurer cost efficiency in the US non-life insurance sector. He et al. (2011) observed that insurers with unusual CEO turnover experience higher revenue and cost efficiency than those without.

Leverty and Grace (2012) found that superior managers can remove their company from regulatory scrutiny faster, lowering insolvency risk and cost. Hsu and Petchsakulwong (2010) revealed that board independence positively affects efficiency, while audit committee size, diligence, board duration, age, and ownership have a negative effect. Kuo et al. (2017) found that the risk management committee’s size, frequency, and independence improve operating efficiency. Ames et al. (2018) found that board risk committees improve financial strength ratings and long-term profitability of US non-life insurers. Adams and Jiang (2016) found no significant influence of outside directors on performance. Instead, they observed that insurers with financial skills on their boards perform better. In a later study, Adams and Jiang (2017) concluded that CEO insurance expertise and financial experience improve insurers’ profitability, solvency, and underwriting performance. Hsu et al. (2015) suggested a positive relationship between the proportion of outside directors on the board and finance committee and insurers’ cash holdings. Additionally, they reported that insurers with larger boards and CEO duality have higher growth opportunities (proxied by Tobin’s Q) and lower expenses when they have excess cash in the previous period. These results indicate that insurers hold excess cash to address the underinvestment problem that emanates from insufficient cash holdings (Hsu et al., 2015).

To summarise, there is evidence that monitoring systems, particularly boards and their committees, impact non-life insurer performance. However, given the variety of corporate governance elements examined in the literature, more evidence is needed to draw firm conclusions, thus calling for further empirical research. Given the importance of cash in the insurance sector due to the unpredictability of future pay-outs, the relationship between corporate governance and cash holdings is one topic that merits additional examination (Xie et al., 2017).

3.1.8. Distribution system and non-life insurer performance

In the non-life insurance sector, insurance products and services are primarily distributed through two channels: the direct writer system, in which agents represent just one insurer, and the independent agency system, in which agents represent multiple insurers (Berger et al., 1997). The market imperfections and product-quality hypotheses have been used to test the coexistence of these two distribution systems in the insurance industry. According to the market imperfections hypothesis, independent agency insurers can stay in business despite having higher costs than direct writers because of market imperfections such as price regulation (Weiss, 1990), search costs, and slow transmission of information in insurance markets (Berger et al., 1989). In contrast, the product-quality hypothesis postulates that the higher prices charged by independent agency writers are a result of the higher service quality or greater service intensity that these professionals bring to their clients in the form of, for example, claims settlement services, a more comprehensive selection of products, and lower search costs.

Barrese and Nelson (1992) is one of the first studies in our database to investigate the performance differences between direct and independent agency writers. According to their findings, exclusive agency insurers have lower costs than independent agency insurers. Although Berger et al. (1997) obtained similar results, their analysis shows that the two systems are equally efficient, consistent with the product quality hypothesis. According to Choi and Elyasiani (2011), direct writers are more efficient than independent agency writers in terms of cost- and revenue-X efficiencies, but they are comparably efficient in terms of cost- and revenue-scale efficiency. Weiss and Choi (2008) reported comparable results; however, their investigation found no significant differences in revenue X-, cost-scale efficiency, and profitability. Shim (2017b) demonstrated using quantile regression that independent agency writers outperform (in terms of profitability) direct writers only in the lower quantiles; otherwise, direct writers are more profitable. Shim (2011a, 2011b) revealed that direct writers are more cost-effective, scaled, technically pure, revenue-efficient, and profitable than independent writers.

Similarly, Cummins and Xie (2016) showed that direct writers outperform independent agency writers in revenue and profit. Nevertheless, their findings revealed no substantial difference between the two distribution systems regarding cost efficiency and productivity, in line with Cummins and Xie (2013), Luhnen (2009), and Garven and Grace (2001), who found that the distribution system has no significant impact on efficiency change.

Overall, empirical research on the effect of distribution systems yields conflicting results. Nonetheless, the findings indicate that direct writers outperform independent agency writers.

3.1.9. Group affiliation and non-life insurer performance

Non-life insurers are firms in the financial services sector. While some firms operate independently, others are affiliated with business groups. It is hypothesized that affiliated insurers would underperform their unaffiliated counterparts because policyholders are expected to pay higher prices for unaffiliated insurers’ products (Cummins & Sommer, 1996; Sommer, 1996). The argument is that a group offers a portfolio of options that would be more valuable to shareholders and less valuable to policyholders if held by a single organisation (Cummins & Sommer, 1996). However, because the group structure enables affiliates to share fixed costs, cross-sell products, and take advantage of internal capital markets, affiliated insurers are expected to perform better than their unaffiliated rivals.

The results from studies that explored the effect of group affiliation on non-life insurer performance are mixed. In Japan, Jeng and Lai (2005) found that keiretsu-affiliated firms (firms with close ties to banks and other financial institutions) are more cost-efficient than non-specialized independent (unaffiliated) insurers. In an earlier study of the Japanese non-life insurance industry, Lai and Limpaphayom (2003) reported that keiretsu-affiliated insurers have lower expenses and higher profit and loss ratios than independent insurers. Huang et al. (2012) found that keiretsu-affiliated insurers have lower costs and technical and allocative efficiency than non-affiliated insurers. Park and Park (2015) revealed that in South Korea, affiliated insurers are less efficient and profitable than their unaffiliated counterparts.

In the US, the findings on the effect of group affiliation on insurer performance predominantly showed a negative (see Garven & Grace, 2001; Lei, 2019; Liebenberg & Lin, 2019; Liebenberg & Sommer, 2008; Shim, 2017a, 2011a, 2011b) and insignificant effect (see Cummins & Xie, 2013; Ma & Elango, 2008; Lei & Schmit, 2010; Elango, 2009; Choi & Weiss, 2005; Shim, 2017b; and Chen et al., 2014). In a multi-country context, Altuntas and Rauch (2017), Berry-Stolzle et al. (2011), and Oetzel and Banerjee (2008) suggested that affiliated insurers outperform unaffiliated insurers, while Davutyan and Klumpes (2008) and Eling and Jia (2018) confirmed an insignificant influence of group affiliation on insurer performance.

3.2. External factors

3.2.1. Market structure and non-life insurer performance

The relationship between market structure and insurer performance has received considerable attention in the insurance literature. Three main hypotheses help explain the link between these two variables. First, structure-conduct-performance (SCP) claims that highly concentrated markets are more profitable than weakly concentrated market because of the possibility of collusion among industry players. Second, the relative market power (RMP) hypothesis posits that insurers with relatively large market share may exercise their market power and charge higher prices and, as a result, earn higher profits. Finally, the efficiency structure (ES) hypothesis claims that insurers with superior management and production technologies capture larger market shares. Because these firms are efficient, they may earn higher profits than their inefficient counterparts.

Forty-three studies in our database explored this relationship. Multi-country studies essentially showed that concentration has a positive influence on insurer performance (see Bahloul & Bouri, 2016; Choi & Weiss, 2005; Diacon, 2001; Ma et al., 2013; Njegomir & Stojić, 2011; Pope & Ma, 2008). Similarly, Ma et al. (2013) found that concentration positively affects insurers’ performance but only in the presence of foreign insurers. Nevertheless, Altuntas and Rauch (2017) revealed that countries with higher concentration levels experience financial instability in their insurance industries. Country-specific studies showed mixed evidence on the effect of concentration on non-life insurer performance. Bajtelsmit and Bouzouita (1998a, 1998b), Choi and Weiss (2005), and Liebenberg and Sommer (2008) found that insurers operating in highly concentrated markets outperform those operating in lowly concentrated markets. Shim (2017b) reported a positive relationship in upper quantiles and a negative relationship in lower quantiles. Carroll (1993) and Shim (2017a) found a negative association between concentration and insurer performance.

The results on the effect of market share on non-life insurer performance are also mixed. Kramer (1996), Maichel-Guggemoos and Wagner (2019), Park and Park (2015), and Zanghieri (2009) provided some evidence that firms with larger market shares achieve higher performance than firms with smaller market shares. Pope and Ma (2008) found that the effect of market share on insurer profitability varies depending on the level of market liberalisation. Specifically, their findings revealed that insurers in concentrated markets engage in collusive behaviour if there are high barriers to entry for foreign insurers. Camino-Mogro and Bermúdez-Barrezueta (2019), Jeng and Yang (2014), and Fenn et al. (2008) showed that market share is negatively related to non-life insurer performance. Choi and Weiss (2005), Berry-Stolzle et al. (2011), Ma et al. (2013), Lee and Lee (2012), and Naini and Nouralizadeh (2012) suggested that market share is not an essential determinant of insurer performance. Alhassan and Biekpe (2016) reported that competition (measured using the Panzar-Rosse H-statistic) positively affects the cost and profit efficiency of non-life insurers operating in South Africa. In a subsequent study, Alhassan and Biekpe (2018) found a non-linear relation (U-shaped) between competition (measured using the Lerner index) and non-life insurers’ solvency (measured using the Z-score).

To conclude, country comparison studies suggested that more concentrated industries outperform less concentrated industries. Studies examining market share’s effect on non-life insurer performance produced mixed results. Notably, only a limited number of studies utilised non-structural measures of competition.⁶. Hence, future studies could adopt these measures.

3.2.2. Macroeconomic factors and non-life insurer performance

The economic environment in which firms operate has a bearing on insurer performance. The most popular macroeconomic factors in the literature are GDP growth, inflation, and interest rates. A growing economy is highly associated with increased demand for insurance products (Outreville, 1990). Thus, it is generally predicted to be accompanied by an improvement in non-life insurer performance, at least in the short term (Pope & Ma, 2008). Changes in inflation and interest rates affect the values of assets and liabilities of non-life insurers. For example, an increase in inflation results in an increase in the value of claims, negatively affecting non-life insurers’ performance.

Studies that examined the impact of economic growth on non-life insurer performance showed mainly a positive (e.g. Davutyan & Klumpes, 2008; Doumpos et al., 2012; Elango & Wieland, 2015; Hsieh et al., 2015; Ishtiaq, 2017; Wang, 2010) and insignificant (e.g. Altuntas & Rauch, 2017; Caporale et al., 2017; Eling & Jia, 2018; Jeng & Yang, 2014) effect. Results from studies that investigated the relationship between inflation and non-life insurer performance indicated largely a negative (e.g., Adams et al., 2019; Altuntas & Rauch, 2017; Chang & Tsai, 2014; Doumpos et al., 2012, 2018; Shiu, 2004) or insignificant (e.g., Altuntas & Rauch, 2017; Browne & Hoyt, 1995; Eling & Jia, 2018) effect. Similarly, studies that analysed the effect of interest rates on non-life insurer performance found either a negative (see Bajtelsmit & Bouzouita, 1998a, 1998b; Gius, 1998; Haley, 1993; Shiu, 2004) or an insignificant (e.g., Altuntas & Rauch, 2017; Browne & Hoyt, 1995; Fields et al., 2012) effect.

Overall, the studies above agree that low economic growth, high inflation, and interest rates are associated with a decline in non-life insurer performance.

3.2.3. Financial sector and institutional development and non-life insurer performance

A few studies that explored the relationship between the financial sector and institutional development and non-life insurer performance are mainly cross-country studies. Among these studies, Doumpos et al. (2012), Hsieh et al. (2015), and Moro and Anderloni (2014) found that stock market capitalization is positively related to insurer performance. Concerning the effect of insurance sector development (measured by the penetration ratio minus gross written premiums divided by GDP) on non-life insurer performance, Ma et al. (2013) reported a negative effect in the Asian insurance markets, while Bahloul and Bouri (2016), Eling and Jia (2018), Moro and Anderloni (2014) and Doumpos et al. (2012) reported an insignificant impact in the EU and global insurance markets, respectively. Doumpos et al. (2012) also considered the effect of banking sector development (measured as bank credit to GDP) on non-life insurer performance. Their results indicate that banking sector development is not an essential driver of the performance of the non-life insurance industry.

Fields et al. (2012) found limited evidence that better investment protection (proxied by anti-director, anti-self-dealing, disclosure index, liability standard index, and creditor rights), higher quality government (measured by the rule of law, anti-corruption, common law, and regulation quality), and greater contract enforcement (proxied using judicial independence and business environmental risk intelligence) collectively affect the underwriting performance of non-life insurers. They concluded that these factors lead to less risk-taking by insurers, preventing managers from expropriating wealth from policyholders and outside stockholders. Boubakri et al. (2008) reported that mergers and acquisitions of firms in countries with weaker investor protection positively correlate with performance. Similarly, Elango and Wieland (2015) concluded that the quality of governance negatively influences insurers’ profitability. Oetzel and Banerjee (2008) confirmed that insurers operating in emerging countries with better regulatory environments outperform insurers operating in environments with poor regulatory quality in terms of profitability (measured by ROA and return on premium). Zanghieri (2009) supported these findings regarding cost and profit efficiency.

Davutyan and Klumpes (2008) showed that regulatory scrutiny (control of corruption) in the EU insurance industry has a positive effect on managerial efficiency, a negative effect on scale efficiency, and an insignificant effect on overall technical efficiency (OTE). Doumpos et al. (2012) revealed that the institutional environment (measured by the institutional development index, enforcement index, financial freedom index, and economic financial freedom index) has no significant effect on non-life insurer performance (measured by a multi-criteria score).

To sum up, the above empirical evidence indicated that the effect of financial sector and institutional development on non-life insurer performance varies with the independent variable proxy, performance measure, sample size, geographical coverage, and methods employed. It is also evident that the studies that explored the effect of these variables on non-life insurer performance are few, which makes it challenging to conclude their impact.

4. Summary and conclusion

The determinants of the performance of non-life insurers have received considerable interest in the insurance literature. However, a comprehensive study is yet to be carried out to synthesise this stream of research. Therefore, this study was to fill this research gap by systematically reviewing 235 studies published between 1990 and 2021. The most popular journals in our dataset were the Journal of Risk and Insurance, The Geneva Papers on Risk and Insurance-Issues and Practice, the Journal of Banking and Finance, the Journal of Insurance Issues and Risk Management, and the Insurance Review. About country-specific studies, 57% focused on the US insurance industry, and cross-country studies are skewed in favour of developed insurance markets. We identified five classes of performance measures Accounting- (traditional financial ratios), efficiency-, market-, insolvency prediction- and ratings-based (financial strength) measures. Traditional financial ratios (99 studies, 42.1%) have been the most frequently used performance measures in the literature, followed by efficiency-based measures with 88 studies (37.4%). Market-, insolvency prediction- and rating-based performance measures were utilized in 15 (6.8%), 24 (10.2%), and 9 (3.8%) studies, respectively.

We identified nine firm-level factors (size, organisational form, diversification, capital structure, risk, reinsurance, corporate governance, distribution system, and group affiliation) and three external factors (market structure, macroeconomic conditions, and financial sector and institutional development) that have been frequently investigated in the literature as the most critical drivers of non-life insurer performance. Insurer size showed a non-linear effect on efficiency, a positive effect on profitability and financial strength performance, and a negative effect on the insolvency of non-life insurers. The findings on the effect of organizational form, reinsurance, leverage, distribution system, group affiliation, and market share on insurer performance are mainly mixed. The mixed results could be due to different performance measures, methodologies, sample sizes, jurisdictions, and proxies. There is a consensus in the literature that capitalisation positively affects performance. Also, insurers that assume higher investment and underwriting risks underperform those with lower risk.

The presence of monitoring mechanisms, particularly boards and their committees, improves non-life insurer performance. Although country-specific studies indicate mixed evidence on the effect of concentration on non-life insurer performance, most multi-country studies showed that concentration positively influences non-life insurer performance. The literature also provided some evidence that low economic growth, high inflation, and interest rates are associated with a decline in non-life insurer performance. The effect of financial sector and institutional development on non-life insurers varies across studies.

Several research gaps emerged from this survey. First, country-specific and multi-country studies concentrated on developed insurance markets. Future research could focus on emerging and developing countries. Second, more recent DEA methodologies, such as dynamic DEA, need to be applied, which provide room for their adoption. Third, in exploring the effect of diversification on non-life insurer performance, it may be essential to consider the conditions under which diversification can result in a discount or premium, for example, organizational form, leverage, reinsurance, and risk. Fourth, some performance determinants, such as corporate governance, financial sector, and institutional development, have received less attention in the literature. Hence, there is a need for more studies to investigate their effect on non-life insurer performance. One specific topic that needs investigation is the nexus between corporate governance and cash holdings. Finally, additional studies could utilise non-structural competition measures, such as the Boone indicator.⁷, to analyse the competition-performance relationship.

One of the implications of this survey is that there may be a need for increased solvency surveillance, especially of smaller insurers, which appear to have a higher risk of insolvency than their larger counterparts. Proper risk management could play a significant role in the operations of non-life insurers. Adopting and enforcing a risk-based supervisory framework may be worthwhile to ensure that policyholders are protected against excessive risk-taking by non-life insurers. Insurance supervisors may also need to ensure that insurance firms have well-functioning boards and committees.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Tafadzwanashe Zinyoro

Prof. Meshach J. Aziakpono is a Professor of Economics in the Department of Economics and Economic History at Rhodes University, South Africa. Before joining Rhodes University, he was a Professor of Development Finance and the Head of Development Finance programmes at Stellenbosch Business School. He holds PhD degree in Economics from the University of the Free State, Bloemfontein in South Africa. His PhD thesis titled: ‘The Depth of Financial Integration and its Effects on Financial Development and Economic Performance of the Southern African Customs Union Countries’ won the Founders’ Medal for the best PhD dissertation in Economics in South Africa. He has published over 50 papers in international peer-reviewed journals and chapters of books.

Meshach Jesse Aziakpono

Tafadzwanashe Zinyoro (PhD) holds in Development Finance at Stellenbosch University Business School. His thesis is titled ‘Insurer Performance and its Determinants: Evidence from Selected African Countries.’ His research interests encompass the efficiency and productivity of insurers and other financial service providers, the regulation of insurance markets, inclusive insurance, as well as climate and disaster risk financing and insurance.

Notes

1 PRISMA is a set of guidelines and a checklist to improve the reporting quality of systematic reviews and meta-analyses. Its purpose is to enhance transparency and accuracy by providing a structured framework for researchers to follow in conducting and reporting these types of research studies (Sarkis-Onofre et al., 2021).

2 We interchangeably use property-liability, property-casualty, general, and non-life because terminology differs from region to region.

3 For inclusion in the final sample. Full articles were examined, mainly if it needed to be made apparent that the study focused on the property-liability industry.

4 Other journal sources include Academia Economic Papers, Acta Universitatis Danubius: Oeconomica, Aestimatio, African Journal of Business and Economic Research, Annals of Operations Research, Applied Economics, Applied Economics Letters, Applied Financial Economics, Asia-Pacific Journal of Risk and Insurance, Asian Economic and Financial Review, Benchmarking, An International Journal Benchmarking: An International Journal, British Accounting Review, British Actuarial Journal, British Journal of Management, China Economic Review, Computers & Operations Research, De Economist, Economic Modelling, Economic Research-Ekonomska Istraživanja, Empirical Economics, Eurasian Economic Review, European Journal of Finance, Financial History Review, Fuzzy Economic Review, Indian Economic Review, Insurance and Risk Management, Insurance Markets and Companies, Insurance: Mathematics and Economics, Intelligent Systems in Accounting, Finance and Management, International Business Review, International Journal of Economic Sciences, International Journal of Industrial Organisation, International Journal of Information Systems in the Service Sector, International Journal of Management, International Journal of Marketing, Financial Services and Management Research, International Journal of Systems Science, International Journal of the Economics of Business, International Review of Accounting, Banking and Finance, International Review of Applied Economics, International Review of Financial Analysis, International Review of Law and Economics, Investment Management and Financial Innovations, Islamic Economic Studies, IUP Journal of Risk and Insurance, Jing Ji Lun Wen Cong Kan, Journal of Accounting and Public Policy, Journal of Applied Statistics, Journal of Asian Business and Economic Studies, Journal of Business, Journal of Business and Economic Studies, Journal of Business Finance and Accounting, Journal of CENTRUM Cathedra, The Business and Economics Research Journal, Journal of Centrum Cathedra: Business and Economics Research Journal, Journal of Developing Areas, Journal of Economic and Administrative Sciences, Journal of Economic Studies, Journal of Economics and Management, Journal of Financial Economics, Journal of Financial Intermediation, Knowledge Management, Journal of Money, Credit and Banking, Journal of Regulatory Economics, Journal of Risk and Financial Management, Journal of Service Management, Journal of Sustainable Finance and Investment, Journal of the Operational Research Society, Management Decision, Management Science, Mathematical Problems in Engineering, Omega, Pakistan Business Review, Service Industries Journal, South African Journal of Economics, Spanish Journal of Finance and Accounting, Strategic Management Journal, Taiwan Economic Review, The Accounting Review, The Annals of the University of Oradea, The British Accounting Review, The European Journal of Finance, The International Journal of Business and Finance Research, The International Journal of Digital Accounting Research, Tijdschrift voor Economie en Management, Total Quality Management and Business Excellence, Zeitschrift für die gesamte Versicherungswissenschaft.Journal of Financial Stability, Journal of Intellectual Capital, Journal of International Business Studies, Journal of International Management.

5 See Cummins and Weiss (2013) for a detailed discussion of the frontier efficiency approaches, including their advantages and disadvantages.

6 These measures do not rely on information on the market structure, such as size and number of firms, but instead directly measure competition (Bikker & Van Leuvensteijn, 2008).

7 It is an indirect measure of competition, posing that firms with diminished marginal costs are more efficient and consequently acquire larger market shares or higher profits (Abel & Marire, 2021). It combines the Lerner Index, which measures a company’s market power, with the Herfindahl-Hirschman Index (HHI), which measures market concentration. The Boone Indicator is denoted by: Boone Indicator = (1 - Lerner Index) * (1 / HHI).

A positive Boone Indicator indicates that a firm has some market power and can set prices above marginal costs. At the same time, a negative Indicator suggests that firms are pricing closer to their marginal costs due to competitive pressures.

Appendix 1.

Summary of frontier efficiency studies on non-life insurance-methodologies, inputs, outputs and average efficiencies: 1990–2021

Table A1. Frontier efficiency studies in the non-life insurance industry-methodologies, inputs, outputs and average efficiencies: 1990–2021.

Study	Country	Sample Period	No. of Insurers	Efficiency Estimation	Inputs	Outputs	Efficiency Type	Average Efficiency Per Year
Country specific studies-developed countries
Fecher et al. (1991)	France	1984–1989	164	SFA	Labour costs, other outlays	Gross premiums	CE	0.412
Cummins and Weiss (1993)	US	1980–1988	261	SFA	Labour, business services, equity capital	PV of expected losses, reserves	CE	0.858
Fecher et al. (1993)	France	1984–1989	164	SFA, DEA	Labour costs, other outlays	Gross premiums	TE (DEA) CE (SFA)	0.537 0.412
Delhausse et al. (1995)	Belgium, France	1984–1988	191, 243	DEA, SFA	Labour costs, other outlays	Premiums	TE	0.506 (SFA), 0.447 (DEA)
Berger et al. (1997)	US	1981–1990	472	DFA	Labour, business services, debt capital, equity capital	PV of losses incurred, invested assets	CE PE	0.387 0.315
Brockett et al. (1998)	US	1989	1,524	RAM-DEA	Surplus previous year, change in capital and surplus, underwriting and investment expenses, debt capital	Rate of return on investments, liquid assets to liability, solvency scores	Efficiency	n/a
Cummins et al. (1999)	US	1981–1990	417	DEA, Cross-Frontier	Labour, business services, debt capital, equity capital	PV of real losses incurred, invested assets	TE CE	0.869 0.600
Garven and Grace (2001)	US	1986–1996	500	Varian’s Weak Axiom of Profit Maximisation	Administrative labour, agent labour, physical capital, financial capital, miscellaneous goods	Net revenue	Efficiency score	—
Cummins and Nini (2002)	US	1993–1998	645	DEA	Administrative labour, agent labour, materials and business services, average equity capital	Present value of losses incurred	TE PTE SE AE CE RE	0.512 0.576 0.888 0.798 0.406 0.271
Worthington and Hurley (2002)	Australia	1998	46	DEA	Labour, information technology, physical capital, equity capital	Net premium revenues, invested assets	PTE SE AE CE	0.762 0.729 0.309 0.296
Brockett et al. (2004)	US	1989	1,524	RAM-DEA	Surplus previous year, change in capital and surplus, underwriting and investment expenses, debt capital	Rate of return on investments, liquid assets to liability, solvency scores	Efficiency	n/a
Ennsfellner et al. (2004)	Austria	1994–1999	53	Bayesian SFA	Operating expenses, debt capital, equity capital	Loss incurred, invested assets	TE	0.719
Turchetti and Daraio (2004)	Italy	1982–2000	45	DEA, TFP	Acquisition, production and organisational costs	Real incurred losses	TE CE AE SE TFP EC TC	0.980 0.780 0.790 0.980 1.060 1.020 1.050
Brockett et al. (2005)	US	1989	1,524	RAM-DEA	Surplus previous year, change in capital and surplus, underwriting and investment expenses, debt capital	Rate of return on investments, liquid assets to liability, solvency scores	Efficiency	n/a
Jeng and Lai (2005)	Japan	1985–1994	19	RAM-DEA, Cross-Frontier	Labour, business services, capital	Number of policies, invested assets	TE	0.972 (Keiretsu), 0.976 (NSIF), 0.911 (SIF)
Dionne et al. (2007)	US	1995–2003	369	SFA	Administrative labour, agent labour, risk management labour, debt capital equity capital	PV of losses incurred, invested assets, level of risk management	Risk Pooling Efficiency	0.481
Cummins and Xie (2008)	US	1994–2003	1,550	DEA	Administrative labour, agent labour, business services, equity capital	PV of losses incurred, average invested assets	PTE AE SE CE RE TFP PTEC PTC SEC STC	0.640 0.710 0.930 0.450 0.390 1.050 1.010 2.440 1.010 0.600
Weiss and Choi (2008)	US	1992–1998	1,119	SF	Administrative labour, agent labour, materials, equity capital	PV of losses incurred, invested assets	CE RE CSE RSE	0.810 0.744 0.877 0.911
Wende et al. (2008)	Germany	1988–2005	40	DEA, Cross-frontier	Labour and business services, debt capital, equity capital	PV of incurred claims, invested assets	TE AE CE	— — —
Cummins et al. (1995)	US	1995–2003	369	SFA	Administrative labour, agent labour, risk labour, materials/ business services, debt capital, equity capital	PV of losses incurred, intermediation (intermediate), risk management	CE	0.480
Luhnen (2009)	Germany	1995–2006	229	DEA	Labour and business service, debt capital, equity	Claims incurred, invested assets	TE SE AE CE TFP TC TEC PTEC SEC	0.800 0.940 0.590 0.480 0.918 0.921 0.997 0.967 1.032
Park et al. (2009)	US	1990–2001	370	SFA	Agent labour, other labours, materials, equity capital	Incurred losses	CE RE	0.687 0.771
Cummins et al. (2010)	US	1993–2006	718	DEA	Administrative labour, agent labour, materials and business services, equity capital	PV of losses incurred, average invested assets	TE PTE SE CE AE RE PE	0.572 0.664 0.870 0.423 0.742 0.336 1.023
Xie (2010)	US	1994–2005	312	DEA	Administrative labour, agent labour, business services, equity capital	PV of losses incurred, average invested assets	TE AE SE CE RE	0.554 0.668 0.758 0.373 0.330
Bikker and Gorter (2011)	Netherlands	1995–2005	195	TFA	Labour, debt capital, equity capital	Net premium, losses incurred, total investments	CE SE	0.170 0.930
Chen et al. (2011)	US	1990–2001	540	DEA	Labour, business services, debt capital, equity capital	Incurred losses, invested assets	TE AE CE TFP TEC TC	0.964 0.892 0.860 1.016 1.005 1.010
He et al. (2011)	US	1995–2006	557	DEA	Administrative labour, agent labour, materials and business services, equity capital	Real PV of losses incurred, average invested assets	CEC REC	1.037 0.998
Huang et al. (2011)	US	2000–2007	28	DEA	Labour, business services, equity capital	Losses incurred, invested assets	TE CE	0.904 0.799
Lin et al. (2011)	US	2002–2004	1,657	SFA	Equity capital, debt capital	Net investment income, losses incurred	CE	—
Shi and Zhang (2011)	US	2001–2006	1,527	SFA	Administrative labour, agent labour, business services and materials	Losses incurred, invested assets	CE	0.579
Shim (2011a)	US	1990–2004	190	DEA	Administrative labour, agent labour, business services and materials, equity capital	PV of losses incurred, average invested assets	PTE SE AE CE RE	0.621 0.809 0.746 0.358 0.406
Cummins et al. (2012)	US	1993–2009	1,260	DEA	Administrative labour, agent labour, business services, equity capital	PV losses incurred, average invested assets	CE RE PE	0.328 0.237 0.884
Huang et al. (2012)	Japan	1992–2005	22	DEA	Labour, business services, equity capital	Net written premiums, invested assets	TE AE CE	0.502 0.527 0.933
Leverty and Grace (2012)	US	1989–2000	1,018	DEA	Administrative labour, agent labour, business services and materials, debt capital, equity capital	PV of losses incurred, average invested assets	PTE SE AE CE RE	0.667 0.898 0.590 0.352 0.233
Brito et al. (2013)	Portugal	1999–2007	13	SFA	Acquisition services, labour and business services, claims, equity capital	Number of policies	CE	0.990
Cummins and Xie (2013)	US	1993–2009	781	DEA	Administrative labour, agent labour, materials and business services, equity capital	PV of losses incurred, average invested assets	PTE SE CE RE AE TFP PTEC PTC SEC STC	0.740 0.900 0.510 0.440 0.770 1.065 1.009 1.096 1.077 0.993
Ma et al. (2013)	US	1993–2008	1,116	DEA	Administrative labour, agent labour, materials and business services, equity capital	PV of losses incurred, invested assets	CE RE	0.470 0.330
Emm (2014)	US	1988–2001	630	Varian’s Weak Axiom of Profit Maximisation	Administrative labour, agent labour, physical capital, business services, equity capital	Net revenue	CE	0.820
Park and Park (2015)	South Korea	2006–2011	17	DEA	Labour, business services, debt capital, equity capital	Incurred losses, invested assets	TE PTE	0.920 0.930
Biener et al. (2016)	Switzerland	1997–2013	51	DEA	Labour and business services, debt capital, equity capital	Losses incurred, total investments	TE CE AE SE RE TFP	0.912 0.852 0.933 0.954 0.835 1.02
Cummins and Xie (2016)	US	1993–2011	768	DEA	Administrative labour, agent labour, materials and business services, equity capital	PV of losses incurred, average invested assets	PTE SE TE AE CE RE PE TFP TC TEC	0.740 0.900 0.660 0.780 0.510 0.440 0.630 1.173 1.165 1.015
Altuntas et al. (2019)	Germany	1999–2009	84	DEA, FE	Labour and business services, debt capital, equity capital	PV of losses incurred, Average invested assets	TE PTE SE CE RE PE	0.859 0.907 0.948 0.505 0.581 0.530
Eling et al. (2020)	Germany	1954–2016	95	TFP, Multiple Regression	Labour, debt capital, equity capital	PV of losses incurred, total investments	TFP	1,003
Country specific studies-Emerging countries
Leverty et al. (2004)	China	1995–2002	14	DEA	Labour, business services and materials, debt capital, equity capital	Losses incurred, invested assets	TE PTE SE	0.866 0.902 0.957
Hwang and Kao (2006)	Taiwan	2001–2002	24	Network DEA	Stage 1: Business and administrative expenses, commissions and acquisition expenses; Stage 2: Direct written premiums, reinsurance premiums received	Stage 1: Direct written premiums, reinsurance premiums received; Stage 2: Net underwriting income, net investment income	Marketing Efficiency Profitability Efficiency TE	0,849 0,596 0,561
Wang et al. (2007)	Taiwan	2000–2002	16	DEA	Labour, business services, debt capital, equity capital	Losses incurred, invested assets	TE AE CE	0.871 0.789 0.716
Kasman and Turgutlu (2009)	Turkey	2000–2005	22	Malmquist Index	Labour, business services, equity capital	Total losses paid	TFP EC TC PEC SEC	1.061 0.745 1.423 0.957 0779
Hsu and Petchsakulwong (2010)	Thailand	2000–2007	17	DEA	Labour, materials and business services, equity capital	Losses incurred, invested assets	TE AE CE RE	0.941 0.626 0.593 0.852
Singh and Kumar (2011)	India	1993–2008	4	DEA	Operating expenses, equity capital	Net written premiums, incurred claims, investment income	TE PTE SE	0.769 0.861 0.893
Naini and Nouralizadeh (2012)	Iran	2003–2010	17	Robust DEA	Labour, general and administrative expenses, equity capital	Losses incurred, ROE	Overall efficiency (CRS)	0.728
Chen et al. (2014)	Malaysia	2008–2011	16	DEA, OLS	Real labour and business services, debt capital, equity capital	Incurred claims + additions to reserves, invested assets	TFP TEC TC	1.212 0.992 1.220
Kweh et al. (2014)	China	2006–2010	32	Dynamic SBM-DEA	Input1 = Operating expenses Carry-over1 = debt capital Carry-over2 = equity capital	Incurred losses + additions to reserves, investment profits	Operating efficiency	0.839
Yaisawarng et al. (2014)	Thailand	2000–2007	46	SFA	Labour, business services, debt capital, equity capital	No. of policies, invested assets	CE (Model A) CE (Model B) Scale Eco, Model A, B	0.793 0. 742 0.922, 0.916
Alhassan and Biekpe (2015)	South Africa	2007–2012	70	DEA,	Labour and business services, Debt capital, Equity capital	Net premium earned, Claims incurred, Investment income	TE (Model 1, Model 2) PTE (Model 1, Model 2) SE (Model 1, Model 2)	0.522, 0.573 0.593, 0.672 0.878, 0.873
Zakery and Afrazeh (2015)	Iran	2006–2012	17	DEA	1^st stage (intellectual capital creation) Owned (financial) resources, no. of employees, general and administrative expenses 2^nd stage (intellectual capital application) Staff education, no. of agencies, product portfolio diversity	1^st stage (intellectual capital creation) Staff education, no. of agencies, product portfolio diversity 2^nd stage (intellectual capital application) ROE, losses incurred	TE	0.769
Alhassan and Biekpe (2016)	South Africa	2007–2012	66	SFA	Labour and business services, Debt capital, Equity capital	Claims incurred, Investment income	CE PE	0.801 0.457
Kuo et al. (2017)	Malaysia	2008–2013	16	Dynamic Network SBM-DEA	Input1 = Labour (staff costs) Input2 = Business services Intermediate1 = Incurred claims + additions to reserves Intermediate2 = Investment assets Carry-over1 = total liabilities Carry-over2 = equity capital	Net income, operating cash flow	Marketing efficiency (Stage 1) Profitability efficiency (Stage 2) Overall operating efficiency	0.854 0.932 0.848
Sinha (2017)	India	2013–2014	16	DEA	Operating expenses, net premium income	Operating income, assets under management, benefit paid	TE (input orientation) TE (output orientation)	0.9001 0.898
Ferro and León (2018)	Argentina	2009–2014	69	SFA	Direct labour costs, commissions to brokers, financial capital (debt and equity)	Total losses	TE	0,423
Ilyas and Rajasekaran (2019)	India	2005–2016	14	DEA	Labour and business services, debt capital, equity capital	Losses incurred, invested assets	TE PTE SE AE CE TFP TC TEC PTEC SEC	0.792 0.895 0.884 0.817 0.818 1.321 1.268 1.042 1.218 0.855
Ilyas and Rajasekaran (2020)	India	2005–2016	15	DEA, Fare-Primont PI	Labour and business services, debt capital, equity capital	Losses incurred, total investments	TFP	1,124
Ilyas and Rajasekaran (2021)	India	2005–2016	15	DEA	Labour and business services (operating expenses), debt capital, equity capital	Incurred claims, total investments	TFP	1,007
Li et al. (2021)	China	2011–2017	47	DEA	Total capital, fixed assets, management expenses	Premium written, investment income, incurred claims	TE TFP	0,905 1,055
Zhao et al. (2021)	China	2013–2017	53	DEA	Compensation expense, tax and surcharges, service charge and commission fee, operation and administrative expense	Premium earned	PE AE TE	0,626 0,746 0,827
Sinha (2021)	India	2010–2018	15	Network DEA	Stage 1: number of branches, operating expenses, equity capital; Stage 2: net premium income	Stage 1: net premium income; Stage 2: assets under management; claims paid	1st Stage 2nd Stage Overall TE	0,955 0,784 0,870
Multi-country studies
Diacon (2001)	Multi-country, 6 EU	1999	431	DEA	Total operating expenses, total capital, total technical reserves, total borrowings from creditors	Net earned premiums, total investment income	TE	0.71
Davutyan and Klumpes (2008)	Multi-country, 7 EU	1996–2002	284	DEA	Labour, business services, equity capital	PV of losses incurred, net earned premiums, average invested assets	TE PTE SE	0.267 0.352 0.759
Fenn et al. (2008)	Multi-country, 14 EU	1995–2001	562	SFA	Labour, debt capital, equity capital	Net incurred claims	CE	0.930
Zanghieri (2009)	Multi-country, 15 EU	1997–2006	not given	SFA	Labour, debt capital, equity capital	Claims paid	CE PE	— —
Eling and Luhnen (2010b)	Multi-country, 36 developed and developing incl. SA	2002–2006	3,566	SFA, DEA	Labour and business services, debt capital, equity capital	Claims + additions to reserves, total investments	TE CE	0.810 0.740
Kasman and Turgutlu (2011)	Multi-country, 19 EU, Switzerland, Turkey, Norway	1995–2005	472	SFA	Debt capital, equity capital	Incurred claims	CE	0.882
Bahloul et al. (2013)	Multi-country, 7 EU	2002–2008	125	SFA	Labour and business services, debt capital, equity capital	Losses incurred, reinsurance reserves, reserves for primary insurance contracts, total investments	CE FTP	0.690, 0.678 1.017
Huang and Eling (2013)	Multi-country, BRIC	2000–2008	819	DEA	Labour, debt capital, equity capital	Net written premiums, invested assets	TE PTE SE TFP TEC TC	0.413 0.492 0.850 0.990 0.970 1.020
Bahloul and Bouri (2016)	Multi-country, 7 EU	2002–2008	125	SFA, FE	Labour and business services, debt capital, equity capital	Losses incurred, reinsurance reserves, reserves for primary insurance contracts, total investments	CE (Model 1) CE (Model 2)	0.691 0.679
Doumpos et al. (2018)	Multi-country, 8 EU	2000–2012	458	Metafrontier RE	not given	not given	TE	0.595

Note: DEA = data envelopment analysis, SFA = stochastic frontier analysis, TFA = thick frontier approach, DFA = distribution free approach, RAM = range adjusted measure, SBM = slacks-based measure, TE = technical efficiency, PTE = pure technical efficiency, SE = scale efficiency, AE = allocative efficiency, CE = cost efficiency, PE = profit efficiency, RE = revenue efficiency, CSE = cost scale efficiency, RSE = revenue scale efficiency, TFP = total factor productivity, EC = efficiency change, TC = technical change, PTEC = pure technical efficiency change, PTC = pure technical change, SEC = scale efficiency change, PV = present value. Source: Authors’ compilation

Appendix 2.

Characteristics of studies that examined drivers of non-life insurer performance

Table A2. Summary of studies that examined drivers of non-life insurer performance.

Authors	Countries	Period	No. of insurers	Measure(s) of performance	Methodology	Determinants
Country-specific studies-Developed countries
Fields et al. (1990)	US	1990 & 1988	36	Stock return/value	Cross-sectional regression	SZ
Fecher et al. (1991)	France	1984–1989	164	TE	Comparison using ratios	SZ, OF, RK, RE
BarNiv and McDonald (1992)	US	1974–1988	294	Insolvency	MDA, LPM, exponential generalised beta distribution	SZ
Shelor et al. (1992)	US	1989	79	Stock price/value	GLS, Modified weighted least squares, Cross-sectional regression	RK
Barrese and Nelson (1992)	US	1978–1990	46	Expenses	Regression-Error components	DS
Carroll (1993)	US	1980–1987	—	Profit Margin	2SLS	SZ, CS, RK, MS, ME
Cummins and Weiss (1993)	US	1980–1988	261	CE	SFA	SZ
Fecher et al. (1993)	France	1984–1989	164	TE	Comparison using ratios	SZ, OF, RK, RE
Haley (1993)	US	1930–1989	56	Underwriting margin	Cointegration techniques-VECM, impulse response	ME
Brockett et al. (1994)	US	1991–1992	243	Insolvency	Logistic regression, neural network approach	CS
Cummins et al. (1995)	US	1989–1991	1,596	Insolvency	Logistic regression	SZ, OF, CS
Delhausse et al. (1995)	Belgium	1984–1988	191	TE, SE	Variance Analysis	OF, RK, RE, PM
Kim et al. (1995)	US	1984–1990	125	Insolvency	Event history analysis	DV, RE, AG, GW
Staking and Babbel (1995)	US	1981–1987	25	Tobin’s Q	Regression Analysis	CS, RK
Kazenski et al. (1995)	US	1985–1991	not given	Insolvency	OLS	RK, MS
Browne and Hoyt (1995)	US	1970–1990	3,676	Insolvency (solvency)	Logistic regression	MS, ME
Grace and Hotchkiss (1995)	US	1974–1990	not given	Combined ratio	VECM, variance decomposition, impulse response	ME
Gron (1995)	US	1972–1983	44	Market share	Multiple regression-weighted least squares	ME
Haley (1993)	US	1949–1992	42	Underwriting profit	Cointegration techniques-VECM	ME
Cagle et al. (1996)	US	1968–1991	27	Various ratios e.g. direct written premiums, NWP, expense ratio, ROA, net underwriting income	Ratio analysis	OF
Lee and Urrutia (1996)	US	1980–1991	164	Insolvency	Logit & hazard model	CS, GW, PM
Kramer (1996)	Netherlands	1992	195	Financial solidity-subjective measures (strong, moderate, weak)	Ordered logit model	MS
Berger et al. (1997)	US	1981–1990	472	CE, PE	Regression	SZ, OG, DS, PM
Brockett et al. (1997)	US	1987–1990	44	Insolvency	Logistic regression, neural network approach	CS
Chidambaran et al. (1997)	US	1984–1993	18	Economic loss ratio	2SGLS	RK, DS, MS
Fok et al. (1997)	US	1991	400	Income to surplus ratio	Factor & regression analysis	RK, RE, GW, MS
Tennyson (1997)	US	1992	64	Expense ratio, market share	Regression analysis	GW, ME
Bouzouita and Young (1998)	US	1989–1992	not given	Rating category	Ordered probit model, OLS	SZ, OF, CS, GW, PM
Fields et al. (1998)	US	1993	46	Stock price/value	Event study, cross-sectional regression	SZ, RE
Brockett et al. (1998)	US	1989	1524	Efficiency	Mann-Whitney Statistic	OF, DS
Gius (1998)	US	1976–1990	not given	Premiums	RE	DV, RK, DS, ME
Bajtelsmit and Bouzouita (1998b)	US	1984–1992	not given	Profit margin	RE	DS, MS, ME
Bajtelsmit, and Bouzouita (1998a)	US	1984–1992	not given	Profit margin	Regression-instrumental variable approach	MS, ME
Colquitt et al. (1999)	US	1993–1995	1,400	Cash holdings	OLS, RNE	SZ, OF, CS, RK, GA
Cummins et al. (1999)	US	1990–1992	268	Insolvency	Logistic regression	SZ, OF
Pottier and Sommer (1999)	US	1996	1,678	Rating category	Ordered probit regression	SZ, DV, CS, RK, RE, GW, PM
Cummins et al. (1999)	US	1981–1990	417	TE, CE	DEA, Cross-frontier	OF
Born (2001)	US	1984–1991	1,440	ROE	Quantile regression, OLS	SZ, OF, DV, CS, RK, AG, PM, MS, ME
Garven and Grace (2001)	US	1986–1996	500	Efficiency score	Tobit	SZ, OF, GA, DS, PM
Cummins and Nini (2002)	US	1993–1998	645	TE, AE, CE, RE, ROE	OLS	SZ, OF, DV, CS, RK, RE, GW, PM
Pottier and Sommer (2002)	US	1996–1998	1,779	Insolvency	Logistic regression	SZ, OF
Worthington and Hurley (2002)	Australia	1998	46	PTE, SE, CE, AE	Tobit	SZ, OF, DV
Lai and Limpaphayom (2003)	Japan	1983–1994	24	ROA, expense ratio, commissions, loss ratio, investment income, free cash flow	Pooled regression, 2SLS	SZ, OF, CS, RK
Li and Greenhood (2004)	Canada	1993–1998	137	ROA	RE	SZ, DV, RK, GW, MS
Shiu (2004)	UK	1986–1999	137	Investment yield, %ge change in shareholders’ funds, return on shareholders’ funds	OLS, RNE, FE	SZ, CS, RE, GW, ME
Brockett et al. (2004)	US	1989	1,524	Efficiency	Mann-Whitney Statistic	OF, DS
Salcedo-Sanz et al. (2004)	Spain	1983–1994	72	Insolvency	Support Vector Machines-based methods	CS
Segovia-Vargas et al. (2004)	Spain	1983–1994	72	Insolvency	Support Vector Machines, Genetic Algorithm & a Simulated Annealing	CS
Gaver and Pottier (2005)	US	1997	80	Rating categories	Ordered probit regression	SZ, CS, RK
Brockett et al. (2005)	US	1989	1,524	Efficiency	Mann-Whitney Statistic	OF, DS
Choi and Weiss (2005)	US	1992–1998	4,777	Underwriting profit-margin	GMM, Heteroskedastic 2SLS	OF, RE, GA, GW, DS, PM, MS
Jeng and Lai (2005)	Japan	1985–1994	19	TE, CE	RAM-DEA, Cross-frontier	OF
Díaz-Martínez et al. (2005)	Spain	1983–1994	72	Insolvency	Non-parametric machine learning techniques, See5 & Rough Set	CS
Hwang and Kao (2006)	Taiwan	2001–2002	24	Marketing & Profitability Eff	Network DEA	SZ, OF, MS
Dionne et al. (2007)	US	1995–2003	369	CE	SUR	SZ, OF, DV, CS, RK, RE, GA, DS, PM
Sharpe and Stadnik (2007)	Australia	1999–2001	68	Financial distress (1 & 2)	Logit regression	SZ, RE, GW, PM
Scordis and Barrese (2007)	US	1994–2004	41	Holding period return, market value, cash dividends	RE, FE	CS, RK, RE
Cummins and Xie (2008)	US	1994–2003	1,550	TFP, PTE, AE, SE, CE, RE	OLS	SZ, OF, DV, CS, GA, PM
Elango et al. (2008)	US	1994–2002	787	ROE, ROA, RAROE, RAROA	FE	SZ, OF, DV, CS, RK, GA, DS, PM
Liebenberg and Sommer (2008)	US	1995–2004	629	ROA, ROE, Tobin’s Q	OLS, 2SLS, Heckman	SZ, OF, DV, CS, RK, GA, PM, MS
Ma and Elango (2008)	US	1992–2000	41	RAROA	FE	SZ, DV
Weiss and Choi (2008)	US	1992–1998	1,119	CXE, RXE, CSE, underwriting profit margin	2SLS, Two-stage Tobit	SZ, OF, DV, RE, GA, GW, DS, PM, MS, ME
Boubakri et al. (2008)	US	1995–2000	177	Market returns	Regression	OF, CG, FSID
Ma and Elango (2008)	US	1992–2000	41	RAROA	FE	OF, RK, RE, GA, DS
Elango (2009)	US	2000–2004	847	ROA	Logistic regression	SZ, OF, DV, RK, GA, AG
Kleffner and Lee (2009)	Canada	1980–2004	159	Insolvency	Logistic regression	SZ, OF, CS, RK, GA, GW
Luhnen (2009)	Germany	1995–2006	229	TE, CE	Truncated regression	SZ, OF, DV, CS, GW, DS
Park et al. (2009)	US	1990–2001	370	CE, RE	Pooled OLS, Two-way FE, Two-way RE	SZ, OF, CS, GA, DS, PM
Barth and Eckles (2009)	US	1998–2005	1,545	Loss ratio	FE	GW
Chen et al. (2010)	US	1985–2008	960	ROA, ROE, Combined ratio (CR), Economic loss ratio (ELR)	Panel regression	SZ, OF, DV, CS, RE, GA, MS
Cummins et al. (2010)	US	1993–2006	718	TE, SE, CE, AE, RE, PE	Multiple regression	SZ, OF, DV, CS, RK, RE, DS, PM
Lei and Schmit (2010)	US	1995–2006	102	RAROA, ROE	RE, FE	SZ, OF, DV, CS, GA
Parente et al. (2010)	US	1992–2000	not given	Market share growth, profitability	OLS	SZ, DV, DS, PM
Yin (2010)	Japan	1989–1995 & 2004–2006	21	Expense ratio	FE, RNE	SZ, OF, RE, AG, PM
Choi (2010)	US	1992–2001	823	Growth rate	Heckman’s 2 stage regression	OF, DV, RE, GA, AG, DS
Xie (2010)	US	1994–2005	312	TE, AE, SE, CE, RE, ROA, loss ratio, expense ratio, premium growth, other financials	FE	OF
Wang (2010)	US	1995–2006	3,088	Rating grades	Least squares regressions	GW, ME
Choi and Elyasiani (2011)	US	1992–1998	953	CXE, CSE, RXE, RSE, ROA, inverse of loss ratio,	GMM	SZ, OF, DV, CS, RE, GA, DS, PM
Eckles and Pottier (2011)	US	1996–2000	1,158	Rating grades	Ordered logit model	SZ, OF, CS
He et al. (2011)	US	1995–2006	557	CE, RE, TFP, TC, TEC	Multivariate regression	SZ, OF, DV, CS, CG, PM
Huang et al. (2011)	US	2000–2007	28	TE, CE	Tobit	SZ, CG
Shi and Zhang (2011)	US	2001–2006	1,527	CE	SFA-Copula regression	SZ, OF
Shim (2011a)	US	1990–2004	190	PTE, SE, AE, CE, RE	FE	SZ, OF, DV, CS, GA, DS, PM
Shim (2011b)	US	1989–2004	190	ROE, ROA, RAROA, RAROE, Z-score	Regression lagged structure model	SZ, OF, DV, CS, RK, GA, DS, PM
Bikker and Gorter (2011)	Netherlands	1995–2005	195	CE, SE	OLS	OF, DV
Chen et al. (2011)	US	1990–2001	540	TE, AE, CE, TFP	OLS	OF
Lin et al. (2011)	US	2002–2004	1,657	CE	SFA	CS, RE, GW
Cheng and Weiss (2012)	US	1994–2005	1,845	Insolvency	Logistic pooled regression	SZ, OF, RK, MS, ME
Cummins et al. (2012)	US	1993–2009	1,260	ROA, ROE, CE, RE, PE	2 way OLS FE	SZ, OF, DV, CS, RK, RE, GA, PM
Huang et al. (2012)	Japan	1992–2005	22	TE, AE, CE	Truncated regression	SZ, OF, DV
Leverty and Grace (2012)	US	1989–2000	1,018	CE, RE, insolvency, cost of insolvency	Heckman 2-stage regression, logistic regression, Tobit	SZ, OF, CG
Lin et al. (2012)	US	2000–2007	63	Tobin’s Q, ROA Underwriting ROA	Treatment-effect model	SZ, DV, CS, RK, RE, GW, PM
Andersson et al. (2013)	Sweden	1903–1939	102	Investment returns	GMM	SZ, OF, CS, RK
Choi et al. (2013)	US	1998–2007	1,717	Liquidity creation	Regression	SZ, OF, DV, CS, RE, GA, PM
Cummins and Xie (2013)	US	1993–2009	781	PTE change, SE change, CE change, RE change, TFP change, RTS	FE, multinomial logit	SZ, OF, DV, CS, GA, DS, PM
Ma et al. (2013a)	US	1993–2008	1,116	CE, RE, ROE, ROA	Regression	SZ, OF, DV, CS, DS, PM
Kelly et al. (2013)	Canada	1991–2008	6	Loss ratio (LR)	3SLS	PM, MS
Chang and Tsai (2014)	US	2006–2010	1,423	Liquidity	OLS, Quantile regression	SZ, OF, DV, CS, RK, RE, GA, GW, ME
Chen et al. (2014)	US	2000–2011	1,479	RAROA, RAROE, CR	FE	SZ, OF, DV, CS, GA, PM
Emm (2014)	US	1988–2001	630	CE	Tobit	SZ, OF, DV, DS
Park and Xie (2014)	US	2003–2009	1,041	Rating, rating downgrade, Stock price/return	Ordered Probit regression, 2SLS, Panel Regression	SZ, OF, CS, RK, RE, GA, AG, PM
Hsu et al. (2015)	US	1997–2002	242	Cash holdings	FE	SZ, CS, RK, CG, AG
Park and Park (2015)	South Korea	2006–2011	17	TE, PTE, ROA	OLS	SZ, OF, DV, CS, GA, DS, PM, MS
Rauch and Wende (2015)	Germany	2004–2011	108	Solvency ratio	OLS, Logistic regression	SZ, OF, DV, CS, RK, GW
Zhang and Nielson (2015)	US	1996–2006	1,422	Insolvency	Logistic regression	DV, CS, RK, GA, ME
Adams and Jiang (2016)	UK	1999–2012	77	PM, ROA, ROE, solvency, CR	2SLS	SZ, OF, DV, RE, CG, AG
Ai et al. (2016)	US	2006–2013	58	ROA, Tobin’s Q	2SLS, Pooled OLS	SZ, OF, DV, CS, RK, GA, PM
Altuntas and Gößmann (2016)	Germany	1999–2011	32	ROA	OLS	SZ, OF, DV, CS, RK, RE, AG, PM
Altunta et al. (2016)	Germany	2004–2012	92	ROA, ROE, RAROA, RAROE	OLS	SZ, OF, DV, CS, RK, GA
Biener et al. (2016)	Switzerland	1997–2013	51	TE, CE, RE	Truncated regression	SZ, OF, DV, CS, AG, GW
Chang and Jeng (2016)	US	1994–2006	2,180	Liquidity	2SLS	SZ, OF, DV, CS, RE, GA, GW
Cummins and Xie (2016)	US	1993–2011	768	CE, RE, PE, TFP	Two-way FE, One-way FE	SZ, OF, DV, CS, DS, PM
Liu et al. (2016)	UK	1994–2011	127	Liquidity	2SLS	SZ, OF, DV, CS, RE, PM
Adams and Jiang (2017)	UK	1999–2012	77	PM	OLS	SZ, OF, RE, CG, AG
Caporale et al. (2017)	UK	1985–2014	167	Default probabilities	Reduced form regression	SZ, OF, DV, CS, RE, GW, ME
Che et al. (2017)	US	1997–2013	639	Investment return	OLS, Heckman model & 2SLS	SZ, OF, DV, CS, RE, GA, PM, MS
Lambalk and de Graaf (2017)	Netherlands	2008–2012	25	Profit, CR, LR, OE, IR	Regression	OF, DV
Shim (2017a)	US	1992–2010	738	Z-score	OLS, 2SLS	SZ, OF, DV, CS, RK, RE, GA, GW, PM, MS, ME
Shim (2017b)	US	1996–2010	1,003	RAROA	Quantile regression	SZ, OF, DV, CS, RK, RE, GA, AG, GW, DS, MS
Ames et al. (2018)	US	2007–2013	396	Rating category, ROE	Ordered logistic regression	SZ, CS, CG
Adams et al. (2019)	UK	1985–2010	145	CR, LR, ELR, ROA	GLS	SZ,DV, CS, RE, PM, ME
Lei (2019)	US	1993–2008	1,505	RAROE	Two-way FE regression	SZ, OF, DV, CS, RK, RE, GA, PM
Liebenberg and Lin (2019)	US	2004–2013	747	ROA	OLS, 2SLS, Heckman’s model	SZ, OF, DV, CS, RK, GA, MS
Maichel-Guggemoos and Wagner (2019)	Germany	2001–2016	91	Profitability, growth, safety	Regression analysis	SZ, OF, GW, MS
Zhang et al. (2019)	US	1992–2011	28	Sharpe ratio, Treynor ratio	OLS, Time-fixed OLS	SZ, CS, GW
Shiu (2020)	UK	1994–2011	129	ROA	Three-equation structural model	SZ, CS, RE, RK, DV
Eling et al. (2020)	Germany	1954–2016	95	TFP	Multiple regression	SZ, OF, MS, ME, DV
Adams and Baker (2021)	UK	1999–2013	73	ROE, Solvency	RE, FE	SZ, OF, RE, CG
Griffith and Liebenberg (2021)	US	2002–2015	1,557	ROA, ROE, loss ratio, expense ratio	OLS	RE
Park et al. (2021)	US	2001–2009	1,321	ROA, ROE, investment return on equity (IRE); underwriting return on equity (URE)	OLS, 2SLS	SZ, OF, CS, RE, DV, GA
Country-specific studies-Emerging countries
Leverty et al. (2004)	China	1995–2002	14	TE, PTE, SE, productivity	Weighted Tobit, WLS regression	SZ, OF, MS
Wang et al. (2007)	Taiwan	2000–2002	16	TE, AE, CE, ROA	Regression	SZ, DV, RE, CG, PM
Kasman and Turgutlu (2009)	Turkey	2000–2005	22	TFP	TFP, EC, TC, PTEC SEC	OF
Hsu and Petchsakulwong (2010)	Thailand	2000–2007	17	TE, AE, CE, RE	Truncated regression	SZ, CG
Shiu (2010)	Taiwan	2001–2003	24	Solvency ratio	Heckman 2 Stage regression, OLS, RE	SZ, DV, CS, RK, RE, PM
Foong and Idris (2012)	Malaysia	2006–2009	24	ROE	Regression	SZ, DV, CS
Naini and Nouralizadeh (2012)	Iran	2003–2010	17	Overall efficiency	Generalised Estimating Equations (GEE)	SZ, OF, CS, MS
Lee and Lee (2012)	Taiwan	1999–2009	15	ROA	Two-equation simultaneous model, 2SLS	SZ, OF, DV, CS, RK, RE, GA, GW, MS
Chen et al. (2014)	Malaysia	2008–2011	16	Malmquist Productivity Index	OLS	SZ, OF, CS, MS
Jeng and Yang (2014)	China	2000–2006	13	ROA, Growth	Two-equation simultaneous model, GMM	SZ, OF, DV, CS, RE, AG, GW, MS, ME
Lee (2014)	Taiwan	1999–2009	15	ROA, operating ratio	RNE	SZ, DV, CS, RK, RE, GA, GW, MS, ME
Kweh et al. (2014)	China	2006–2010	32	Operating efficiency	OLS	SZ, CS
Iqbal and Rehman (2014)	Pakistan	2002–2011	22	Loss ratio, Expense ratio	Pooled OLS, FE, RNE	RE
Yaisawarng et al. (2014)	Thailand	2000–2007	46	CE, TFP	SFA	PM
Alhassan and Biekpe (2015)	South Africa	2007–2012	70	TE, PTE, RTS	Truncated, Logistic regression	SZ, DV, CS, RE, AG
Alhassan et al. (2015)	Ghana	2007–2011	22	ROA	PCSE, FE	SZ, CS, RK, MS, ME
Sandada et al. (2015)	Zimbabwe	n/a	n/a	Non-financial measures	Regression	CG
Öner Kaya (2015)	Turkey	2006–2013	24	Technical profitability ratio, sales profitability ratio	FE	SZ, CS, RK, RE, AG, GW, PM
Alhassan and Biekpe (2016b)	South Africa	2007–2012	66	CE, PE	FE	SZ, DV, CS, RK, RE, AG, MS
Asare et al. (2017)	Ghana	2007–2011	22	ROA, UPM	OLS-PCSE	SZ, CS, RK
Lee (2017)	Taiwan	1999–2010	15	RAROA, RAROE	FE	SZ, DV, CS, RE, GA, GW, PM, MS
Sinha (2017)	India	2013–2014	16	TE	Truncated regression	OF, CS
Kuo et al. (2017)	Malaysia	2008–2013	16	Marketing efficiency, profitability efficiency	Truncated regression	CG
Ishtiaq (2017)	Pakistan	2009–2013	40	ROA	Multiple regression	ME
Ferro and León (2018)	Argentina	2009–2014	69	TE	Average scores	SZ
Alhassan and Biekpe (2018)	South Africa	2007–2012	37	Z-score	ISUR, OLS-PCSE, QR, GMM	SZ, OF, DV, CS, RE, MS
Barua et al. (2018)	Bangladesh	2000–2014	16	ROA, ROE	FE, RNE, Pooled OLS, PMG	SZ, CS, RK, AG, GW
Hasan et al. (2018)	Bangladesh	2009–2015	32	ROA, ROE	FE, RNE	SZ, CS, RK, AG, ME
Ilyas and Rajasekaran (2019)	India	2005–2016	14	TE, PTE, SE, CE, AE	Truncated regression, Mann-Whitney U test, Wilcoxon rank-sum tests	SZ, OF, DV, CS, RK, RE, AG
Camino-Mogro and Bermúdez-Barrezueta (2019)	Ecuador	2001–2017	25	ROA, investment income, profit after tax	PCSE	OF, DV, CS, RK, MS, ME, FSID
Ilyas and Rajasekaran (2020)	India	2005–2016	15	TFP	Mann-Whitney U test	OF
Li et al. (2021)	China	2011–2017	47	TE, TFP	DEA, Fuzzy set qualitative comparative analysis	OF, CS, RE, RK
Wu and Li (2021)	China	2009–2015	53	Solvency	Step-wise regression	RE
Andoh and Yamoah (2021)	Ghana	2008–2018	20	ROA	RE	SZ, RE, ME
Zhao et al. (2021)	China	2013–2017	53	PE, AE, TE	DEA, Tobit	SZ, DV
Ilyas and Rajasekaran (2021)	India	2005–2016	15	TFP	DEA, Bootstrap truncated regression	SZ, DV, RE
Sinha (2021)	India	2010–2018	15	1st Stage, 2nd Stage, Overall TE	Network DEA, censored & probit regression	OF
Multi-country studies
Diacon (2001)	Multi-country, 6 EU	1999	431	TE	Tobit	SZ, OF, CS, RE, PM, MS
Florez-Lopez (2007)	Multi-country, 14 EU	1997–1999	257	Rating grades	Multivariate models-MDA, logit, C4.5, CART Gini	SZ
Davutyan and Klumpes (2008)	Multi-country, 7 EU	1996–2002	284	TE, PTE, SE	FE	SZ, OF, DV, CS, GA, ME, FSID
Fenn et al. (2008)	Multi-country, 14 EU	1995–2001	562	CE	SFA	SZ, MS
Oetzel and Banerjee (2008)	Multi-country, 31 Emerging	1998–2003	307	ROE, ROA, Return on sales	GLS RE	SZ, OF, DV, GA, ME, FSID
Pope and Ma (2008)	Multi-country, 23 developed & developing	1996–2003	15	UPM	2-way FE	PM, MS, ME
Zanghieri (2009)	Multi-country, 15 EU	1997–2006	not given	CE, PE	SFA (translog cost function)	SZ, DV, PM, MS, FSID
Eling and Luhnen (2010b)	Multi-country, 36 developed & developing	2002–2006	3,566	TE, CE	Conditional Mean Approach	SZ, OF, CS
Kasman and Turgutlu (2011)	Multi-country, 19 EU, Switzerland, Turkey, Norway	1995–2005	472	CE	SFA (translog cost function)	SZ
Berry-Stolzle et al. (2011)	Multi-country, 12 countries	2003–2007	319	Profit margin	OLS, 2SLS	OF, RE, GA, GW, PM, MS
Njegomir and Stojić (2011)	Multi-country, 11 Eastern Europe	2004–2008	not given	Profit margin	FE	PM, MS, ME
Doumpos et al. (2012)	Multi-country, 91 developed & developing	2005–2009	1,836	Multicriteria score	FE	SZ, RE, ME, FSID
Fields et al. (2012)	Multi-country, 66 developed & developing	1992–2006	312	UPM	OLS	SZ, CS, RK, RE, GW, ME, FSID
Huang and Eling (2013)	Multi-country, BRIC	2000–2008	819	TE, PTE, SE	Truncated regression	SZ, CS, RK, CG
Ma et al. (2013)	Multi-country, 4 Asian	2003–2008	36	UPM	GMM	SZ, DV, RE, AG, MS, FSID
Moro and Anderloni (2014)	Multi-country, 9 EU	2004–2012	198	ROA, ROE	RE	SZ, OF, DV, CS, RK, RE, GW, FSID
Hsieh et al. (2015)	Multi-country, 62 developed & developing	1995–2009	43	ROA, ROE, Tobin’s Q, Z-score	Dynamic Panel GMM	SZ, DV, CS, RE, GW, ME, FSID
Elango and Wieland (2015)	Multi-country, 32 countries	2001–2007	2,293	ROA	Hierarchical linear modelling techniques (HLM)	ME, FSID
Biener et al. (2016)	Multi-country, various	2003–2013	283	RAROA, RAROE, CE	LSDV, Tobit, truncated regression	SZ, DV, CS, RE, GW, ME, FSID
Bahloul and Bouri (2016)	Multi-country, 7 EU	2002–2008	125	CE	FE	CG, MS, ME, FSID
Altuntas and Rauch (2017)	Multi-country, 29 developed & emerging	2004–2012	1,600	Z-score, RAROE	OLS	SZ, OF, CS, RK, RE, GA, MS, ME
Doumpos et al. (2018)	Multi-country, 8 EU	2000–2012	458	Metafrontier efficiency	RE	SZ, OF, DV, RE, MS, ME, FSID
Eling and Jia (2018)	Multi-country, 16 EU	2006–2013	770	Probability of failure	Logistic regression	SZ, OF, CS, RK, GA, GW, ME, FSID
Olarewaju and Msomi (2021)	SADC	2008–2019	58	ROA	2SLS,FE, RE, 2-step system GMM	SZ, CS, RK

Notes: TE = technical efficiency, PTE = pure technical efficiency, SE = scale efficiency, AE = allocative efficiency, RE = revenue efficiency, PE = profit efficiency, TFP = total factor productivity, TC = technical change, TEC = technical efficiency change, PTEC = pure technical efficiency change, CXE = cost X-efficiency, RXE = revenue X-efficiency, CSE = cost scale efficiency, RSE = revenue scale efficiency, RTS = returns to scale, ROA = return on assets, ROE = return on equity, RAROA = risk adjusted return on assets, RAROE = risk adjusted return on equity, CR = combined ratio, LR = loss ratio, ELR = economic loss ratio, UPM = underwriting profit margin, OE = operating expenses, IR = investment return, OLS = ordinary least squares, FE = fixed effects, RNE = random effects, PM = profit margin, GMM = generalised method of moment, GLS = generalised least squares, PCSE = panel corrected standard errors, 2SLS = two-stage least squares, LSDV = least squares dummy variable, SFA = stochastic frontier approach, MDA = multiple discriminant analysis, QR = quantile regression, LPM = linear probability model, NPDM = non-parametric discriminant model, MPG = pooled mean group, SZ = size, OF = organisational form, DV = diversification, CS = capital structure, RK = risk, RE = reinsurance, GW = growth, GA = group affiliation, DS = distribution system, CG = corporate governance, MS = market structure, ME = macroeconomic, FSID = financial sector and institutional development.

Source: Authors’ compilation.