The quality of risk disclosure and Chief Risk Officers : a longitudinal study : evidence from European insurance companies

(1)

Amsterdam Business School

The quality of risk disclosure and Chief Risk Officers:

A longitudinal study

Evidence from European insurance companies

Name: Victor Thijssen

Student number: 0297550

Date: 21-06-2014

1st_{supervisor :} _{Prof. dr. D.M. Swagerman} 2nd_{supervisor :} _{Dr. ir. S.P. van Triest}

MSc Accountancy & Control, variant Control Amsterdam Business School

(2)

2 Abstract

Corporate risk reporting is a very important aspect in the field of accounting and investment practice.

This thesis investigates if the quality of risk disclosure of European insurance companies with an appointed Chief Risk Officer increased in the period 2005-2013.

First, there has never been done previous research specific to companies with an appointed CRO and the quality of risk reporting.

Second, little definitive research has been done at the academic or practitioner level on risk reporting by insurance companies (Klumpes et al, 2013). Finally, CRO’s in this industry are widely introduced.

This study therefore chose to investigate risk reporting quality of this specific industry. First, there will be a selection of companies which have an appointed CRO.

The selection is made of European listed companies which are member of the CRO forum, an institute formed in 2004 to advance risk management practice in the insurance industry.

The quality will be measured from the risk management sectors of annual reports from 2005, 2009 and 2013. The year 2005 is chosen because it is short after the CRO forum was formed.

From the insurance companies which were member in 2005 there can be investigated how the quality of risk reporting evolved in the period 2005-2013.

The quality of risk reporting is measured by the method of word count and other qualitative measures since merely word count is not sufficient to measure risk reporting quality.

Results show that three of five hypotheses are supported. Regarding the concerned gravities of the observed factors one can conclude that in general there is a significant increase in risk disclosure quality of European insurance companies with an appointed CRO.

Further research has to be done to investigate how risk disclosure quality of European insurance companies without an appointed CRO evolved in the period 2005-2013. This can contribute to academic science, namely to the fact if a CRO can or should be included as a possible new variable for risk disclosure quality.

(3)

3 Table of contents

Introduction ... 4

1. Literature review ... 6

1.1 Relevance ... 6

1.2 Corporate Governance Theories ... 6

2. Enterprise Risk Management ... 9

2.1 Evolution and role of the Chief Risk Officer ... 10

2.2 Studies regarding ERM implementation ... 12

2.3 Quality of risk reporting ... 13

3. Hypotheses development ... 17

4. Empirical methodology and results ... 20

4.1 Sample Selection ... 20

4.2 Model of Analysis ... 21

4.3 Method of Analysis ... 22

5. Empirical Results ... 23

5.1 Overall data analysis and overview ... 23

5.1.1 Quantity of words ... 24

5.1.2 Number of risks ... 25

5.1.3 Number of controls ... 25

5.1.4 Number of quantitative substantive disclosure ... 27

5.1.5 Number of forward looking risk disclosure ... 27

5.1.6. Hypothesis testing ... 29

6. Conclusion , limitations and further research ... 37 Literature list ...

(4)

4 Introduction

Corporate governance is a main topic in various studies and current events.

According to the OECD Principles of Corporate Governance ''Corporate governance involves a set of relationships between a company’s management, its board, its shareholders and other stakeholders. Corporate governance also provides the structure through which the objectives of the company are set, and the means of attaining those objectives and monitoring performance are determined."

The financial scandals from the past affecting major firms, such as Enron, WorldCom, Parmalat and Arthur Andersen, and the resulting loss of confidence by the investing public in the stock market have led to dramatic declines in share prices and substantial financial losses to millions of individual investors. Both the public and the experts have identified failed corporate governance as a principal cause of these scandals. To effectively fulfill the demands of their stakeholders, listed companies have been

improving the communication of their long-term value-generation capabilities by increasing the amount of information disclosed with regard to the risks faced and their expected impact on future profits. However, disclosing current financial risks will not provide sufficient information about the financial status of a company because financial performance is also affected (and even more so) by strategic and operating risks (EIU- MMC, 2001).

Monitoring of application of corporate governance in firms all over the world is made possible by introducing corporate governance codes.

Many countries have introduced their own corporate governance codes where the concerning listed companies do have to comply with.

Part of Corporate Governance is reporting about risk. In the US, risk reporting is mainly driven by the Sarbanes-Oxley Act which was introduced in 2002.

In recent years, scholars have shown increasing interest in risk disclosure.

However, risk disclosure is still one of the most ambiguous and unexplored areas of corporate disclosure. In particular, we know too little about quality of risk disclosure (Miihkinen, 2012).

Past studies regarding risk reporting quality find mixed results about the quality of the disclosure. There is a lot of discussion about the way to measure quality of (risk) reporting and some even doubt if there is a simple and objective one. Most of the studies have tended to measure quality by quantitative means (word count, analyzing sentences). In later years attempts have been made to look beyond this type of quantitative means whereby several frameworks have been developed.

(5)

5

There has never been done research to the fact if a CRO has a positive influence on the quality of risk reporting.

Nowadays many companies have contracted a CRO to be the main responsible for risk management and, as a part of it, risk reporting. Some major companies are even a member of institutes like the CRO forum, which is a group of professional risk managers from the insurance industry that focuses on developing and promoting industry best practices in risk management.

The main goal of this research is to analyze if risk disclosure of European insurance companies with an appointed Chief Risk Officer increased in the period 2005-2013.

The main question of this research is:

Does the quality of risk disclosure increased within European insurance companies with an appointed Chief Risk Officer?

In the next chapter, this study will outline some theories relating corporate governance and describe the concept of ERM and the evolution and (supposed) role of a CRO.

Next there will be an overview of various studies that have been done regarding ERM and risk reporting (quality).

In chapter 3 the hypothesis is formed and in chapter 4 the sample selection is outlined and the results of the research are presented.

(6)

6 1. Literature Review

1.1 Relevance

The topic of risk reporting is still an actual one, which is evolving year by year.

This study contributes to the previous research by extending prior work on risk disclosure quality and tries to examine whether the presence of a CRO in a firm has an influence on the quality of risk reporting. This hasn’t been researched in earlier papers. Much research is focused on measuring risk reporting quality by looking at the quantity of disclosure whereas this thesis is applying an combination of methods.

1.2 Corporate Governance Theories

Risk reporting disclosure is influenced by several corporate governance theories. Corporate governance includes all types of firms and its definitions could extend to cover all of the economic and non-economic activities. There are various fundamental theories underlining corporate governance which range from the agency theory and expanded into stewardship theory, stakeholder theory, resource dependency theory, transaction cost theory, political theory and ethics related theories such as business ethics theory, virtue ethics theory, feminists ethics theory, discourse theory and postmodernism ethics theory. (Masdoor, 2011).

(7)

7 Stakeholder Theory

Stakeholder theory stems from the positive theories.

A Positive theory can be described as a theory, which seeks to explain and predict (Deegan 2005). Stakeholder theory was embedded in the management discipline in 1970 and gradually developed by Freeman (1984) incorporating corporate accountability to a broad range of stakeholders. Stakeholder theory can be defined as “any group or individual who can affect or is affected by the achievement of the organization’s objectives”. Stakeholder theorists suggest that managers in organizations have a network of relationships to serve – this include the suppliers, employees and business partners. (Abdullah, Valentine, 2009)

Risk reporting is related to the stakeholders’ theory, because stakeholders have influence over the content of the report. Besides the SOx (2002) and Dodd Frank Act (2010) almost every country has his own corporate governance code describing the aspects a risk report should contain.

Because a CRO is specially appointed for risk reporting he or she should have more expertise and time for stakeholder management. This should result in better stakeholder management and thus better risk reporting quality.

(8)

8 Stewardship Theory

Stewardship theory has its roots from psychology and sociology and is defined by Davis, Schoorman &

Donaldson (1997) as “a steward protects and maximizes shareholders wealth through firm performance, because by so doing, the steward’s utility functions are maximized”. In this perspective, stewards are company

executives and managers working for the shareholders, protecting and making profits for the shareholders. The stewardship perspective suggests that stewards are satisfied and motivated when organizational success is attained.

Based on the stewardship theory the person in charge of risk management should do his utmost best to ensure that risk reporting is from the highest possible quality.

(9)

9 2. Enterprise Risk Management

Enterprise risk management (ERM) has been the topic of increased media attention in recent years. figure 1 (Liebenberg, Hoyt, 2003)

Various definitions of ERM exist.

Lam (2000), one of the pioneers in this topic , defined ERM as “an integrated framework for managing credit risk, market risk, operational risk, economic capital, and risk transfer in order to maximize firm value”. He stated ERM should consist of seven components which include corporate governance, line management, portfolio management, risk transfer, risk analytics, data and technology resources and stakeholder management.

ERM benefits firms by decreasing earnings and stock-price volatility, reducing external capital costs, increasing capital efficiency, and creating synergies between different risk management activities. (Liebenberg, Hoyt, 2003) and provides a significant source of competitive advantage for those who can demonstrate a strong ERM capability and discipline. (Beasley et al, 2005). It is important for companies to have a well functioning ERM system because it can generate substantial benefits. CRO's likely have the necessary communication skills that are required to promote the importance of ERM to the board and to inform outside stakeholders of the firm's risk profile. (Liebenberg, Hoyt, 2003)

(10)

10 2.1 Evolution and role of the Chief Risk Officer

Emergence of the CRO

The emergence of the chief risk officer goes back to august of 1993.

This particular month and year GE Capital gave James Lam a job that brought together management of credit risk, market risk and liquidity risk, and he coined the term chief risk officer for his business cards. (Economist, 2005)

James Lam’s appointment was before the era of severe financial and corporate governance scandals occurred like Enron, Worldcom, Barings Bank and others.

But even at the time of Lam’s appointment the environment in which financial institutions were operating was becoming broader and more complex.

After Lam’s appointment many would follow.

The role of CRO is now widely adopted in risk intensive businesses like financial institutions and energy firms (Lam, 2000)

Reasons of appointing a CRO

(11)

11

Increasing trust between the business and its shareholders is also pointed out by Linsley and Shrives as an important potential benefit :

’’The most important potential benefit arising from improved risk disclosures by firms is a

reduction in the cost of capital ’’.That is, if risks are disclosed providers of capital may remove a part of the premium that is incorporated in the cost of capital to cover for uncertainty concerning the firm’s risk position’’. (Linsley and Shrives, 2006). In times of financial crisises this is an important benefit due to the relatively higher interest rates on loans and bonds during such a period.

Role of the CRO

In his study Lee (2000) describes his vision of how the role of the CRO emerged and points out that a CRO has the following goals :

● To create a risk-aware culture

● To develop a center of excellence for rnanaging risk, statred by highly competent and highly valued individual risk managers

● To create an efficient approach for financing risks

● To communicate to stakeholders and be an advisor to other executives and managers

The last goal is important specifically having to do with risk reporting (risk paragraph in annual report , annual risk report etc).

CRO and stakeholder management

One mentioned aspect of Enterprise Risk Management in both Lam and Lee’s article is stakeholder management. Risk reporting is an important part of it.

’’Risk Management is not only an internal management process, but it should be used to improve risk transparency to key stakeholders.Equity analysts and rating agencies need risk information to develop their investment and credit opinions. In communicating and reporting to these key stakeholders, an important

objective for management is to assure them that appropriate risk management strategies are in effect’’. This is in line with Linsley and Shrives (2006) assumptions that improved risk disclosure can result in a lower risk premium.

Summarizing, a good ERM system could generate substantial benefits.

(12)

12

management and risk reporting as a part of it.

As a CRO being main responsible for ERM this study hypotheses that a CRO makes a positive difference regarding ERM and thus risk management and risk disclosure practices.

2.2 Studies regarding ERM implementation

Various studies exist regarding ERM implementation and possible factors influencing it.

Yazid et al. (2012) attempted to define the determining factors for companies implementing ERM. Their literature review resulted in several factors that could possibly influence any company/organization to eventually implement ERM including the appointment of a CRO.

They found a positive relationship between a CRO with ERM engagement.

Beasley et al. (2005) examined the factors associated with the stage of ERM implementation at a variety of US and international organizations and formulated the following seven research questions:

 Is the presence of a Chief Risk Oﬃcer positively associated with an enterprise's stage of ERM deployment?

 Is a higher percentage of board of director members who are independent positively associated with an enterprise’s stage of ERM deployment?

 Are explicit calls from the CEO or CFO for internal audit involvement in ERM positively associated with an enterprise’s stage of ERM deployment?

 Is the presence of a Big Four auditor positively associated with an enterprise’s stage of ERM deployment?

 Are larger ﬁrms more likely to have further-developed ERM deployments?

 Are entities in the banking, education, or insurance industries more likely to have further-developed ERM deployments?

 Are non-US enterprises more likely to have further-developed ERM deployments?

They surveyed chief audit executives to obtain data related to ERM deployments and other organizational characteristics.

They received 175 survey responses, a rate of 10.3% and their final sample consisted of 123 organisations. To address their seven research questions, they used an ordinal logistic regression model.

(13)

13

The results showed a positive relationship between the presence of a Chief Risk Officer and the extent of ERM deployment. This finding suggests that the presence of a ‘‘risk champion’’ among the senior management team significantly increases the entity's stage of ERM deployment

2.3 Quality of risk reporting

There are number of studies trying to measure the quality of risk disclosures.

Several methods have been developed to make it possible to convert the information out of risk reports into metrics, which are more comparable and easier to assign a measure of quality.

A vast literature is related to the quantity of corporate disclosures.

The quantity of risk reporting can be defined by the number of words, the number of sentences, the number of pages and the percentage of pages. (Leitner-Hanetseder, 2012)

Though, Beretta & Bozzolan (2004) contend in their paper that the quantity of disclosure is not a satisfactory proxy for the quality of disclosure.

In their proposed framework, quality of disclosure depends both on the quantity of information disclosed and on the richness offered by additional information.

While the quantity of disclosure has been discussed in previous literature, little attention has been paid, until now, to the richness of the information in quality. (Beretta & Bozzolan, 2004).

With their paper they intend to fill that gap.

To measure the quality of disclosure they implemented four dimensions to assign a score to the voluntary disclosure; the first dimension being the quantity of disclosure and the second, density of disclosure. The relative quantity (RQT) is known to be influenced by size and industry. Density (DEN) they define as the ratio between the number of sentences in which risk information is provided over the total number of sentences included in the Management Discussion and Analysis or Operating and Financial Review. The disclosure of expected impacts of considered risks and the orientation of management were split up into; first dimension was depth and the second, outlook profile. The depth (DPT) dimension is the content of risk information disclosed regarding the expected economic impact of the identified risks upon future performance. The fourth and last dimension is outlook profile (OPR). This dimension concerns how management communicates the approach it adopted to face identified risks.

(14)

14

The summarized equation resulting from the dimensions mentioned above is as follow;

Source: Beretta & Bozzolan (2004) page 274

In her paper, Leitner-Hanetseder (2012) attempted to establish a scoring model for the analysis of the

information quality of risk reporting from companies listed in the German and Austrian stock exchange. After determining the information quality from the different risk reports she investigated if the quality of risk reporting is influenced by the quantity of risk reporting, firm size, and/or the index a company is listed with the following regression model:

The results of her applied multiple regression model indicate that quantity of risk reporting is a statistically significant factor in determining the information quality of risk reporting disclosures. Besides quantity, the stock index is also a statistically significant in determining the information quality of risk reporting.

Over the years research has served to remind readers that the quality of information is more important than the quantity (Abraham & Shrives, 2014). In their paper the overall objective is to discuss how best to improve reporting of principal risk factors by public listed companies. They used proprietary costs theory and

institutional theory to explain the evolution of risk reporting disclosure and to help explain why earlier research found that risk disclosure is of poor quality and lacking in relevance.

Proprietary costs theory

The proprietary cost theory embodies the benefits and costs of disclosure.

When managers have to choose a disclosure strategy, they have to make a tradeoff between the benefits from expanded disclosure against the costs of disclosing potentially damaging information.

’’Proprietary costs theory suggests that managers prefer the safety of general routine disclosures which do not change from year to year and hence do not need to be evaluated. While this behavior is understandable, such disclosures will be of limited use and so this theory enables us to identify the second theme in providing quality risk disclosures’’ (Abraham & Shrives, 2014).

(15)

15

Institutional theory

Abraham and Shrives (2014) state that the Institutional Theory can clarify our thinking about disclosure in different ways.

’’First, because of the cost/benefit uncertainties of disclosure, understandably, managers may consider mimicking other companies’ disclosures (Dillard, Rigsby, & Goodman, 2004), particularly companies with good reputations. By mimicking information in their disclosures, companies can signal that their risk management systems are

equivalent to the industry standard’’

Second, “institutional pressures can drive organizations to engage in routine social actions” (Cormier et al., 2005, p. 13). ‘’This suggests that once managers have decided on their risk disclosures, however they are derived, they become reluctant to make changes to existing disclosures, particularly where the consequences of those extra or altered disclosures are unclear.’’ This also combines with the proprietary costs theory, because changing disclosures cost money.

Though in a short horizon this type of disclosure could be acceptable and will probably appear to be right, on the long term horizon the content of disclosure can’t be sustainable, due to the fact that risks are dynamic and they always change. (Abraham & Shrives, 2014).

They suggest that that preparers need to address three questions in their risk disclosures.

 Is risk information specific to the company and are there changes to reported risks in risk factor statements over time?

 Are significant events identified in prior risk factor statements?

 Are significant observed events discussed in subsequent risk factor statements?

This model is applied on four companies in the food production and processing industry to assess risk reporting quality.

They conclude that disclosure of risk factors tends to be quite general and routine and is typical of the type of disclosure predicted by both theories.

Summarizing the above, in the early beginning risk reporting quality was measured by the quantity, consisting of word sentences counts or some kind of checklist.

(16)

16

Yet, over the years more researchers have tried to assess risk reporting quality by not just looking at the quantity, like for instance Beretta and Bozzolan (2004), Leitner-Hanetseder (2012) and Abraham and Shrives (2014).

Though there are not yet uniform generally accepted guidelines to assess risk reporting quality.

In my view there has to be a combination of different measures to assess risk disclosure quality and not just word or sentences counts.

In this thesis therefore a combination of word count together with some aspects of the Beretta and Bozzolan (2004) and Abraham & Shrives’ (2014) model is applied to assess the quality of risk reports over the period 2005 until 2013.

(17)

17 3. Hypothesis development

Due to the type of research question as well as to achieve a high internal validity this paper chose use hypotheses testing as a research method which is further outlined in chapter four.

The CRO, a person publicly known and head responsible for risk reporting, will try to maximize the quality of risk reporting to benefit the companies' stakeholders.

Lee states that a payoff for the CRO is the satisfaction of providing strategic value to the enterprise and the rewards of being a strategic contributor, not just a skilled technician. These assumptions are in line with the stewardship theory.

One of the roles of the CRO is to implement a set of risk metrics and reports, including losses and incidents, key risk exposures and early warning indicators and that one of the aspects of Enterprise Risk Management is stakeholder management. (Lam, 2000)

Lee argues that the role of the CRO is to communicate with all stakeholders concerned and be an advisor to other executives and managers (Lee, 2000).

Beasley et al. (2005) and Yazid et al. (2012) find that the stage of ERM implementation is positively related to the presence of a chief risk officer.

As stakeholder management is a part of ERM and based on the assumptions mentioned above companies who have appointed a CRO should have a better implemented ERM and therefore a higher quality of risk reports. The selected insurance companies from the sample were all member of the CRO forum since 2005. The CRO Forum was formed in 2004 to bring risk management practice in the insurance industry to a higher level. It consists of professional risk managers of the insurance industry that focuses on developing and promoting industry best practices in risk management. (CRO Forum, 2014). This mission statement suggests that risk management practices will be enhanced and risk disclosure as a part of it.

This results in the following statement:

(18)

18

Risk disclosure quality is measured by the following hypotheses derived from the papers of Beretta and Bozzolan (2004) / Leitner-Hanetseder (2012) and Abraham & Shrives (2014).

In this paper risk disclosure quality is measured by the quantity as well as the specific relevance.

Quantity part (Beretta & Bozzolan, 2004)/ Leitner-Hanetseder, 2012)

Though Beretta and Bozzolan state that quantity is not a satisfactory proxy to measure quality, it still is one of the relevant factors also pointed out in the paper of

Leitner-Hanetseder (2012) . Therefore the first hypothesis is as follows :

H1: In the period 2005-2013 there will be an increase in (relative) risk disclosure quantity.

The increase in absolute and relative quantity will be investigated by calculating the total number of words of risk management information and the total words from the annual report. Hereby it is possible to assess if risk management information gained relatively more or less attention than in previous years.

The principle of insurance is transferring risk from one party (the policyholder) to another (insurer).Nowadays insurance companies offer insurance policies for various risks like health , property, plant , equipment , fire , water. Standard type of risks mentioned in most reports are health risks, interest rate risks, currency risks, equity price risks and liquidity risks. Since the year 2009 other types of risks are mentioned like strategic risks , political risks and catastrophic risk.

For investors, it is relevant what type of risks a company bears. The increasing globalizing environment during the last decade resulted in the surge of more risks. So in this study there is a supposed increase in number of risk reported over the years. This results in the following hypothesis:

H2 : In the period 2005 - 2013 there will be an increase in the reported risks.

Risk disclosure should not only contain what type of risk there exist, but also how they tend to mitigate them. This way investors can assess if the risks are controlled properly.

(19)

19

An in- or decrease in the number of controls has to be seen in perspective with the in-or decrease of the number of risks. An increase in reported risks should generally lead to an increase in the number of controls reported. This results in hypothesis number three:

H3 : In the period 2005-2013 there will be an (relative) increase in the controls reported

Studies like Woods and Marginson (2004); Beretta and Bozzolan (2004, 2008); Dobler, Lajili, and Zéghal (2011); Lajili and Zéghal (2005); Linsley and Shrives (2006) and the Accounting Standards Board (2007 & 2009) highlight that risk disclosures are not sufficiently forward-looking (Abraham & Shrives , 2014) in spite of the increasing focus on forward-looking information in corporate annual reports, for example, from regulation (Abraham & Shrives, 2014). Therefore an increase in forward-looking risk disclosure can be seen as a proxy for increasing risk disclosure quality.

This leads to the fourth hypothesis :

H4 : In the period 2005-2013 there will be an increase in forward-looking risks reported.

Substantive part (Abraham & Shrives, 2014)

In analyzing the risk disclosures, they similarly divide the information into two broad groups:

(1) Factors that are general in nature which apply to any business or any business within the industry (symbolic) (2) Factors that are company specific (substantive)

They mention a sensitivity analysis as an example of substantive disclosure.

Linsley and Shrives (2006) state that a lack of quantitative information is considered to be of limited help to users of disclosure information. Berretta and Bozzolan (2004) argue that quantitative information is more relevant than qualitative information. Abraham and Shrives (2014) state that substantive information is more valuable to investors than symbolic information. As a result in this paper is argued that an increase in quantitative as well as substantive (company specific) information can be seen as a proxy for increased risk disclosure quality.

Therefore I will investigate the number of quantitative company specific disclosure reported This leads to the last hypothesis :

(20)

20 4 Empirical methodology and results

4.1 Sample selection

To examine risk disclosure practices I chose to examine the European insurance industry.

In this industry all is about managing risk and due to the globalization and other factors like global warming , terrorism and increasing regulation, the insurance industry has to deal with more risks than ever. There also is little definitive research at the academic or practitioner level on risk reporting by insurance companies (Klumpes et al, 2013).

Since this paper investigates how risk reporting quality from insurance companies evolves with CRO’s in the lead it selected European insurance companies which also are a member of the CRO forum.

The CRO Forum was formed in 2004 to bring risk management practice in the insurance industry to a higher level. It consists of professional risk managers from the insurance industry that focuses on developing and promoting industry best practices in risk management.(CRO Forum, 2014).

In the sample this study selects the 2005, 2009 and 2013 annual reports, since the hypothesized effect of the CRO foundation on the quality of risk-disclosure.

By selecting three different years it is possible to make a longitudinal study of the practices of risk disclosure and examine the development of quality in the period selected.

In the insurance industry , there almost is no authoritative guidance on risk reporting practices by insurance. Neither the Actuarial Standards Board, the UK accounting standards board, or the International Financial

Standards Board provide any detailed promulgation concerning specific or even general guidance for risk reporting for insurance enterprises (Klumpes et al, 2013).

Therefore in this study there is not expected any bias in the risk reports due to specific regulation. Most of the researched annual reports have a separate part for risk disclosure.

The UK reports were mostly separated in both a financial risk as a non-financial risk part in the notes of the consolidated financial statements.

If an annual report didn’t have a separate part for risk disclosure, the separated paragraphs containing risk

management information were aggregated. From 2009 the annual reports mostly contained extra paragraphs of risk information besides the principal risk section.

The total sample contained a total of 273.962 words. These aggregated parts of risk information were further analysed. In total 21 annual reports were analyzed.

(21)

21

Due to the small sample size one should be aware of the limited reliability of the results.

Given a medium to large effect size, 30 participants per cell should lead to about 80% power. (Wilson Van Voorhis and Morgan, (2007).

This study has a sample size of 21 , thus seven participants per cell and will yield power of approximately 50 percent. Therefore the presented results are either meaningful but have to be interpreted with caution.

4.2 Model for analysis

Though there exist many studies regarding risk disclosure quality, many researchers have measured quality relating it to the quantity disclosed.

Moving from the basic idea that disclosure quantity and quality are not separable information in many empirical settings and that disclosure quality is inherently immeasurable (Botosan,

2004), it is generally assumed that the extent of disclosure (i.e., quantity) is an adequate measure of the quality of disclosure (Beretta, 2004).

Due to the objectivity and simplicity of the method, word count is chosen as a measure for quantity.

In Beretta and Bozzolan’s (2004) framework quality of disclosure depends both on the quantity of information disclosed as on the richness offered by additional information.

For instance , they investigate the density and outlook profile of risk disclosure.

An increase in the quantity of risk reporting disclosures is not essentially related to an increase of information quality (Leitner-Hanetseder, 2005). For investors not only the type of risks a company bears is relevant, but also the approach to mitigate the particular risks.

To capture an extra measure, this study will apply a combination of models for the assessment of the quality of risk disclosure.

1. Word count (quantity and relative quantity) , where quantity is derived from the papers of Beretta and Bozzolan (2004) and Leitner-Hanetseder, (2012).

2. Outlook profile (management approach of mitigating or dealing with risks, expectation of future risks , which is derived from the paper of Beretta and Bozzolan (2004)

(22)

22

4.3 Method for analysis

The selected sample contains three annual reports of seven European Insurance companies.

All annual reports were downloaded from the company’s website and the total words were counted. Words of all the 21 annual reports totaled 2.823.513. Consequently the Risk report paragraphs were copied to a another separate file and words were also counted. From 2009 annual reports sometimes contained multiple parts of risk

management information. The total amount of words counted from the risk reports was 273.692. Furthermore, the following factors were also counted and noted :

- number of specified risks - number of controls mentioned.

- number of quantitative substantive disclosure - number of forward-looking disclosure

All data was analyzed using IBM SPSS Statistics 22.

Without any doubt there can be said that coding is an subjective type of research. No individual will exactly present the same outcome. However there are different ways to mitigate the level of subjectivity. One aspect is that the method should be reliable.

The three types of reliability that can be identified are stability, accuracy and reproducibility (Beattie et al, 2004). Stability is about the consistency of coding the data over time.

Though in this paper the data is coded once, fact is that the insurance companies disclosed their risk information in a consistent manner through the years. All risk reports were coded company wise to make it possible to retest some coding when a discrepancy was noticed in the subsequent years. If there were any, the differences were resolved. The second element of reliability is accuracy which refers to how the coding compares to a pre-set standard. In this study accuracy is guaranteed by the following : word count is a uniformly accepted measure. Controls were accounted for if a sentence literally or in the context, contained the words ‘‘manage(d)’’ , ‘‘control(led)’’ or ‘‘mitigate(d)’’. Forward looking risk disclosure was accounted for when a sentence literally or in the context, contained a risk or program of risk in a subsequent year. Risks were accounted for if the report literally contained the word ‘‘risk’’, as specified in chapter 3. Quantitative substantive disclosure was accounted for if a sentence or table literally or in the context, contained numbers or digits related to the specific companies risk.

The last element, reproducibility is only at stake when there are multiple coders involved which in this specific research is not.

(23)

23 5. Empirical results

In this chapter the empirical results are discussed resulting from the content analysis of the 273.692 words of risk disclosure out of the 21 selected annual reports in the European Insurance Industry. The hypothesis are also tested. First of all there will be an overview and discussion of the overall data followed by an analysis of the (relative) quantity of disclosure, the specified risks and the number of controls reported.

Finally there will be a part showing the statistical tests performed to validate all hypotheses.

5.1 Overall data analysis and overview

From various parts of the annual reports a total of 273.692 words were counted containing risk management information with an average of 13.033 words. The number of words of all the annual reports totaled 2.823.513 with an average of 134.453. For all measured factors there is an increase in the average numbers in the period 2005-2013. Only the number of controls shows a slight decrease between 2009 and 2005-2013. Regarding the quantity , there is a substantial increase between the years 2005 and 2009. For example, all 2005 annual reports had a total of 647.046 words while in 2009 the total words amounted 1.064.099. The total number of words from the risk paragraphs respectively increased from 47.311 in year 2005 to 91.823 in 2013. The overall output from SPSS is provided in Appendix 1.

Figure 1 : summary of overall descriptive statistics

N Minimum Maximum Mean Std. Deviation Number of words annual

report 21 38059 216388 134453,00 52979,312

Number of times word

''Risk'' in Annual report 21 175 1047 714,48 269,647

Number of words risk

paragraphs 21 3198 29450 13032,95 7078,139

Number of risks

21 9 35 19,86 8,266

Total number of controls to

mitigate risk ₂₁ ₀ ₉₈ _38,43 _26,275

Number of forward looking

risk disclosure 21 0 12 2,38 3,170

Number of quantitative

company specific factors 21 0 71 25,05 17,223

(24)

24 5.1.1 Quantity of words

The absolute quantity of words in the risk paragraphs increased steadily in the period 2005-2013. In the year 2005 the average number of words was 6.759 which increased to 13.118 words in 2009 respectively 19.223 words in the year of 2013.

Where in the annual reports from 2005 just a couple of pages was dedicated to the disclosure of risk information, in the years 2009 and 2013 the number of pages increased substantially. From 2009,in various annual reports the number of pages containing risk disclosure totaled 50 or more.

Figure 2 : descriptive statistics absolute quantity per year

To put this absolute increase in quantity in perspective, a calculation is necessary of the relative increase in quantity of words. This is done by dividing the total number of words from the risk paragraphs by the words of the annual reports.

Doing this, it is possible to assess the relative increase in the number of words.

As shown by figure 3 below there is an increase in the relative quantity of words in the period 2005-2013.

Figure 3 : descriptive statistics relative quantity per year

The significance of the increase in (relative)quantity of words as well as the other measurements will be tested and shown in a later paragraph.

N Minimum Maximum Mean Std. Deviation Number of words risk

paragraphs 2005 7 3198,0 10780,0 6758,714 3080,5572

paragraphs 2009 7 6977,0 16559,0 13117,571 3556,9061

paragraphs 2013 7 10379,0 29450,0 19222,571 7363,8806

Valid N (listwise) 7

N Minimum Maximum Mean Std. Deviation

Number of words risk report 2005/

words annual report 2005 7 ,028 ,131 ,08140 ,031865

(25)

25 5.1.2 Number of risks

In the period 2005-2013 the average number of reported risks increased from 15 in the year 2005 to 24 in the year 2013 as seen in figure 4 below.

Figure 4 : descriptive statistics number of reported risks per year

5.1.3 Number of controls

The average number of reported controls increased from 27 in the year 2005 to 45 in 2009. From the year 2009 to 2013 there is shown a little decrease from 45 to 44 controls mentioned. This decrease is mainly caused by Aviva which reported 77 controls in the year 2009. 35 controls of them were included in a separate overall scheme. Neither in 2005 and 2013 this type of scheme was included in their risk report. Neglecting this 35 controls would have resulted in average numbers of respectively 27, 40 and 44 controls over years 2005, 2009 and 2013. Munich Re and Swiss Re reported the most controls of all companies. In this paper this is not further investigated. An example of a control is illustrated by the following Aviva 2013 annual report on page 223: ’’For the UK annuities business interest rate exposure is mitigated by closely matching the duration of liabilities with assets of the same duration.’’

Another example is from the Prudential annual report 2013 on page 49: ’’Persistency risk is mitigated by appropriate training and sales processes and managed proactively post sale’’.

Totaal number of mentioned risks 2005 ₇ ₉ ₂₅ _15,14 _6,962

Totaal number of mentioned risks 2009 ₇ ₁₂ ₃₄ _20,14 _8,214

Totaal number of mentioned risks 2013

7 16 35 24,29 7,931

(26)

26

Figure 4 : descriptive statistics number of controls reported per year

Figure 5 : descriptive statistics number of controls reported per company

An in- or decrease in the number of controls has to be seen in perspective with the in-or decrease of the number of risks. An increase in mentioned risks should generally lead to an increase in the number of controls reported. To see this in perspective it is more relevant to calculate the number of controls per risk.

Figure 6 : descriptive statistics number of controls reported per company

N Minim um Maxim um Mean Std. Deviation Total num ber of controls

taken to m itigate ris k 2005 ₇ _0,0 _71,0 _27,000 _22,9637

Total num ber of controls taken to m itigate ris k 2009

7 16,0 90,0 44,714 29,4941

Total num ber of controls taken to m itigate ris k 2013

7 16,0 98,0 43,571 25,9734

Valid N (lis twis e) ₇

Aviva 3 29 77 47,67 25,716 Allianz 3 10 28 18,67 9,018 Aegon 3 38 46 41,00 4,359 Munich Re ₃ ₄₂ ₇₁ _54,67 _14,844 Prudential ₃ ₀ ₁₆ _10,67 _9,238 Swiss Re ₃ ₂₀ ₃₈ _26,67 _9,866 Zurich Re 3 21 98 69,67 42,336 Valid N (listwise) 3

N Minimum Maximum Mean Std. Deviation Total number of controls

taken to mitigate risk

2005/number of risks 2005 7 0,00 3,80 1,73 1,30233

Total number of controls taken to mitigate risk

2009/number of risks 2009 7 1,10 4,05 2,19 1,15426

Total number of controls taken to mitigate risk

2013/number of risks 2013 7 1,00 2,88 1,78 ,78015

(27)

27

Figure 6 illustrates that the average of controls per reported risks increased from 1,7 in 2005 to 2,2 in 2009. Though there is observed a decrease to 1,78 controls per reported risk in 2013.

5.1.4 Number of quantitative substantive disclosure

An example of quantitative and company specific information is illustrated by the Munich annual report 2013 on page 121: ’’As in 2012, the largest natural catastrophe exposure for Munich Re is the 2.9 bn currently retained for the ’’Atlantic Hurricane’’ scenario (value at risk for a 200-year return period’’. Much but not all quantitative disclosure is also company specific. This is illustrated by the Aegon annual report 2009 on page 120 showing a table of closing exchange rates of various currencies. Though this is quantitative, it is not specific to Aegon and therefore not accounted for. The figure below shows an increase in the absolute average number of quantitative substantive disclosure in the period 2005-2013 with an average in year 2005 of 13 increasing to 33 in year 2013. This increase is partly caused by the various tables which were included in the risk reports since 2009. These are mainly present in the note on risk factors of the financial statements part of the annual report.

Figure 7 : descriptive statistics number of quantitative company specific factors reported per year

5.1.5 Number of forward looking risk disclosure

There is relatively paid little attention to forward-looking disclosure considering the few numbers disclosed in the risk paragraphs. Only Allianz has embodied a clear forward looking part in their risk disclosure section of the annual report. An example is illustrated by the following part out of their annual report 2009.

’’With respect to risk management, our general objectives for 2010 include three priorities’’.

Number of quantitative company specific factors

mentioned 2005 7 0,0 26,0 12,714 8,1387

mentioned 2009 7 11,0 53,0 29,429 14,4090

mentioned 2013 7 12,0 71,0 33,000 21,0079

(28)

28

And : ’’Our third priority is to do what is necessary to meet Solvency II internal model requirements by the end of 2012-which is one of the Allianz Group’s top ten priorities set by the Board of Management of Allianz SE during 2009’’. (Allianz , 2010 , p202)

Figure 8 and 9 show an increase in the number of outlook items mentioned over the years 2005-2013 as well as the average number of forward looking items over the entire period.

Figure 8 : number of forward-looking risk disclosure per company per year

Figure 9 : descriptive statistics number of forward-looking risk disclosure per year

Total number of forward

looking risk disclosure 2005 7 0 5 1,14 2,035

(29)

29 5.1.6 Testing of hypothesis

In this part statistical tests are performed to validate each of the hypothesis. Consequently the results of them are discussed.

To test for significance of the results in the previous paragraphs the Wilcoxon signed Ranks test is used which is a non-parametric test. The advantage with Wilcoxon Signed Ranks Test is that it neither depends on the form of the parent distribution nor on its parameters. It does not require any assumptions about the shape of the distribution like other tests like the t-test. For this reason, this test is often used as an alternative to t test's whenever the population cannot be assumed to be normally distributed (Explorable.com, 2009). In this study the population is not normally distributed. Several researchers have used the Wilcoxon signed Ranks test in their papers

investigating risk disclosure (Greco, 2012, Rajab, 2009, Linsley and Shrives, 2006)

There will be tested whether there is a significant increase in absolute as well as relative risk disclosure quantity in the period 2005-2013.

(30)

30

The first hypothesis that will be tested is H1: in the period 2005 until 2013 there will be an increase of (relative) risk disclosure quantity.

Figure 10 : Wilcoxon signed ranks test for absolute risk disclosure quantity

As can be seen in the figure 10 above the difference in quantity of disclosure is significant at a five percent confidence level between 2005 and 2013 (p-value = 0,009) as well as for the period 2005 until 2009 (p-value = 0,009) and 2009 until 2013 (p-value = 0,014). These findings support the hypothesis that there is an increase in risk disclosure quantity in the period 2005 until 2013. When the relative quantity is observed there is also a significant increase at five per cent confidence level (p-value = 0,022) in the period 2005 and 2013.

N Mean Rank Sum of Ranks

Negative Ranks ₀a _0,00 _0,00 Positive Ranks ₇b _4,00 _28,00 Ties ₀c Total 7 Negative Ranks ₁d _1,00 _1,00 Positive Ranks ₆e _4,50 _27,00 Ties ₀f Total ₇ Negative Ranks ₀g 0,00 0,00 Positive Ranks ₇h 4,00 28,00 Ties ₀i Total 7

Total number of words risk paragraphs 2009 - Total number of words risk paragraphs 2005

Z _-2,366b

-2,197b -2,371b

Asymp. Sig. (2-tailed) ,018 ,028 ,018

b. Based on negative ranks.

h. Total number of words risk paragraphs 2013 > Total number of words risk paragraphs 2005 i. Total number of words risk paragraphs 2013 = Total number of words risk paragraphs 2005

Test Statisticsa

a. Wilcoxon Signed Ranks Test

c. Total number of words risk paragraphs 2009 = Total number of words risk paragraphs 2005 d. Total number of words risk paragraphs 2013 < Total number of words risk paragraphs 2009 e. Total number of words risk paragraphs 2013 > Total number of words risk paragraphs 2009 f. Total number of words risk paragraphs 2013 = Total number of words risk paragraphs 2009 g. Total number of words risk paragraphs 2013 < Total number of words risk paragraphs 2005 Total number of words

risk paragraphs 2009 - Total number of words risk paragraphs 2005 Total number of words risk paragraphs 2013 - Total number of words risk paragraphs 2009 Total number of words risk paragraphs 2013 - Total number of words risk paragraphs 2005

a. Total number of words risk paragraphs 2009 < Total number of words risk paragraphs 2005 b. Total number of words risk paragraphs 2009 > Total number of words risk paragraphs 2005

(31)

31

This is illustrated by figure 11 below.

Figure 11 : Wilcoxon signed ranks test for relative risk disclosure quantity

There can be concluded that there is enough evidence to support H1.

Negative Ranks ₃a _3,00 _9,00 Positive Ranks ₄b _4,75 _19,00 Ties ₀c Total ₇ Negative Ranks ₀d _0,00 _0,00 Positive Ranks ₇e _4,00 _28,00 Ties ₀f Total ₇ Negative Ranks ₁g _2,00 _2,00 Positive Ranks ₆h _4,33 _26,00 Ties ₀i Total ₇

Total number of words risk report 2009/words annual reports 2009 - Total number of words risk report 2005/words annual reports 2005

Z _-,845b

-2,366b -2,028b

Asymp. Sig. (2-tailed) ,398 ,018 ,043

Total number of words risk report 2009 / words annual reports 2009 - Total number of words risk report 2005 / words annual reports 2005

a. Total number of words risk report 2009/words annual reports 2009 < Total number of words risk report 2005/words annual reports 2005

b. Total number of words risk report 2009/words annual reports 2009 > Total number of words risk report 2005/words annual reports 2005

c. Total number of words risk report 2009/words annual reports 2009 = Total number of words risk report 2005/words annual reports 2005

d. Total number of words risk report 2013/words annual reports 2013 < Total number of words risk report 2009/words annual reports 2009

e. Total number of words risk report 2013/words annual reports 2013 > Total number of words risk report 2009/words annual reports 2009

f. Total number of words risk report 2013/words annual reports 2013 = Total number of words risk report 2009/words annual reports 2009

g. Total number of words risk report 2013/words annual reports 2013 < Total number of words risk report 2005/words annual reports 2005

h. Total number of words risk report 2013/words annual reports 2013 > Total number of words risk report 2005/words annual reports 2005

i. Total number of words risk report 2013/words annual reports 2013 = Total number of words risk report 2005/words annual reports 2005

Test Statisticsa

(32)

32

The second hypothesis is H2 : in the period 2005-2013 there will be an increase in the reported risks.

Figure 12 : Wilcoxon signed ranks test for number of risks

As illustrated by figure 12 there is a significant increase in the number of risks measured at a confidence level of five percent in the years 2005-2013. The corresponding p-value is 0,009. There can be concluded that there is enough evidence to support H2.

Negative Ranks ₀a 0,00 0,00 Positive Ranks ₇b _4,00 _28,00 Ties ₀c Total 7 Negative Ranks ₀d _0,00 _0,00 Positive Ranks ₆e _3,50 _21,00 Ties ₁f Total ₇ Negative Ranks ₀g _0,00 _0,00 Positive Ranks ₇h _4,00 _28,00 Ties ₀i Total 7

Total number of risks 2009 - Total number of

risks 2005

Total number of risks 2013 - Total number of

risks 2009

Total number of risks 2013 - Total number

of risks 2005

Z _-2,388b

-2,232b -2,375b

Asymp. Sig. (2-tailed) ,017 ,026 ,018

h. Total number of risks 2013 > Total number of risks 2005 i. Total number of risks 2013 = Total number of risks 2005

Test Statisticsa

c. Total number of risks 2009 = Total number of risks 2005 d. Total number of risks 2013 < Total number of risks 2009 e. Total number of risks 2013 > Total number of risks 2009 f. Total number of risks 2013 = Total number of risks 2009 g. Total number of risks 2013 < Total number of risks 2005 Total number of risks 2009 -

Total number of risks 2005

Total number of risks 2013 - Total number of risks 2009

Total number of risks 2013 - Total number of risks 2005

a. Total number of risks 2009 < Total number of risks 2005 b. Total number of risks 2009 > Total number of risks 2005

(33)

33

The third hypothesis is H3 : in the period 2005-2013 there will be an (relative) increase in the controls reported.

Figure 12 : Wilcoxon signed ranks test for number of controls

Figure 12 above shows that the number of controls didn’t significantly increase in the period 2005-2013 as well as for the period 2005-2009. When observing the relative number of controls also no significant increase is shown at a five percent confidence level. This is illustrated by figure 13 below.

Negative Ranks ₁a _5,00 _5,00 Positive Ranks ₆b _3,83 _23,00 Ties ₀c Total ₇ Negative Ranks ₂d _4,50 _9,00 Positive Ranks ₄e _3,00 _12,00 Ties ₁f Total ₇ Negative Ranks ₁g _6,00 _6,00 Positive Ranks ₆h _3,67 _22,00 Ties ₀i Total ₇

Total number of controls taken to mitigate risk 2009 -

Total number of controls taken to mitigate risk 2005

Z _-1,521b

-,314b -1,357b

Asymp. Sig. (2-tailed) ,128 ,753 ,175

Total number of controls taken to mitigate risk 2009 - Total number of controls taken to mitigate risk 2005

a. Total number of controls taken to mitigate risk 2009 < Total number of controls taken to mitigate risk 2005 b. Total number of controls taken to mitigate risk 2009 > Total number of controls taken to mitigate risk 2005 c. Total number of controls taken to mitigate risk 2009 = Total number of controls taken to mitigate risk 2005 d. Total number of controls taken to mitigate risk 2013 < Total number of controls taken to mitigate risk 2009 e. Total number of controls taken to mitigate risk 2013 > Total number of controls taken to mitigate risk 2009 f. Total number of controls taken to mitigate risk 2013 = Total number of controls taken to mitigate risk 2009 g. Total number of controls taken to mitigate risk 2013 < Total number of controls taken to mitigate risk 2005 h. Total number of controls taken to mitigate risk 2013 > Total number of controls taken to mitigate risk 2005 i. Total number of controls taken to mitigate risk 2013 = Total number of controls taken to mitigate risk 2005

Test Statisticsa

(34)

34

Figure 13 : Wilcoxon signed ranks test for relative number of controls

Negative Ranks 3a 3,00 9,00 Positive Ranks 4b 4,75 19,00 Ties 0c Total 7 Negative Ranks 4d 5,50 22,00 Positive Ranks 3e 2,00 6,00 Ties 0f Total 7 Negative Ranks 4g 3,50 14,00 Positive Ranks 3h 4,67 14,00 Ties 0i Total 7

Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009 - Total number of controls taken

to mitigate risk 2005 / total amount of risk mentioned 2005

Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 - Total

number of controls taken to mitigate risk 2009 / total amount

of risk mentioned 2009

Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 -

Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

Z _-,845b

-1,352c ,000d

Asymp. Sig. (2-tailed) ,398 ,176 1,000

b. Based on negative ranks. c. Based on positive ranks.

d. The sum of negative ranks equals the sum of positive ranks.

h. Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 > Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

i. Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 = Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

Test Statisticsa

c. Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009 = Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

d. Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 < Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009

e. Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 > Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009

f. Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 = Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009

g. Total number of controls taken to mitigate risk 2013 / total amount of risk mentioned 2013 < Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009 - Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

a. Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009 < Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

b. Total number of controls taken to mitigate risk 2009 / total amount of risk mentioned 2009 > Total number of controls taken to mitigate risk 2005 / total amount of risk mentioned 2005

(35)

35

Summarizing the above, there is not enough evidence to support H3, neither for the absolute number nor for the relative number of controls.

The fourth hypothesis is H4 : in the period 2005-2013 there will be an increase in forward-looking risks reported.

Figure 14 : Wilcoxon signed ranks test for number of forward looking disclosure

Figure 14 above illustrates that there is not a significant increase in forward looking risk disclosure in the period 2005-2013 at a five per cent confidence level. (p-value = 0,0575).

Negative Ranks ₂a _2,75 _5,50 Positive Ranks ₃b _3,17 _9,50 Ties ₂c Total 7 Negative Ranks ₀d 0,00 0,00 Positive Ranks ₆e _3,50 _21,00 Ties ₁f Total 7 Negative Ranks ₁g 3,00 3,00 Positive Ranks ₅h _3,60 _18,00 Ties ₁i Total 7 Total number of forward looking disclosure items 2009 - Total number of forward looking disclosure items 2005 Total number of forward looking disclosure items 2013 - Total number of forward looking disclosure items 2009

Total number of forward looking disclosure items 2013 - Total number of forward looking disclosure items 2005 Z _-,542b -2,207b -1,577b

Asymp. Sig. (2-tailed) ,588 ,027 ,115

h. Total number of forward looking disclosure items 2013 > Total number of forward looking disclosure items 2005 i. Total number of forward looking disclosure items 2013 = Total number of forward looking disclosure items 2005

Test Statisticsa

c. Total number of forward looking disclosure items 2009 = Total number of forward looking disclosure items 2005 d. Total number of forward looking disclosure items 2013 < Total number of forward looking disclosure items 2009 e. Total number of forward looking disclosure items 2013 > Total number of forward looking disclosure items 2009 f. Total number of forward looking disclosure items 2013 = Total number of forward looking disclosure items 2009 g. Total number of forward looking disclosure items 2013 < Total number of forward looking disclosure items 2005 Total number of forward

looking disclosure items 2009 - Total number of forward looking disclosure items 2005

Total number of forward looking disclosure items 2013 - Total number of forward looking disclosure items 2009

Total number of forward looking disclosure items 2013 - Total number of forward looking disclosure items 2005

a. Total number of forward looking disclosure items 2009 < Total number of forward looking disclosure items 2005 b. Total number of forward looking disclosure items 2009 > Total number of forward looking disclosure items 2005

(36)

36

Over the period 2009-2013 there is (p-value = 0,0135)

Summarizing the above, there is not sufficient evidence for support of H4.

The fifth and last hypothesis is H5 : in the period 2005-2013 there will be an increase in quantitative substantive risk disclosure.

Figure 15 : Wilcoxon signed ranks test for number of quantitative company specific disclosure

Figure 15 above shows a significant increase in quantitative substantive disclosure in the period 2005-2013 (p-value = 0,0315). Summarizing, there is sufficient evidence for the support of H5.

Negative Ranks ₁a _2,00 _2,00 Positive Ranks ₆b _4,33 _26,00 Ties ₀c Total 7 Negative Ranks ₂d 4,50 9,00 Positive Ranks ₅e 3,80 19,00 Ties ₀f Total 7 Negative Ranks ₂g _1,50 _3,00 Positive Ranks ₅h _5,00 _25,00 Ties ₀i Total 7 Number of quantitative company specific factors

2009 - Number of quantitative company specific factors 2005 Number of quantitative company specific factors 2013 - Number of quantitative company specific factors 2009 Number of quantitative company specific factors 2013 - Number of quantitative company specific factors 2005 Z _-2,043b -,847b -1,859b

Asymp. Sig. (2-tailed) ,041 ,397 ,063

h. Number of quantitative company specific factors 2013 > Number of quantitative company specific factors 2005 i. Number of quantitative company specific factors 2013 = Number of quantitative company specific factors 2005

Test Statisticsa

c. Number of quantitative company specific factors 2009 = Number of quantitative company specific factors 2005 d. Number of quantitative company specific factors 2013 < Number of quantitative company specific factors 2009 e. Number of quantitative company specific factors 2013 > Number of quantitative company specific factors 2009 f. Number of quantitative company specific factors 2013 = Number of quantitative company specific factors 2009 g. Number of quantitative company specific factors 2013 < Number of quantitative company specific factors 2005 Number of quantitative

company specific factors 2009 - Number of quantitative

company specific factors 2005 Number of quantitative

company specific factors 2009 Number of quantitative

company specific factors 2005

a. Number of quantitative company specific factors 2009 < Number of quantitative company specific factors 2005 b. Number of quantitative company specific factors 2009 > Number of quantitative company specific factors 2005