Disclosure theories and disclosure measures

Disclosure theories and disclosure measures * Teorías y medidas de la divulgación de la información Francisco Bravo Urquiza **. Universidad de Sevilla María Cristina Abad Navarro. Universidad de Sevilla Marco Trombetta. Instituto de Empresa ABSTRACT Theories have been developed in the disclosure literature to explain the reasons behind the decision to disclose more information. Empirical evidence does not consistently support disclosure theories and results found are contradictory. The diffi culty in measuring voluntary disclosure might be one of the reasons infl uencing on these divergences. In this paper, we investigate two key questions related to disclosure measurement. First, we aim to empirically test if use of disclosure indices that measure different information attributes determines validity of disclosure theories. Second, we investigate how disclosure indices design infl uences results. Results show that determinants of more specifi c information attributes are different than those that infl uence less specifi c attributes. Furthermore, independently of the information attribute that is measured, disclosure measure design infl uences results. KEY WORDS Disclosure indices; Disclosure determinants; Forward-looking information; Disclosure theories. RESUMEN En la literatura sobre divulgación de información se han desarrollado diversas teorías para explicar los motivos que llevan a las empresas a divulgar más información. La evidencia empírica no apoya consistentemente las teorías y los resultados encontrados son contradictorios. Estas divergencias podrían estar ocasionadas por problemas implícitos a la medición de la información voluntaria. En este trabajo, se estudian dos cuestiones clave relacionadas con este problema. Primeramente, investigamos si el uso de índices de revelación que captan diferentes atributos informativos determina la validez de las teorías enunciadas sobre la divulgación de información. Por otro lado, analizamos la infl uencia de utilizar distintas medidas de la información. Los resultados muestran que los determinantes de atributos informativos más específi cos son distintos de aquellos factores que infl uyen en la divulgación de atributos menos específi cos. Además, independientemente del atributo informativo que es medido, el diseño de un índice de revelación infl uye en los resultados obtenidos. PALABRAS CLAVE Divulgación de información; Determinantes de la divulgación; Información previsional; Teorías sobre divulgación. Recibido 10-03-09 · Aceptado 03-03-10 · Copyright © 2001 Asociación Española de Contabilidad y Administración de Empresas · ISSN: 0210-2412 REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD Vol. XXXIX, n.º 147 · julio-septiembre 2010 · pp. 393-415 ARTÍCULOS DOCTRINALES 393 * Acknowledgements: The authors are grateful for the valuable comments and suggestions made by the anonymous referees in the preparation of this article. The authors also want to thank Emma García-Meca and Juan Manuel García Lara for their discussions on previous versions of this paper. ** Corresponding author: Francisco Bravo Urquiza. Departamento de Contabilidad y Economía Financiera. Universidad de Sevilla. Avenida Ramón y Cajal, 1, 48018 Sevilla (Spain). Tel.: (+34) 954556067. Fax: (+34) 954557569. E-mail: pacobravo@ us.es Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 394 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 1. INTRODUCTION Various theories have been developed in the disclosure literature to explain the reasons behind the decision to disclose more and/or better information. Empirical tests of disclosure theories identify which characteristics of the companies determine disclosure (Singhvi and Desai, 1971; Firth, 1979; Cooke, 1992; Raffournier, 1995; Giner, 1997; Archambault and Archambault, 2003; Prencipe, 2004). Although the methodologies applied are similar, not only do the results lack consistency in their support of disclosure theories but they are also often contradictory (Ahmed and Courtis, 1999; Chavent et al., 2006). One possible explanation for these contradictory results lies in the diffi culty inherent to the measurement of information disclosure. The diffi culty in measuring voluntary disclosure is one of the most important problems encountered by empirical researchers in disclosure studies (Healy and Palepu, 2001). In this paper, we tackle this problem by investigating two key questions related to disclosure measurement. To the best of our knowledge, these questions remain unchallenged in the literature on disclosure determinants. First, we aim to empirically test which determinants are related to which specifi c information attributes. In other words, we want to test how using disclosure indices that measure different information attributes affect the signifi cance of some of the determinants proposed by disclosure theories. With this purpose in mind, we compare results obtained when using various disclosure indices that measure different information attributes in a regression analysis of disclosure determinants. The information attributes considered are quantity, coverage and quality. Second, we study how disclosure index design infl uences the results of empirical analysis of disclosure theories. To this end, we compare two disclosure indices that capture the same information attribute, but which are designed in a different way. Our empirical analysis is conducted on a sample of listed Spanish companies and focuses on the disclosure of forward-looking information. We fi nd that size is the only determinant that signifi cantly affects all the disclosure attributes considered. The signifi cance of other determinants depends on the attributes of disclosure which are measured by the specifi c index adopted. Moreover, we will show that, even when we focus on a specifi c attribute of disclosure, the design of the index used for the measurement of that attribute affects the results of the analysis. Disclosure theories do not generally provide an exhaustive defi nition of the term «disclosure». Theories assume that, under specifi c circumstances, companies disclose more information because perceived expected benefi ts exceed costs. However, the theoretical arguments are enunciated in a generic way and fail to explicitly defi ne what exactly is meant by «more information» or «an increase in disclosure». Disclosure is a multidimensional concept that integrates various attributes (Beattie et al., 2004). Consequently, it is expected that the determinants of disclosure of specifi c information attributes will be different to those of other attributes. For instance, the factors that infl uence disclosure quantity can differ from the determinants of information quality. We test whether the use of disclosure measures that capture different information attributes results in empirical evidence showing that the Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 395 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 disclosure determinants also differ. Disclosure indices that capture dissimilar information attributes —such as quantity, coverage or quality— coexist in the literature (Entwistle, 1999; Hussainey et al., 2003; Wallace et al., 1994; Giner, 1997; Prencipe, 2004; Beretta and Bozzolan, 2005; Cerbioni and Parbonetti, 2007). We expect that the general theoretical arguments will be verifi ed when less specifi c disclosure measures are used (e.g.: quantity or coverage). However, the theoretical predictions on disclosure determinants may lack full verifi cation when more specifi c measures (such as information richness or depth) are employed. This being the case, then the implications for research on disclosure determinants are substantial. It would highlight the need to enunciate theories on disclosure determinants more explicitly, by taking information attributes into consideration. In the same way, disclosure indices should be designed on the basis of specifi c information characteristics, in order to obtain empirical results of a more conclusive nature. Disclosure index design is a complex task, affected by subjectivity. Several diffi culties have to be faced when building a disclosure index, such as item selection and weighting (García- Meca and Martínez, 2004). As Ahmed and Courtis (1999) suggest, the use of diverse disclosure measures is one possible explanation for the contradictory results found in empirical research on disclosure determinants. Self-constructed indices that capture coverage are the most commonly used measures. A common approach in this kind of study is to design an index that captures coverage of several informational items and to relate it to several company characteristics. Despite the controversial nature of this methodology, as far as we know there are no studies about the impact of different index designs on the results in this research area. Empirical evidence on this matter will highlight the need for the specifi c disclosure measure to be considered when interpreting results, and more importantly, when comparing these results with those in the prior literature. This research contributes towards our understanding of the disclosure measurement and of the research on disclosure determinants. It sheds some light on the reasons why contradictory results on voluntary disclosure determinants have been documented. «Disclosure» as a concept is still not clearly defi ned in terms of specifi c information attributes. This lack of consensus on the disclosure index design may explain divergences in the results. The implications for future research include the necessity to be more precise about information attributes when developing theoretical hypotheses and the need to search for more clearly defi ned disclosure measures that provide comparable results. This paper is organized as follows. The literature regarding disclosure theories and disclosure indices is reviewed in the next section. Section 3 describes data collection and the sample, whereas Section 4 explains the research method. Section 5 discusses the results of the empirical analysis and Section 6 summarizes the contributions of the paper. 2. LITERATURE REVIEW Disclosure of fi nancial information is an area that has triggered a great deal of academic interest in recent years. Among other topics, research is focused on the corporate characteristics that help to predict a fi rm’s disclosure level. Even though similar methodologies and corporate characteristics have been used in the prior research on disclosure determinants, Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 396 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 some disagreements have appeared in relation to some of these characteristics (Chavent et al., 2006). In some cases, results fail to support theoretical arguments; in others, results, from empirical studies that are designed in a similar fashion, differ (Cooke, 1989; Raffournier, 1995; Prencipe, 2004). These contradictory results suggest the existence of some open questions in this research area. One of the most important drawbacks encountered by researchers has to do with the problems that are inherent to the measurement of voluntary disclosure (Healy and Palepu, 2001). We hypothesize that there are two main factors that might be causing these contradictory results: fi rst, disclosure theories use a concept of disclosure that is not specifi cally defi ned in terms of information attributes; second, different disclosure measures are employed in each of the empirical studies. In this section, both the literature about disclosure theories and the literature on disclosure measure design are reviewed. 2.1. VOLUNTARY DISCLOSURE THEORIES The determinants of voluntary disclosure have been subjected to major analysis. Several theories explain the reasons for companies to reveal voluntary information (under the assumption that fi rms perceive benefi ts from disclosure), including agency theory, signalling theory and political process theory, among others. Proprietary costs as well as costs derived from the collection and preparation of information must also be considered from a theoretical perspective (Prencipe, 2004). To a lesser extent, other types of costs have been found as limitations to the disclosure of information, such as corporate governance and monitoring, capital needs, litigation costs, and audit fi rm reputation (Ahmed and Courtis, 1999). The theoretical arguments on the determinants of voluntary information disclosure are summarized below. Agency theory defi nes an agency relationship as a contract under which one or more persons (principals) engage another person (agent) to perform some service on their behalf which involves delegating some decision-making authority to the agent (Jensen and Meckling, 1976). It is expected that the agent will not always act in the best interest of the principal. Agency theory claims that confl icts are expected to arise when there is incomplete and asymmetric information between principal and agent in a company. Both parties may have different interests and this problem could be minimized by providing more information. Some determinants of voluntary information that have been commonly associated to the agency problem are size, leverage, profi tability and listing status. First, given that larger fi rms carry out a greater number of contracts which are more complex than smaller fi rms, agency costs depend on company size (Rodríguez Perez, 2004). Larger fi rms are expected to reveal more voluntary information to reduce these costs. Second, agency costs are higher when the proportion of debt increases. Agency theory predicts that a highly leveraged company has more of an obligation to satisfy the information needs of long-term and short-term creditors, and hence it may provide more detail to meet those needs than would a less leveraged fi rm (García-Meca et al., 2005). Third, higher margin and higher profi tability lead to a greater level of disclosure in order to ob- Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 397 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 tain and justify better contractual conditions. Managers will disclose detailed information to improve their compensation arrangements (Giner, 1997). Finally, listed companies are expected to provide more information due to the higher information requirements they face, or due to agency costs (Giner, 1995). Specifi cally, international listing status is also expected to infl uence disclosure. Disclosure serves to control the agency costs that appear when ownership is more disperse (García-Meca et al., 2005). Signalling theory indicates that asymmetric information between a company and the investors causes adverse selection. To avoid this situation, companies disclose information voluntarily, providing signals to the market (Watts and Zimmerman, 1986). Size, profi tability and growth are factors that infl uence the decision to disclose voluntary information to avoid adverse selection. Information asymmetries will be larger for big companies, which justify more disclosure for mitigation purposes (Rodríguez Pérez, 2004). Moreover, fi rms with a high profi tability will have a higher tendency to disclose more information to the markets, to increase investor confi dence (Singhvi and Desai, 1971) and prevent undervaluation of their shares (Giner, 1997). Finally, growth and disclosure are expected to be positively related (Lev and Penman, 1990) since companies with high growth rates provide more information to be more attractive in the market. Political process theory suggests that regulators make decisions based on the information disclosed by fi rms (Watts and Zimmerman, 1986). Companies disclose voluntary information to minimize these political costs. Size and profi tability are incentives for companies to reveal more information to reduce these costs. Larger fi rms are subject to higher political costs, leading to a greater level of disclosure (Watts and Zimmerman, 1986). Higher information disclosure is expected to justify a fi rm’s large profi ts and thus avoid legal obligations (Lang and Lundholm, 1993) and as a justifi cation of the company’s profi t level (Giner, 1997). Political costs and the competitive environment also infl uence the level of information disclosed in an industry (Mora and Rees, 1996). In addition, proprietary costs are considered as one of the most important limitations to information disclosure. Competitive disadvantages infl uence the decision to provide private information. Smaller fi rms are sensitive, to a great extent, to the disadvantages that, in terms of competitive edge, are derived from a higher disclosure level (Singhvi and Desai, 1971; Giner, 1995). The previous literature has also considered the costs derived from the collection and preparation of information as an obstacle to revealing more voluntary information. Company size plays an important role to minimize these costs, which decrease for larger fi rms (Land and Lundholm, 1993). In summary, under the framework of these theories, the prior research has employed variables, such as size, leverage, profi tability, growth and listing status as determinants of voluntary information disclosure. However, theories refer to information disclosure in general terms, while this concept embodies several attributes or dimensions. The reasons to increase or improve the disclosure of a specifi c information attribute might be different to the rationale for enhancing another attribute. Disclosure indices that capture different information attributes concur in the literature. Our objective is to test whether the deter- Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 398 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 minants of a specifi c information attribute disclosure differ from those of other attributes. This being the case, existing disclosure theories will not remain applicable in all cases. 2.2. MEASUREMENT OF DISCLOSURE Disclosure is measured in the prior research by following various approaches, all under the implicit assumption that disclosure quality is what is being measured (Beattie et al., 2004). One approach is the use of scores provided by analysts according to the amount of information disclosed-such as scores elaborated by the Association for Investment Management and Research [AIMR (1)], and Standard & Poor’s (S&P). However, it has been argued that they are biased owing to analysts’ subjectivity or the number of companies that compose the AIMR or S&P samples (Healy and Palepu, 2001). Another approach is the use of self-constructed indices. This approach is the most frequently employed due to the need for disclosure measures that are valid regardless of the companies selected or the kind of information analysed. Self-constructed indices commonly measure voluntary information in general (Chow and Wong-Boren, 1987; Raffournier, 1995; Botosan, 1997; Depoers, 2000), or mandatory information (Wallace and Naser, 1995, Chen and Jaggi, 2000). However, some studies have employed disclosure indices to measure specifi c information, like intellectual capital (Cerbioni and Parbonetti, 2007; Li et al., 2008), environmental and social (Williams, 1999) or forward-looking information (Beretta and Bozzolan, 2005; Aljifri and Hussainey, 2007). As mentioned in previous sections, diverse information attributes are considered when designing self-constructed measures. Some disclosure indices are based on information quantity (Entwistle, 1999; Williams, 1999; Hussainey et al., 2003; Aljifri and Hussainey, 2007; Li et al., 2008). Measures of a more specifi c and complex nature have recently been developed, including several dimensions related to information richness (Beretta and Bozzolan, 2005; Cerbioni and Parbonetti, 2007). Nevertheless, the majority of studies employ indices that capture coverage of several information categories. Although most of disclosure indices do not assign any weight to the information items (Chow and Wong-Boren, 1987; Cooke, 1989; Raffournier, 1995; Giner, 1997; Depoers, 2000), some of them select weighted items (Singhvi and Desai, 1971; Buzby, 1975; Chow and Wong-Boren, 1987). Several indices have also considered information specifi city to obtain a proxy for information scope by assigning a higher value to quantitative information (Botosan, 1997; Robb et al., 2001). Despite the proliferation of information indices of similar design, the results from research on disclosure determinants are contradictory. Most of the studies agree on the existence of a positive relationship between size and disclosure, indicating that larger companies disclose more information (Buzby, 1975; Firth, 1979; McNally et al., 1982; Cooke, 1989; Wallace et al., 1994; Giner, 1997; Depoers, 2000; Alsaeed, 2005). However, other studies show no evidence on this relationship (Malone et al., 1993; Ahmed, 1996, Prencipe, 2004). Disagreements are also found in studies on forward-looking information; some research fi nds a (1) As of 9 May 2004, AIMR changed its name to CFA institute. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 399 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 signifi cant relationship (Kent and Ung, 2003; Hossain et al., 2005), whereas other studies do not support this result (García and Monterrey, 1993; Aljifri and Hussainey, 2007). Other widely studied factors, such as profi tability, leverage, auditor or industry, show more contradictory results. There is no consensus in previous research on how profi tability infl uences disclosure. Giner (1997) documents a positive relationship, while some others do not (McNally et al., 1982; Raffournier, 1995; Prencipe, 2004). A negative association between profi tability and disclosure level has also been documented (Wallace et al., 1994). Research on forward-looking information disclosure also shows mixed results (Hossain et al., 2005; Aljifri and Hussainey, 2007). The results from research on leverage and disclosure are also confusing: some authors fi nd a signifi cant relationship (Mora and Rees, 1996), others fi nd no evidence (Chow and Wong- Boren, 1987; Wallace et al., 1994; Raffournier, 1995, among others). Studies on forwardlooking information disclosure fi nd both a signifi cant relationship (Aljifri and Hussainey, 2007) and no evidence about this variable (Kent and Ung, 2003). Some research fi nds a positive relationship between auditor fi rm and disclosure (Ahmed, 1996; Giner, 1997). Nevertheless, empirical evidence is mixed, since other fi ndings support no such positive relationship (Firth, 1979; McNally et al., 1982; Wallace et al., 1994). Regarding industry, whereas some studies show a positive relationship between industry and voluntary disclosure (Cooke, 1992), others do not (McNally et al., 1982; Meek et al., 1995). Mixed results are also found in forward-looking information studies (García and Monterrey, 1993; Aljifri and Hussainey, 2007). Above all, a major diffi culty faced by researchers in this area is the measurement of disclosure. Although all diverse information attributes should be taken into account to make up a comprehensive disclosure measure, existing disclosure indices generally concentrate on coverage. Moreover, there are no general guidelines for the design of an index, which is usually constructed by the researchers themselves to meet specifi c research needs. As pointed out earlier, the results of prior research are contradictory and hardly comparable. In an attempt to offer some explanation, we study the use of disclosure indices that are designed to capture similar information attributes and their effects on empirical results. 3. SAMPLE SELECTION The sample is composed of the listed companies that make up the Spanish IBEX 35 stock market index during the period 2000-2004. Listed and larger fi rms are more likely to provide voluntary information, which benefi ts further analysis. For a company to be included in the sample, it had to be one of the companies appearing in the index in at least one of those years. Over that period, 36 companies belonged at least once to the set of companies that make up the stock market index. We excluded fi nancial companies due to the fact that their special characteristics could bias the results. The information was gathered for the years: 2002, 2003 and 2004. The fi nal sample included 36 companies; therefore there are 108 fi rm-year observations. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 400 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 4. RESEARCH METHOD We test whether the theories on disclosure determinants are applicable when specifi c information attributes are considered. Several disclosure measures capturing different information attributes are compared in an empirical investigation of voluntary disclosure determinants. The factors causing disclosure in terms of a specifi c information attribute may differ from the reasons that determine another attribute. To this end, we compare disclosure indices that measure information quantity, coverage and quality. In addition, we compare two disclosure indices that capture coverage —but which are designed differently— to study if the disclosure measure design determines the results obtained. The information disclosed by companies was coded according to specifi c criteria that had been previously established when designing the indices (see Section 4.1.). The sources of information examined were the annual reports. Previous research indicates that annual reports provide a good proxy for the level of information disclosed by a company (Botosan, 1997) and that the information contained in annual reports is positively correlated with the information provided in other sources (Land and Lundholm, 1993). The code unit chosen was the sentence, since using sentences for both coding and measurement seems likely to provide complete, reliable and meaningful data for further analysis (Milne and Adler, 1999). The information analysed was the forward-looking information disclosed in the annual report. We focused on only one kind of information since otherwise, coding every sentence in each of the annual reports in the sample would imply an endless task. Forward-looking information was chosen due to its relevance for external users. As Kieso and Weygant (1986) indicate, forward-looking information disclosure is benefi cial for several reasons: since historical information is insuffi cient for appropriate decision making, then information risk would reduce, thereby improving investment decisions and future cash-fl ows estimation. Moreover, forward-looking information is included in most of the corporate disclosure frameworks. The usefulness of forward-looking fi nancial information for investors has been pointed out in several reports such as the Trueblood Report (AICPA, 1973) and the Corporate Report (ICAEW, 1975), and more recently by the FASB (2001) report titled Improving Business Reporting: Insights into Enhancing Voluntary Disclosure, and by the ICAEW (2002) report titled Prospective Financial Information: Guidance for UK Directors, and by the CICA (2002) report titled Management’s discussion and analysis; Guidance on preparation and disclosure. Finally, we decided to analyze forward-looking information thanks to the availability of a more complex measure that captures different dimensions specifi cally related to forward-looking information disclosure quality (designed by Beretta and Bozzolan, 2005). The coding process started with the execution of a preliminary test so as to set up several coding rules. Two annual reports (excluded from the sample) were examined independently by two different researchers, with the purpose of identifying possible disagreements and of measuring reliability. The whole sample (108 annual reports) was then coded according to the criteria established when designing each index. At the end of the process, a reliability test was performed by employing the alpha agreement coeffi cient proposed by Krippendorf (1980). Once the whole sample was coded and the robustness of this process was verifi ed, an empirical analysis was performed through a regression analysis which examined forward-looking information determinants. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 401 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 4.1. DESIGN OF THE INDICES We compare several indices that capture different information attributes. Disclosure determinants may differ depending on the information attribute that is being considered. Traditional disclosure theories are satisfi ed when the term «higher disclosure» is defi ned in general terms, but might not be satisfi ed if more complex information attributes are considered. Theories are expected to be satisfi ed when measuring information quantity. However, companies perceive higher costs when providing more specifi c information, and the determinants of attributes related to information richness might not agree with existing disclosure theories. To this end, we use three indices that differ on the specifi city of the information attributes captured (from simpler to more complex and detailed). First, we chose a simple index that captures information quantity (quantity index). Second, given that most of the indices employed in the prior literature capture informational coverage, we designed an index (scope index) that refl ects coverage of specifi c topics. Third, we include in the analysis the multidimensional index proposed by Beretta and Bozzolan (2005) (2), which is claimed to be a proxy for disclosure quality (named quality index). Moreover, each of the dimensions that make up the quality index is also considered in the analysis. In addition, we suggest that the previous contradictory results from research on disclosure determinants might be caused by the use of different disclosure measures. To evaluate this claim, we compare two indices that are similar in design; the scope index and the coverage dimension of the quality index. The coding process required the analysis of 108 annual reports in order to fi nd sentences with forward-looking information. These sentences were coded according to the criteria mentioned below for each index. 4.1.1. Quantity index The quantity index (QNI) measures the amount of information by taking into account the number of sentences with forward-looking information. Its purpose is to capture disclosure quantity —which is standardised in order to make it relative to the sample— in accordance with the following formula: QNIi = (fl i – min) / (Max – min) (1) Where: fl i: is number of sentences with forward-looking information disclosed by company i. Max: is maximum number of sentences with forward-looking information disclosed by a company across the sample. min: is minimum number of sentences with forward-looking information disclosed by a company across the sample. The quantity index ranges between 0 and 1. (2) Beretta and Bozolan (2005) quality index robustness has been assessed by different methods in order to support its validity to capture disclosure quality. An updated version of the working paper (Beretta and Bozolan, 2005) has been published recently (Beretta and Bozzolan, 2008; Beretta et al., 2009). Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 408 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 TABLE 3 MULTIVARIATE LINEAR REGRESSIONS (FIXED EFFECTS MODEL) (108 fi rm-year observations. Years 2002, 2003 & 2004, t-statistics in parentheses) DIi = α + β1SIZEi + β2 ROEi + β3 LEVi + β4 OR_GROWTHi + β5 FE_LISTINGi + ε DEPENDENT VARIABLEA: DISCLOSURE INDEX (DII) Explanatory VariablesbQuantity index Scope index Quality index Coverage dimension Intercept-0.011 (-0.206) 0.502*** (7.809) 0.407*** (9.858) 0.401*** (7.680) SIZE 0.672*** (7.377) 0.347*** (3.037) 0.386*** (3.153) 0.581*** (5.315) ROE 0.014 (0.197) -0.075 (-0.852) 0.043 (0.459) 0.019 (0.231) LEV 0.226*** (3.188) 0.117 (1.325) 0.108 (1.146) 0.158* (1.873) OR_GROWTH 0.166** (2.295) -0.009 (-0.105) 0.038 (0.395) -0.070 (-0.814) FE_LISTING -0.160* (-1.756) 0.174 (0.904) -0.166 (-1.364) -0.113 (-1.035) Adjusted R20.499 0.216 0.100 0.276 F (p-value) 16.082 (0.000) 5.200 (0.000) 2.074 (0.013) 9.172 (0.000) a The quantity index measures information amount taking into account the number of sentences with forward-looking information. The scope index is obtained by dividing the number of information items disclosed by a company by the total number of information disclosure items that might be disclosed [six forward-looking information items: environment (1), evolution (2), goals, strategies and business policies (3), information related to dividends policy (4), information about future investments (5), organization and corporate structure (6)]. The quality index (Beretta and Bozzolan, 2005) is an aggregated index composed of two dimensions, relative quantity and richness. The relative quantity index (RQT) is calculated as the difference between observed disclosure for a company and estimated disclosure. The richness index (RCN) is designed to capture disclosure quality, and is calculated by averaging two new dimensions associated to information richness: width (WID) and depth (DEP). Width (WID) depends on both coverage (COV) of important topics and on disclosure dispersion (DIS) (which measures concentration of items disclosed). Depth (DEP) depends on the type of measure used in an information unit, on the communication of the economic sign of the disclosed items and on information profi le. b SIZE is defi ned as operating revenues; ROE is return on equity; LEV is calculated dividing total debt between total assets; OR_ GROWTH is growth in operating revenues; FE_LISTING is a dummy variable with a value of 1 if the company is listed in the New York Stock Exchange or the London Stock Exchange, and 0 otherwise. Operating revenues are in millions of euros. * Signifi cant at 10% level; ** Signifi cant at 5% level; *** Signifi cant at 1% level. The quantity index was defi ned in general terms, simply by capturing information quantity. When using the quantity index (QNI), results in Table 3 indicate that size, leverage, and growth are positively related to forward-looking information disclosure, at 1%, 1% and 5%, respectively. Similar to the evidence found in previous studies, this result supports the hypothesis, suggested by agency, signalling, and political process theories, that larger fi rms have incentives to reveal more voluntary information. It also corroborates the agency theory prediction that highly-leveraged companies will provide more information. Finally, signifi cance of the growth variable confi rms the signalling theory hypothesis that companies with high growth rates have incentives to increase disclosure to convey that information to the market. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 409 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 TABLE 4 MULTIVARIATE LINEAR REGRESSIONS: QUALITY INDEX DIMENSIONS (FIXED EFFECTS MODEL) (108 fi rm-year observations. Years 2002, 2003 & 2004, t-statistics in parentheses) DIi = α + β1SIZEi + β2 ROEi + β3 LEVi + β4 OR_GROWTHi + β5 FE_LISTINGi + ε DEPENDENT VARIABLEa: DISCLOSURE INDEX (DIi) Explanatory Variablesb Coverage (COV) Dispersion (DIS) Width (WID) Economic Sign (ESM) Outlook Profi le (OTL) Depth (DEP) Intercept0.401*** (7.680) 0.481*** (11.031) 0.441*** (9.946) 0.581*** (8.289) 0.790*** (13.086) 0.680*** (12.704) SIZE 0.581*** (5.315) 0.542*** (4.710) 0.605*** (5.449) 0.006 (0.050) -0.010 (-0.083) -0.002 (-0.014) ROE 0.019 (0.231) 0.132 (1.488) 0.074 (0.871) 0.125 (1.265) 0.095 (1.016) 0.134 (1.394) LEV 0.158* (1.873) 0.107 (1.206) 0.145* (1.700) -0.020 (-0.205) -0.127 (-1.360) -0.079 (-0.820) OR_GROWTH -0.070 (-0.814) -0.073 (-0.800) -0.076 (-0.870) -0.125 (-1.228) -0.310*** (-3.243) -0.260** (-2.639) FE_LISTING -0.113 (-1.035) -0.380*** (-3.303) -0.248** (-2.234) 0.118 (0.915) -0.140 (-1.052) -0.004 (-0.032) Adjusted R20.276 0.198 0.253 -0.003 0.107 0.052 F (p-value) 9.172 (0.000) 6.278 (0.000) 8.244 (0.000) 0.933 (0.463) 3.565 (0.005) 2.175 (0.063) a The dependent variables in the different models are the dimensions that compose the quality index of Beretta and Bozzolan (2005). The quality index is an aggregated index composed of two dimensions, relative quantity and richness. The relative quantity index is calculated as the difference between observed disclosure for a company and estimated disclosure. The richness index (RCN) is designed to capture disclosure quality, and is calculated by averaging two new dimensions associated to information richness: width (WID) and depth (DEP). Width (WID) depends on both coverage (COV) of important topics and on disclosure dispersion (DIS) (which measures concentration of items disclosed). Depth (DEP) depends on the type of measure used in an information unit and on the communication of the economic sign of the disclosed items (EMS index) and on information profi le (OTL). The ESM index is calculated by aggregating the MSR and ES sub-indices. The MSR index captures the type of measure, which is defi ned as the ratio between the number of phrases (forward-looking) containing a measure and the total number of pieces of forward-looking information considered in the annual report. The expected impact on future performance is measured by the economic sign index (ES), which counts the number of pieces of forward-looking information disclosed containing an indication of the sign of effective or expected impact on actual or future performance of the item disclosed out of the total number of phrases (forward-looking) considered in the annual report. The outlook profi le (OTL) attribute is calculated by aggregating two other dimensions. Information nature is captured by the ACP index, which is calculated dividing the number of sentences with forward-looking information on decisions, programs, and actions into the total number of sentences with forward-looking information. Information relevance is captured by the FL index, which considers sentences with forward looking information that include information useful to make earnings forecasts (FL). b SIZE is defi ned as operating revenues; ROE is return on equity; LEV is calculated dividing total debt between total assets; OR_ GROWTH is growth in operating revenues; FE_LISTING is a dummy variable with a value of 1 if the company is listed in the New York Stock Exchange or the London Stock Exchange, and 0 otherwise. Operating revenues are in millions of euros. . * Signifi cant at 10% level; ** Signifi cant at 5% level; *** Signifi cant at 1% level. When the scope (SCI) index or the coverage (COV) dimension is used as the dependent variable, determinants of forward-looking information disclosure differ, since only size (and leverage for the coverage dimension) appears as signifi cant. When disclosure is measured in terms of scope/coverage, disclosure theories are supported but only insofar as the hypotheses that refer to size and leverage. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 410 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 When the dependent variable in the regression model is the quality index (QLI), size appears as the only determinant of an increase in disclosure quality. Moreover, the R2s values indicate that the model explains almost 50% of quantity, but just 10% of quality, which shows again that some of the theoretical predictions on disclosure determinants are not confi rmed when specifi c information attributes are considered. Nevertheless, the quality index was designed as composed of several disclosure dimensions that capture specifi c information attributes. The company characteristics that determine an increase in disclosure quality differ depending on the information attribute considered. The results of the estimation of the regression model by considering each of the quality index dimensions as the dependent variable are shown in Table 4. Some of the results for the dimensions that form the quality index are inconsistent with disclosure theories, making it necessary to bear in mind that those theories refer to the total disclosure for a company, and not to specifi c disclosure attributes. The width (WID) dimension and its components —coverage (COV) and dispersion (DIS)— show, as expected according to disclosure theories, a positive relationship with size. This coeffi cient is signifi cant at 1% for all these dimensions. However, the specifi c dimensions that capture information profi le, which are closely associated to information richness, show unexpected results. The coeffi cient on size is not signifi cant when disclosure is measured through the depth (DEP) indices, including the economic sign (ESM) and outlook profi le (OTL) dimensions. This fi nding contradicts the disclosure theories that support a positive association between size and disclosure. Larger companies disclose information of a higher quality in terms of coverage and dispersion, but fail to provide more information about plans, decisions or actions that will affect the future development of the company or information that is useful to make forecasts. This supports our hypothesis that when more detailed attributes of the information are considered, then the existing disclosure theories may be rendered invalid. When the components of the width dimension are considered, it is found that foreign exchange listing holds a negative relationship with forward-looking information disclosure when it is measured with the dispersion and width indices. A possible explanation is the design of this variable, which implies that only a few companies are assigned a value of 1. Regarding the components of the depth dimension, operating revenue growth shows a negative association with forward-looking information disclosure (outlook profi le (OTL) and depth (DEP) dimensions). The coeffi cient of the depth index is negative (-0.260) and signifi - cant at 5%. This is consistent with some theoretical arguments which claim that companies with high growth rates might be reluctant to provide voluntary information to prevent competitive disadvantages (Prencipe, 2004). However, this result is unexpected according to signalling theory, since it indicates that companies with higher growth rates provide less information of a forward-looking nature about plans, decisions, actions and less information that may be useful for forecasting purposes. The depth dimension is designed in terms of information richness, without considering information quantity or coverage. The results obtained for this measure contradict theoretical predictions, since the depth index and its components are not much explained by the selected characteristics. Different information Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 411 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 attributes are explained by different determinants, as it was expected. As a consequence, different disclosure attributes must be explained with different theories. The second objective was to examine how disclosure measure design infl uences the results obtained from research on disclosure determinants. To this end, two indices that measure disclosure in terms of coverage, which both pay attention to the number of topics disclosed for a given company, are compared. The two indices are the scope (SCI) index and the coverage dimension (COV) of the quality index, which are designed in a similar fashion. The results of the estimation of the regression model with each of these two indices as the dependent variable are shown in Table 3. The determinants of coverage are not the same as the determinants of scope, since when measuring forward-looking information disclosure through coverage (COV), then size and leverage seem to be related to disclosure, but when it is measured with the scope index (SCI), then only size appears as signifi cant. Moreover, the coeffi cient on size differs (0.347 when the dependent variable is the scope index and 0.581 if the coverage dimension is used). Therefore, even though similar indices are used (the coverage and the scope indices), results differ, which in turn indicates the impact that the choice of a particular disclosure measure has on this type of study. 6. CONCLUDING REMARKS Research on the determinants of disclosure has so far failed to provide conclusive results. The complexity underlying disclosure measures may explain the confl icting evidence. Disclosure indices that capture diverse information attributes and have different designs coexist in this literature. We study whether the determinants of a specifi c information attribute differ from those that determine the disclosure of another information attribute. This being the case, disclosure theories will not unfailingly hold true, as they characterize disclosure in general terms, without making a clear distinction between specifi c information attributes. Moreover, we also explore the infl uence of disclosure index design on the results of the research on disclosure determinants. If different designs of measures that capture the same information attribute cause differences in determinants, then the evidence will highlight the need to search for measures of a more comparable nature. An empirical analysis of forward-looking information disclosure determinants for a sample of listed Spanish companies was performed. Several disclosure indices, which measure disclosure of different information attributes (such as quantity, coverage and quality), were compared. In addition, we also compared two indices that capture the same information attribute (coverage, since this is the most commonly measured information attribute in existing literature) but which are different in design. Our results show that size is the only determinant that is always signifi cant, no matter which disclosure attribute is being analysed. However the determinants of information attributes of a more specifi c nature are different to those that infl uence attributes of a less specifi c nature. Size, leverage and growth determine information quantity. However, when disclosure is characterized in terms of quality, size is the only variable that explains disclosure. In addition, if attention is paid to the quality index components, no variable appears as a determinant of disclosure for the depth dimension, which is the dimension Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 412 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 that captures information richness. Companies perceive higher costs when providing more detailed information, and the reasons for disclosing information attributes that are more related to information richness differ from those that infl uence the disclosure of generic information attributes. Consequently, general disclosure theories fail to hold true for all the information attributes. Since the determinants of disclosure of some information attributes do not coincide with the determinants of another attributes, disclosure theories should therefore pay specifi c attention to the information attribute that is being measured. Our results help to clarify why prior evidence appears to be contradictory and to refi ne theoretical arguments on the reasons behind disclosure decisions. Disclosure indices should clearly defi ne the information attribute that is captured, as it will infl uence the results obtained. Furthermore, regardless of the information attribute that is measured, disclosure measure design infl uences results as well. Our empirical evidence indicates that results differ when two indices that capture the same information attribute, but that are different in design, are used. Empirical results are conditional upon the measure used. Therefore, care must be taken when comparing the results from disclosure studies that employ different measures of the same attribute. In terms of the implications for future research, our results highlight the need to search for a consensus on index design that would help towards comparability and enable results of a more conclusive nature to be attained. Another future development could be the study of the impact of using different disclosure indices on the effects of disclosures, such as cost of capital, analysts’ forecasts and stock returns. 7. REFERENCES AMERICAN INSTITUTE OF CERTIFIED PUBLIC ACCOUNTANTS (AICPA). 1973. Trueblood Report. New York. AMERICAN INSTITUTE OF CERTIFIED PUBLIC ACCOUNTANTS (AICPA). 1994. Improving Business Reporting-A customer focus. New York. AHMED, K. 1996. Disclosure policy choice and corporate characteristics: a study of Bangladesh. Asia- Pacifi c Journal of Accounting 3 (1): 183–203. AHMED, K., and COURTIS, J. K. 1999. Associations between corporate characteristics and disclosure levels in annual reports: a meta-analysis. British Accounting Review 31: 35-61. ALJIFRI, K., and HUSSAINEY, K. 2007. The determinants of forward-looking information in annual reports of UAE companies. Managerial Auditing Journal 32 (9): 881-894. ALSAEED, K. 2005. The association between fi rm-specifi c characteristics and disclosure: the case of Saudi Arabia. Journal of American Academy of business 7 (1): 310-321. ARCHAMBAULT, J. J., and ARCHAMBAULT, M. E. 2003. A multinational test of determinants of corporate disclosure. The International Journal of Accounting 38 (2): 173–194. BALATA, P., and BRETON, G. 2005. Narratives vs. numbers in annual report: are they giving the same message? Review of Accounting and Finance 4 (2): 5-25. BEATTIE, V.; MCINNES, B., and FEARNLEY, S. 2004. A methodology for analysing and evaluating narratives in annual reports: a comprehensive descriptive profi le and metrics for disclosure quality attributes. Accounting Forum 28: 205-236. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 413 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 BELSLEY, D.; KUH, E., and WELSCH, R. 1980. Regression diagnostics: identifying infl uential data and sources of collinearity. New York, NY: John Wiley and Sons. BERETTA, S., and BOZZOLAN, S. 2005. Forward looking disclosure and fi nancial analysts´ forecasts: an international study. Working paper, Bocconi University and University of Padova. BERETTA, S., and BOZZOLAN, S. 2008. Quality versus quantity: the case of forward-looking disclosure. Journal of Accounting, Auditing and Finance 23 (3): 333-375. BERETTA, S.; BOZZOLAN, S., and TROMBETTA, M. 2009. Forward looking disclosure, fi nancial verifi ability and analysts' forecasts: A study of cross-listed European fi rms. European Accounting Review 18 (3): 435-473. BHOJRAJ, S. 1999. Management earnings forecasts: A theoretical and empirical analysis. Doctoral dissertation, University of Florida. BOTOSAN, C. A. 1997. Disclosure level and the cost of equity capital. The Accounting Review 72 (3): 323-349. BUZBY, S. 1975. Company size, listed versus unlisted stocks, and the extent of fi nancial disclosure. Journal of Accounting Research 13 (1): 16-37. CANADIAN INSTITUTE OF CHARTERED ACCOUNTANTS (CICA). 2002. Management's discussion and analysis: Guidance on preparation and disclosure. CERBIONI, F., and PARBONETTI, A. 2007. Exploring the effects of corporate governance on intellectual capital disclosure: an analysis of European biotechnology companies. European Accounting Review 16 (4): 791-826. CHAVENT, M.; DING, Y.; FU, L.; STOLOWY, H., and WANG, H. 2006. Disclosure and determinants studies: an extension using the Divisive Clustering Method (DIV). European Accounting Review 15 (2): 181-218. CHEN, C. J. P., and JAGGI, B. 2000. Association between independent non-executive directors, family control and fi nancial disclosures in Hong Kong. Journal of Accounting and Public Policy 19: 285-310. CHOW, C. W., and WONG-BOREN, A. 1987. Voluntary fi nancial disclosure by Mexican corporations. The Accounting Review 62 (3): 553-541. COOKE, T. E. 1989. Disclosure in the corporate annual reports of Swedish companies. Accounting and Business Research 19 (74): 113-124. Spring. COOKE, T. E. 1992. The impact of size, stock market listing and industry type on disclosure in the annual reports of Japanese listed corporations. Accounting and Business Research 22 (87): 229-237. DEPOERS, F. 2000. A cost-benefi t study of voluntary disclosure: some empirical evidence from French listed companies. European Accounting Review 9 (2): 245-263. DURBIN, J., and WATSON, G. 1950. Testing for serial correlation in least squares regression-I. Biometrika, 37: 409-428. ENTWISTLE, G. M. 1999. Exploring the R&D disclosure environment. ACCOUNTING HORIZONS 13 (4): 323- 341. FINANCIAL ACCOUNTING STANDARDS BOARD (FASB). 2001. Improving Business Reporting: Insights into Enhancing Voluntary Disclosure. Steering Committee Report. Business Reporting Research Project. FIRTH, M. 1979. The impact of size, stock market listing, and auditors on voluntary disclosure in corporate annual reports. Accounting and Business Research Autumn: 273-280. GARCÍA, M. A., and MONTERREY, J. 1993. La revelación voluntaria en las compañías españolas cotizadas en Bolsa. Revista Española de Financiación y Contabilidad 23 (74): 53-70. GARCÍA-MECA, E., and MARTÍNEZ, I. 2004. Divulgación voluntaria de información empresarial: índices de revelación. Partida Doble 157: 66-77. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta 414 ARTÍCULOS DOCTRINALES Disclosure theories and disclosure measures REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD. Vol. XXXIX, n.º 147 · julio-septiembre 2010 GARCÍA-MECA, E.; PARRA, I.; LARRÁN, J., and MARTÍNEZ, I. 2005. The explanatory factors of intellectual capital disclosure to fi nancial analysts. European Accounting Review 14 (1): 63-94. GINER, B. 1995. La divulgación de información fi nanciera: una investigación empírica. Instituto de Contabilidad y Auditoría de Cuentas. Ministerio de Economía y Hacienda, Madrid. — 1997. The infl uence of company characteristics and accounting regulation on information disclosed by Spanish fi rms. European Accounting Review 6 (1): 45-68. GREENE, W. 1997. Econometric analysis. Prentice Hall, International Edition. HEALY, P. M., and PALEPU, K. G. 2001. Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics 31: 405-440. HOSSAIN, M.; AHMED, K., and GODFREY, J. M. 2005. Investment opportunity set and voluntary disclosure of prospective information: A simultaneous equations approach. Journal of Business Finance and Accounting, 32: 871-907. HUSSAINEY, K.; SCHLEICHER, T., and WALKER, M. 2003. Undertaking large-scale disclosures studies when AIMR-FAF ratings are not available: the case of prices leading earnings. Accounting and Business Research 33 (4): 275-294. INSTITUTE OF CHARTERED ACCOUNTANTS IN ENGLAND AND WALES (ICAEW). 1975. Corporate Report. INSTITUTE OF CHARTERED ACCOUNTANTS IN ENGLAND AND WALES (ICAEW). 2002. Prospective fi nancial information: Guidance for UK directors. JENSEN, M., and MECKLING, W. 1976. Theory of the fi rm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3 (4): 305-360. KIESO, D. E., and WEYGANDT, J. J. 1986. Intermediate Accounting. John Wiley & Sons. New York. KENT, P., and UNG, K. 2003. Voluntary disclosure of forward-looking earnings information in Australia. Australian Journal of Management 28 (3): 273-285. KRIPPENDORF, K. 1980. Content analysis: an introduction to its methodology. Beverly Hills, CA: Sage Publications. LANG, M., and LUNDHOLM, R. 1993. Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research 31 (2): 246-271. LEV, B., and PENMAN, S. H. 1990. Voluntary forecast disclosure, nondisclosure, and stock prices. Journal of Accounting Research 31 (2): 49-76. LI, J.; PIKE, R., and HANNIFA, R. 2008. Intellectual capital disclosure and corporate governance structure in UK fi rms. Accounting and Business Research 38 (2): 137-159. MALONE, D.; FRIES, J., and JONES, T. 1993. An empirical investigation of the extent of corporate fi nancial disclosure in the oil and gas industry. Journal of Accounting, Auditing and Finance 8: 249-273. MCNALLY, G. M.; ENG, L. H., and HASSELDINE, H. R. 1982. Corporate fi nancial reporting in New Zeland: an analysis of user preferences, corporate characteristics and disclosure practices for discretionary information. Accounting and Business Research 13 (49): 11-20. MEEK, G. K.; ROBERTS, C. B., and GRAY, S. J. 1995. Factors infl uencing voluntary annual report disclosures by U.S., U.K. and continental European multinational corporations. Journal of International Business Studies 26 (3): 555-572. MILNE, M. J., and ADLER, R. W. 1999. Exploring the reliability of social and environmental disclosures content analysis. Accounting, Auditing & Accountability Journal 12 (2): 237-256. MORA, A., and REES, W. 1996. Un análisis empírico sobre la práctica de la consolidación contable en las empresas españolas. Revista Española de Financiación y Contabilidad 25 (86): 161-178. Francisco Bravo Urquiza, María Cristina Abad Navarro and Marco Trombetta Disclosure theories and disclosure measures ARTÍCULOS DOCTRINALES 415 SPANISH JOURNAL OF FINANCE AND ACCOUNTING. Vol. XXXIX, n.º 147 · july-september 2010 PRENCIPE, A. 2004. Proprietary costs and determinants of voluntary segment disclosure: evidence from Italian listed companies. European Accounting Review 13 (2): 319-340. RAFFOURNIER, B. 1995. The determinants of voluntary fi nancial disclosure by Swiss listed companies. European Accounting Review 4 (2): 261-280. ROBB, S. W. G.; SINGLE, L. E., and ZARZESKI, M. T. 2001. Nonfi nancial disclosures across Anglo-America countries. Journal of International Accounting 10 (1): 71-83. RODRÍGUEZ PÉREZ, G. 2004. Factores explicativos de la revelación voluntaria de información sobre fuentes de ventaja competitiva empresarial. Revista Española de Financiación y Contabilidad 23 (122): 705-739. SINGHVI, S., and DESAI, H. 1971. An empirical analysis of the quality of corporate fi nancial disclosure. The Accounting Review 46: 129-138. January. WALLACE, R. S. O.; NASER, K., and MORA, A.. 1994. The relationship between the comprehensiveness of corporate annual reports and fi rm characteristics in Spain. Accounting and Business Research 25 (97): 41-53. WALLACE, R. S. O., and NASER, K. 1995. Firm-specifi c determinants of comprehensiveness of mandatory disclosure in the corporate annual reports of fi rms listed on the stock exchange of Hong Kong. Journal of Accounting and Public Policy 14 (4): 311-368. WATTS, R. L., and ZIMMERMAN, J. L. 1986. Positive Accounting Theory. Englewood Cliffs, NJ: Prentice- Hall. WHITE, H. 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48 (4): 817-838. WILLIAMS, S. M. 1999. Voluntary environmental and social accounting disclosure practices in the Asia- Pacifi c region: An international empirical test of political economy theory. The International Journal of Accounting 34 (2): 209-238. Discussion on Disclosure theories and disclosure measures (by F. Bravo Urquiza, M. C. Abad Navarro and M. Trombetta) Juan Manuel García Lara. Universidad Carlos III de Madrid 1. WHAT IS DISCLOSURE QUALITY? In the presence of information asymmetries, potential investors cannot properly evaluate the expected profi tability and risk of the available investment opportunities. Firms provide information to reduce the adverse effects of these information asymmetries (ineffi cient pricing and adverse selection problems, as described by Akerlof, 1970). This provision of information from fi rms to the potential providers of capital funds is done through the fi nancial statements and also through other communication channels. An impressive amount of literature has analysed how the information reported compulsorily in the fi nancial statements should be to reduce the problems created by information asymmetries. One example of this literature is the large stream of studies about whether accounting numbers should be more or less conservative (LaFond & Watts, 2008, García Lara, García Osma & Penalva, 2011). Similarly, there are many studies analysing whether and how voluntary disclosure contributes to reduce adverse selection problems. A recent example of this voluntary disclosure literature is Francis, Nanda & Olsson (2008), who study whether voluntary disclosure contributes to decrease cost of capital. An increased cost of capital is one of the clearest examples of the deadweight losses introduced by information asymmetries. Even if the problems of adverse selection are resolved through the ex-ante provision of information, there is still an unresolved issue. Once that investors allocate their funds to the fi rms, current investors will require information about fi rm’s activities to avoid being expropriated by the managerial team or the dominating shareholder. Consequently, besides the role of (accounting) information to resolve problems related to the allocation of resources in the economic system, accounting information plays also a role in resolving agency confl icts between the different parties in the fi rm (Jensen & Meckling, 1976). This provision of information to monitor managerial decisions can also be done through the fi - nancial statements, and also through other voluntary channels. Prior accounting research has analysed which should be the desirable characteristics of the information included in the fi nancial statements to permit a better monitoring over managers’ decisions. Recent examples of this literature include García Lara, García Osma & Penalva (2009) or Biddle, Hilary & Verdi (2009). Regarding disclosure, there is also a large stream of research stu- Recibido 5-04-10 · Aceptado 15-05-10 · Copyright © 2001 Asociación Española de Contabilidad y Administración de Empresas · ISSN: 0210-2412 REVISTA ESPAÑOLA DE FINANCIACIÓN Y CONTABILIDAD Vol. XXXIX, n.º 147 · julio-septiembre 2010 · pp. 417-420 ARTÍCULOS DOCTRINALES 417