Accounting Conservatism in Portugal: Similarities and Differences Facing Germany and the United Kingdom

This paper examines the existence of conservative practices in the Portuguese accounting system, and whether these conservative practices affect the comparability of financial information provided by companies. We particularly examine whether the book value figure can be understated due to conservative practices to protect creditors' interests (balance sheet conservatism) and whether accountants delay the recognition in earnings of good news, while they recognize immediately bad news (earnings conservatism). Using a Basu (1997) type reverse regression and a simple adaptation of the Ohlson (1995) valuation model, the paper gives evidence concerning the existence of both definitions of conservatism in Portuguese accounting practices. A sample of non-financial Portuguese, German and British companies was used. Our results also show the larger earnings conservatism of British firms relative to Portugal and Germany, and surprisingly, that Portugal is more BS (larger understatement of shareholders' equity) conservative than the United Kingdom. The results have implications for accounting standard setting and they can be useful for both the European Commission and the IASB since they provide some insight into the properties of accounting figures in Portugal.


INTRODUCTION: CONSERVATISM AND THE COMPARABILITY OF ACCOUNTING INFORMATION
This paper examines the existence of conservative practices in the Portuguese accounting system, and whether these conservative practices affect the comparability of financial information provided by Portuguese companies with respect to that disclosed by companies in other countries in the European Union.We particularly examine whether the book value figure can be understated due to conservative practices to protect creditors' interests and whether accountants delay the recognition in earnings of economic events that affect the firm positively, while immediately recognizing those that affect negatively.
The motivation that has encouraged us to begin this study relates directly to the European accounting harmonization process.Given that the European Union is attempting to set up a new single securities market in Europe, the financial information provided by the companies of countries that will join this new institution must be completely comparable.This is one of the main reasons why the European Commission requires all listed European Union companies to prepare their consolidated accounts following the new IASB standards.These movements towards a single stock market have led to the alliance between the London and Frankfurt Stock Exchanges, as well as to the creation of Euronext, a new securities market formed by the merger of the Amsterdam, Brussels and Paris Stock Exchanges.The Stock Exchange of Lisbon and Oporto (Bolsa de Valores de Lisboa e Porto) joined this institution, and therefore Portugal is involved directly in the movements towards the future European single stock market.Our objective is thus to provide some empirical evidence on whether Portuguese companies' accounting information is comparable to that of the rest of Europe, and so, to explore whether the differences in conservatism make a claim for the usage of a common set of standards that could reduce these differences or whether they are not necessary.
This study focus on two rather different definitions of conservatism that affect accounting numbers.
Firstly, we address the definition of conservatism raised by Feltham and Ohlson (1995).This definition views accounting as being conservative if the market value of the firm exceeds on average the book value, i. e., if the market to book ratio is consistently greater than one.Feltham and Ohlson (1995), Beaver and Ryan (2000) and Zhang (2000) have theoretically described the effects of this notion of accounting conservatism, which we denominate balance sheet conservatism.Following Zhang (2000), we assume that there exists balance sheet conservatism if… Where oa is operating assets and V is the market value of operating assets.Zhang (2000) describes conservatism in terms of operating assets, as Feltham and Ohlson (1995) also did, given that we can assume that perfect or unbiased accounting holds for financial assets and liabilities since there are perfect markets for them.However, although the existence of balance sheet conservatism is attributable to operating assets, the relation continues to hold if we use the total book value of the firm and the market value of the firm (market capitalization).Few studies have tested the existence of this type of accounting conservatism.Stober (1996) and Givoly and Hayn (2000) analyze the market to book ratio in the United States and conclude that there is a conservatism bias, i. e., that it is always greater than one, and the level of balance sheet conservatism has increased consistently during the last three decades.Joos and Lang (1994) analyze the book to market ratio in Germany, France and the UK, for the period covering [1982][1983][1984][1985][1986][1987][1988][1989][1990], and their results show that it is consistently lower than one.They also find (using a Wilcoxon test) that Germany shows a statistically significant larger balance sheet conservatism bias than France and the UK.Finally Joos (1997) uses a simplification of the Bernard (1995) empirical application of the theoretical models by Ohlson (1995) and Feltham and Ohlson (1995), obtaining similar results to those in Joos and Lang (1994).
The other definition of conservatism that we address is that stated by Basu (1997), who interprets conservatism as capturing accountants' tendency to require a higher degree of verification for recognizing good news than bad news in earnings.Basu (1997) uses the reverse regression approach proposed by Beaver, Lambert and Ryan (1987) and proposes, as a proxy for news, the rate of return of the firm.For example, in all GAAP regimes unrealized losses are recognized earlier than unrealized gains.This is the case of a change in the expected lifespan of a tangible fixed asset.If the expected lifespan of a fixed asset increases, the firm is economically better off, but the depreciation charges that would have been taken in the current and future periods are spread out over the remaining life of the asset.However, if there is a decline in the expected lifespan of the asset, accountants normally record an asset impairment that produces a sharp reduction in earnings in the current period.His results (he analyses only the US accounting system) are consistent with the existence of this type of conservatism, which we define as earnings conservatism.Another important paper in this line of research is Ball, Kothari and Robin (2000).They extend Basu's analysis to seven countries: the United States, Canada, Australia, the United Kingdom, France, Germany and Japan; and they argue that common-law based countries are more earnings conservative than code-law based countries.They explain this difference by pointing to the different economic role of financial statements in both types of countries.In common-law based countries, the ownership of the company is spread over a wide number of shareholders, who can only know how the firm is performing through financial statements.This is why they demand that the information reflected in financial statements is timely and exact.In these countries, investors will be willing to suit managers or auditors in case that they do not disclose bad news in a timely way through financial statements.Furthermore, in these countries legislation seems to be to some extent favorable to plaintiffs (shareholders).Conversely, in code-law based countries the main providers of finance are financial institutions.There is a very close relationship between the company and the main users of financial statements.This is why there is not a demand for exact and timely information in financial statements, since users already know this information before it is disclosed through financial statements.The authors also use a reverse regression, incorporating into the model the differential effect of each country through dummy variables.Their results show the extreme conservatism of the US accounting model relative to France, Germany and Japan.The United Kingdom is in an intermediate position between the US and France, Germany and Japan.Other studies analyzing the effect of earnings conservatism are Pope and Walker (1999) who focus on the differences between the United States and the United Kingdom, and Givoly and Hayn (2000) and Holthausen and Watts (2001) who analyze the evolution of earnings conservatism.These two latest studies conclude that there has been a consistent increase in this type of conservatism over the last 30 years and the latter finds that the level of earnings conservatism is associated with the level of litigation risk.Finally, Giner and Rees (2001) analyze the asymmetry in several European countries, and Gigler and Hemmer (2001) give an alternative explanation for the asymmetry, which does not imply the existence of conservative practices.Other more recent papers on accounting conservatism include Watts (2003a and b), Ryan and Zarowin (2003), Ball, Robin and Wu (2003), Raonic, McLeay and Asimakopoulos (2004), García Lara and Mora (2004), Roychowdhury andWatts (2004), García Lara, García Osma andMora (2005), Ball and Shivakumar (2005) and Dietrich, Muller and Riedl (2005).
The differences we find between Portugal and Germany and the UK will be affected by the compulsory use of the IASB standards (International Financial Reporting Standards, IFRS) from 2005 onwards.However, it is unlikely that the differences will disappear.With respect to balance sheet conservatism, the IFRS permit the use of fair value accounting for assets.However, the decision to revaluate an asset will depend on managers' incentives, which are likely to differ between countries.With respect to earnings conservatism, the situation is similar.As pointed out by García Lara et al. (2005) managers in Germany have very different incentives than managers in the UK.While managers tend to manage earnings upwards in the UK to beat or meet certain earnings thresholds, in Germany they will very likely be more concerned with reporting low earnings numbers due to dividend, taxation and investment policies, and with smoothing earnings.These incentives will not disappear by using IFRS and consequently differences will remain.Managers' incentives in Portugal are probably closer to those of German managers and the institutional framework and ownership structure of firms is similar.We describe the Portuguese accounting and institutional framework in the following section.

THE PORTUGUESE INSTITUTIONAL FRAMEWORK
Historically, there has been a strong code-law influence on accounting in Portugal.In fact, the Portuguese standard setting process started almost 40 years ago with the establishment of tax laws.As in other continental European countries, the Portuguese accounting regime is based on a codified system of law.The first significant attempt to harmonize accounting practices occurred in 1963 with a profound tax reform which contained some implicit accounting regulations.But only in 1977 was an Official Accounting Plan (POC -Plano Oficial de Contabilidade) revealed, based mainly on the 1957 French Plan Comptable.In the past 25 years, Portugal has witnessed two Official Accounting Plans, several tax laws, fourth and seventh Directive introductions in Portuguese law, and the legislation and revision of the securities market regulation.Professional bodies have been created for the auditing profession and, more recently, for the accounting profession, and an Accounting Standards Board has been also created (1) .
The main factors that influence financial reporting and the Portuguese GAAP have been the Official Accounting Plan, the tax law, the EU Directives (4th and 7th) and IAS.Portuguese accounting standards are issued and enforced by government, mainly by enacting the Official Accounting Plan (where regulations set out the information that companies must report and the principles and rules for the elaboration of that information), establishing taxation rules (that include some accounting practices), and ruling the characteristics of the companies that have to be audited as well as both the audit and accounting professions.
In the following subsections, we summarize the influences on the accounting system in Portugal, analyzing three of the main factors pointed out by Nobes (1983), namely the legal influence on accounting and financial reporting, who are the providers of finance, and the influence of taxation.

The Legal Influence
The Portuguese standard setting process started about 40 years ago with the establishment of tax laws.In fact, only recently did Portugal begin to realize the importance of a comprehensive set of accounting rules.This process took a step forward with the establishment of the CNC (Comissão de Normalização Contabilística -the accounting standards board) and the approval of the first Official Accounting Plan (POC) in 1977, in Decree-law Nr. 47/77.In November, 1989, a Revised Version of the POC was issued to comply with the EU fourth Directive.Besides the Official Accounting Plan, 29 accounting standards (DC -Directrizes Contabilísticas) have been approved, covering specific issues.Accounting standards are developed through a very close due process that involves only the members of the CNC and a few well-known individuals.The due process is weak.The agenda is established internally, there is very little formal information to the public and only a certain period reserved for comments from specific parties, namely the Portuguese Institute of Public Auditors (OROC -Ordem dos Revisores Oficiais de Contas) and Accounting Institute (CTOC -Câmara dos Técnicos Oficiais de Contas).
Regarding the Portuguese Accounting Standards Board (CNC), it was set up in 1974 by law and its structure and functions have been revised several times since then, the last being in 1999.The main objectives of the CNC are standard setting, standard interpretations, mostly by request and representing Portugal in international forums and joint working groups.The CNC has 41 members representing 36 institutions (public entities, professional associations, schools and representatives of economic sectors) and operates, both financially and administratively, under the Ministry of Finance, although it has technical independence.
Portugal's entry into the EU (January 1986) also led to a major reform of commercial and company law.The Companies Law, published in November 1986, is the key amending regulation in the field of company law, substituting the previous legislation (dating back to 1888), and was adapted to EU Directives.
Presently, the accounting disclosure of Portuguese firms derives mainly from the Companies Law and from the Official Accounting Plan.Portuguese companies are required to present a balance sheet, a profit and loss account, a statement of cash flows and a complete set of notes on financial statements.Companies below the following size criteria can present abridged financial statements: when two of the following three limits have not been exceeded for two successive years: total assets of Euros 1,500,000; turnover of Euros 3,000,000 and average number of employees 50.These annual financial statements, together with the management's report and, if appropriate, the auditor's report, must be made public by submitting them to the Business Registry.The layout of the accounts of certain kind of companies differs from those in the POC: banks and other financial institutions whose accounting rules are issued by the Bank of Portugal and also insurance companies whose accounting principles, procedures, and financial reports depend on the Insurance Institute of Portugal.With regard to consolidated accounts, the amendment to the Official Accounting Plan, approved in 1991 following the seventh Directive, made consolidation mandatory for all groups controlled by a parent company.However, some groups are exempt owing to reasons of size or because they belong to larger groups with parent companies in EU member countries.In any case, parent companies of listed groups must always present their consolidated annual financial statements.
The harmonization effort also brought about the adoption of IAS.The influence of such standards can now be observed in almost every DC issued and in the disclosure requirements of the Stock Exchange authority.Thus, in DC 18 a hierarchy of GAAP is presented: Official Accounting Plan; Accounting Standards (DC); and International Accounting Standards.Therefore, the latter must be applied in case of absence of Portuguese accounting rules.And from 2005 on, they should apply to consolidated accounts of listed companies.
In fact, the CNC has been strongly influenced by the IAS when preparing the directrizes contabilísticas.The CNC issued its first accounting guideline in 1991, and in 1999 these guidelines became of compulsory application.Portuguese accounting guidelines are very close in content to comparable IAS issued by the IASB so that among the 29 Portuguese accounting guidelines issued by the CNC, over two thirds of them were developed in line with IAS, for instance, Accounting for business combinations (DC 1); Elimination of profits and losses resulting from transactions, between group undertakings (DC 6), and Deferred taxation (DC 28) among others.
Portuguese accounting rules may differ from IAS/IFRS because there is no accounting standardization in some topics.Concerning these topics, Directriz Contabilística No. 18 includes a formal reference for using IAS where Portuguese accounting rules do not exist.However, this reference has not been extensively applied by companies.Such evidence can be found in literature.Gomes, Serra and Ferreira (2006) approached the harmonization of Portuguese listed companies in terms of intangibles accounting treatment according to IAS 38.In this study little evidence is found in terms of enforcement in the sample of such rule.Authors justify it on non-compulsory character, although DC18 states differently.Even when the IAS has been translated to national normative, companies tend not to apply them.Ferreira, Pinto, Isidro and Alves (2004) compare Portuguese GAAP, IAS and tax law assessing the degree of compliance with the international accounting regulation and the effectiveness of enforcement mechanisms in case of observed non-compliance.The results show that few Portuguese listed companies are disclosing and computing deferred taxes, hence companies are not applying DC 18 and IAS 12, and those enforcement mechanisms are not working properly.They found that there still remained some doubts about the obligation to follow international standards in the absence of national rules possibly because there was not a strong and clear accounting framework or accounting standard with legal power to compel companies.Authors also observed that, although IAS must be applied in the absence of a national standard, auditors are not sanctioning companies for their failure to comply with IAS 12.
Such conclusions are in line with Giner and Rees (2005).According to them, the ASB has to convince those with regulatory and enforcement power to approve and mandate the use of international rules since such a board (IASB) can not do it on its own.In fact, as the authors pinpoint, this will lead rules to be enforced differently or with different effectiveness, losing many of its benefits.Moreover, Flower (2004) calls attention to what he believes will be the future of such a process of harmonization.According to this author, financial reporting in Europe will develop differently whether we consider major European enterprises of an international character or small and medium-sized home country implanted companies.In his vision, the former will move towards a global accounting framework where IASB will have a major role, while for the latter more national diverse accounting practices will prevail.
Finally, concerning the accounting and auditing professions and their influence on accounting in Portugal, we should start by remarking that, as is the case in other European countries, this country has separate professions, and not just separate professional bodies.The CTOC represents all accounting professionals in any matter related to the profession.Membership is required to trade as an accountant in Portugal.The Law created this body in 1995 and slightly revised it in 1999.The CTOC activity is focused on the profession, professional ethics and ongoing education, but it also plays an important role in standard setting through the CNC.At the moment there are nearly 80,000 members, turning this institution into the largest professional association in the country with a large budget, which may anticipate its increasing influence.Traditionally, accountants have had a remarkable relation with tax laws and tax authorities, assuming several responsibilities regarding taxes and professional liabilities.This relationship can justify, at least partially, the use of tax rules over accounting standards.
The Law established the OROC, which represents the certified and statutory auditors, in 1974.This law has been revised several times.All members have to follow the rules approved by the Institute, in accordance with EU eighth Directive.The main functions of the OROC are the issuing of auditing standards, the continuing updating of these standards in compliance with market needs, EU legislation and advances in auditing standards.Auditors have to present a report in which they express their opinion as to whether the financial statements present a true and fair view in accordance with the Portuguese GAAP.
The accounting and auditing professions have not been an important group in anticipating the application in Portugal of IAS.Furthermore, neither the accounting nor the auditing professional bodies have developed a set of accounting principles and procedures that are considered as generally accepted.

Providers of Finance and Capital Markets Regulation
Portuguese firms present a high debt-to-equity ratio, since capital has traditionally been channeled through the banking system.Consequently, firms must periodically send accounting information to the banks (e.g., quarterly and annual financial statements).Nevertheless, the importance of the stock market in financing the activities of companies -although still small -has increased in recent years.After 1994, and until Portugal joined Euronext, there were two stock exchanges (Lisbon and Oporto) under one direction only, on which corporate shares and bonds are traded.Since 1991, the Securities Market Commission (CMVM -Comissão do Mercado de Valores Mobiliários) has assumed the function of regulation, supervision and promotion of the securities market and similar activities carried out by market participants.An Executive Board, appointed by the Council of Ministries by indication of the Ministry of Finance, governs the CMVM.The Executive Board is assisted by the Advisory Board, which represents all market participants, and is overseen by the Supervisory Board.The CMVM is entitled to approve regulatory provisions, the technical rules and the instructions necessary for a regular functioning of the securities market, as well as to propose to government legislative changes, to ensure the continuous updating of the legal structure regarding the demands of the market and the EU requirements.The CMVM can also establish some rules regarding financial reporting and influence the accounting standards through its representation in the CNC.

Influence of Taxation
Traditionally, accounting in Portugal has been highly influenced by tax regulations.In fact, the first significant attempt to harmonize accounting practices occurred in 1963 with a profound tax reform, which contained some implicit accounting regulations.Furthermore, the 1977 Portuguese Official Accounting Plan was strongly influenced by tax-collecting aspects.It was arranged in order to facilitate tax inspections and to justify a company's income tax fairness, (since the basis for the calculation of the tax is accounting earnings), more than to disclosing information on the company's financial position, performance and cash flows.This orientation towards tax continued with the Tax Law of 1988 and with its various revisions.It provides a set of rules for the recording and measurement of items in order to assess taxable income.Taxation affects accounting standards regarding many issues, such as inventory valuation, revaluation of fixed assets, and depreciation of fixed assets.

Conclusion
From what has been said, we can conclude that the Portuguese accounting system must be regarded as code-law based.Accounting standards have been developed mainly to protect creditors' interests, since the main providers of finance even through financial markets are financial institutions.Besides, accounting earnings are regarded as the pie to be distributed among the government (through taxes), debt holders (financial institutions) and a restricted number of shareholders.In all cases, these parties receive the information regarding the company not through financial statements but directly through other more timely sources of information (they behave as insiders).Thus, the economic role of financial statements is strongly attached to tax purposes, guiding in most of the cases managers' accounting choices.Consequently, although it was not included in Nobes (1983), we can include Portugal among the so-called Macro countries (Code-law based in Ball et al., 2000, terms), and within them, among those that he referred to as Plan-based countries, i. e., France, Spain and Belgium.This is why we expect that in our empirical analyses the results regarding Portugal will be similar to those obtained in previous comparative studies in this line of research that included these Planbased countries, such as Joos and Lang (1994); Joos (1997); Ball et al. (2000) or Giner and Rees (2001).

RESEARCH DESIGN Hypotheses Development
We test the following four hypotheses: H1: there exists a balance sheet conservatism bias in Portugal; H2: there exists an earnings conservatism bias in Portugal; H3: the balance sheet conservatism bias of Portuguese firms is more pronounced than that shown by firms in the UK (purest common-law country type in Europe) but less than the one of the German firms (purest code-law type country); H4: Portuguese earnings conservatism bias is less pronounced than that of the UK.
Hypothesis 1 and Hypothesis 2 are single-country hypotheses, while Hypothesis 3 and Hypothesis 4 are multi-country hypotheses.
Hypothesis 1: Balance sheet conservatism bias in Portugal.We expect to find a balance sheet conservatism bias in Portugal, i. e., the market to book ratio will be always greater than one.Among other factors, the strict application of historical cost accounting, the non allowance to the usage of fair value estimates for operating assets and the non recognition or understatement of certain intangible assets (such as research and development costs, brands, software, pensions and other post-retirement benefits) provoke the existence of a difference between market estimates of firm value and the book value of the firm.We define this difference as unrecorded goodwill, and explain its existence in accounting standards themselves.As in other code-law based countries, Portuguese standards have been developed mainly to protect the interests of lenders and other creditors, who demand low figures of book value in order to satisfy themselves that firms will be able to meet their obligations.
Hypothesis 2: Earnings conservatism bias in Portugal.We expect to find earnings conservatism bias in Portuguese firms; namely, there is an asymmetric recognition of good and bad news in accounting earnings.Bad news is recognized on a timelier basis.Basu (1997) defined (earnings) conservatism as capturing the tendency of accountants to require a higher degree of verification for recognizing good news than bad news in earnings.This asymmetry will affect all accounting systems, due to the usage of basic accounting principles.Portugal is no exception and unrealized losses will be recorded immediately, while the recognition of unrealized gains will be recorded only once they have been transformed into cash flows.
Hypothesis 3: Portugal accounting system is more balance sheet conservative than the UK, but less than Germany.We expect to find more pronounced balance sheet conservatism bias in Germany than in the UK, with Portugal in an intermediate position that is closer to Germany.As we have described in Section 2, although it was not included among the countries studied in Nobes (1983), we include Portugal among code-law based countries.These countries are supposed to be more balance-sheet conservative since the main providers of finance are financial institutions, whose main interest in financial statements is in the book value, in order to decide whether or not to invest in the firm, with the assurance that they will have their money back.Contrarily, in common-law countries (the UK is the most clear example in Europe) the main providers of finance are investors in capital markets, and consequently accounting systems have been developed to protect their interests.In these countries, shareholders are more focused on income statement figures make take their investment decisions.One clear example of the differences between the countries under study is the case of the revaluation of tangible fixed assets.While the inclusion of tangible fixed assets in the balance sheet at revalued amount has been a common practice in the UK (extreme example of commonlaw based country), it is completely forbidden in Germany (extreme example of code-law based country), and in Portugal (closer to code-law based country) it can only be put into practice if allowed by the government and using preestablished correction numbers for each year or according to a special technical evaluation (Directriz Contabilística n.º 16).This is why in a similar way to Joos and Lang (1994) and Joos (1997) we expect that common-law countries (in this case the UK) will be less balance sheet conservative.We also expect Germany to be more balance sheet conservative than Portugal due to a stricter application of historical cost accounting.
Hypothesis 4: Accounting earnings are less conservative in Portugal than in the UK.We expect the United Kingdom to be significantly more earnings conservative than Portugal.Due to the differences between the accounting systems commented on Section 2, we expect that the asymmetry in the recognition of good and bad news in earnings will be much more pronounced in the UK than in Portugal, due mainly to the different economic role of financial statements in the two countries and to the larger litigation risk faced by managers and auditors in the UK.On the other hand, we do not expect to find any significant difference in this type of conservatism between Portugal and Germany.

Sample Selection
The lack of availability of data in the traditional databases for Portugal has obliged us to work directly with the annual reports published by the Bolsa de Valores de Lisboa e Porto (BVLP).For the United Kingdom and Germany we use the Extel Company Analysis database.We use the period 1994-1998, given that BVLP reports are only available from that date on, and also because Lisbon and Oporto exchanges were operating separately before that date.We do not use data after 1998 because there IAS may be adopted in case of absence of domestic rules and, as a result, there will be a mix of effects in accounting practices.We have excluded financial firms from the analysis.We also delete from the sample all observations with missing values for any of the variables used.We have only worked with firms with an accounting period length of between 335 and 395 days.We also exclude the two extreme percentiles of each variable.This deletion of outliers is always done by countries, and not for the whole sample.The deletion of outliers have been expanded in the Portuguese case to five per cent of the sample owing to greater variability of the variables, probably due to a much smaller sample.

Methodology Hypothesis 1: Balance Sheet Conservatism in Portugal
Following Feltham andOhlson (1995), Stober (1996), Beaver and Ryan (2000), Zhang (2000) and Givoly and Hayn (2000), we use the market to book ratio as a proxy for conservatism.Givoly and Hayn (2000) argue that using what they call aggregate market to book ratio, which is the aggregate market value of all firms in the sample divided by their aggregate book value at year-end, has the advantage over the simple average ratio across individual companies of being independent of the cross-sectional variance in the ratio.However, if afterwards we wish to compare the ratio across countries, we find it more appropriate to use the simple average ratio, to be thus able to use afterwards a median test, for example the Mann-Whitney test.In our research we adopt two specifications: the aggregate market to book ratio and the single average ratio.However, we do not expect to find differences between them.In both cases, we expect the market to book ratio to be greater than one in every year.

Hypothesis 2: Earnings Conservatism in Portugal
To test Hypothesis 2 we use the simple model proposed by Basu (1997), which was… Where X is earnings per share, P is share price, D is a dummy variable with a value of zero if the rate of return is positive and one if it is negative, and u is the disturbance term.
We expect that the differential effect of bad news with respect to good news, that is, β 3 will be significantly positive, and also that the total effect of bad news, that is, β 2 + β 3 will be also significantly positive.To test this latter extent we use a Wald test.We also expect that the intercepts will be reflecting the effect of prior period good news that is incorporated to this year's earnings.Although the intercept could be reflecting the effect of other omitted variables (for example as Pope & Walker, 1999, point out, the cost of capital) we think that the most important factor affecting the intercept will be prior period news, since when the measurement window is enlarged to test the explanatory power of prior period news on earnings the adjusted coefficient of determination is raised from around 10 to around 60 per cent (see Easton, Harris, & Ohlson, 1992;Kothari & Sloan, 1992).

Hypothesis 3: Larger Balance Sheet Conservatism Bias in Portugal with Respect to the United Kingdom
To test Hypothesis 3 we used two different methodologies, and we do not expect to find different results: These methodologies are a test of the statistical significance of the difference between the market to book ratios across countries (Median test), and usage of a valuation model including book value and earnings.1) Test of the statistical significance of the difference between the market to book ratios across countries (Median test).Since in Hypothesis 1 we have calculated the simple average of the market to book ratio for the three countries under study, we will be able to compare the differential effect of balance sheet conservatism in each country using a Mann-Whitney (median test).To do that, we assume that the ratios follow, in each year, a normal distribution.With this test we will be able to check whether the difference in the market to book ratio across countries is statistically significant.
2) Usage of a valuation model including book value and earnings.Joos and Lang (1994) and Joos (1997) also check the differential effect of balance sheet conservatism across countries using a simplification of the valuation models proposed by Ohlson (1995) and Feltham and Ohlson (1995).One empirical specification of these models, developed by Bernard (1995) for valuation and fundamental analysis purposes, was: Where V is the intrinsic value of the firm; BV is the book value; r is the risk free rate of return and x is the earnings figure.
Adapting this model to an association or value relevance study Joos and Lang (1994) and Joos (1997) use the following simplification, using only contemporaneous values and earnings, instead of a proxy for abnormal earnings: Where P is share price, BV is book value per share, NI is earnings per share and u is the disturbance term.
We expect that due to the larger understatement of book value in Portugal with respect to the United Kingdom, the coefficient γ will be significantly larger for Portugal.To check the statistical significance of this difference we build the following comparative model, which includes also the differential effect of balance sheet conservatism in Germany with respect to the UK: (5) Where P: Share price, DGER: Dummy variable with value 1 if the country is Germany and 0 otherwise, DPOR: Dummy variable that takes value 1 if the country is Portugal and 0 otherwise, NI: Earnings per share, BV: Book value per share, and u: Disturbance term.
We expect that γ 1 and γ 2 will be positive and statistically significant, showing the larger understatement of book value in Germany and Portugal with respect to the United Kingdom.

Hypothesis 4: Larger Earnings Conservatism Bias in the UK relative to Portugal
To test Hypothesis 4 we use the comparative model developed by Ball et al. (2000) from the simple model by Basu (1997).The model is as follows: (6) RAC, 2 a. Edição Especial 2007 Where: EPS: Earnings per share, R: Rate of return of the firm, CD j : Dummy variable of country j.It takes value 0 in the case of the United Kingdom and 1 if it is country j.We use the United Kingdom as the value reference country as it is the most different, a priori, of the three countries under study.RD: Dummy variable that takes value 0 if there is good news, namely, when the rate of return is positive, and 0 in the case of bad news (negative rate of return), or rate of return equal to zero, u: Disturbance term.
Contrarily to Ball et al. (2000), and following Pope and Walker (1999), we use earnings after extraordinary items per share.We use this definition of earnings, and not earnings before extraordinary items, as is usual in the vast majority of studies, to try to avoid our results being influenced by a different classification of good and bad news within financial statements across countries.The inclusion of extraordinary items will not bias the coefficients of the model since we use a reverse regression, and so the measurement error in earnings is placed in the disturbance term (See Beaver et al., 1987).We expect β 3j , which shows the differential effect of bad news in Portugal and Germany with respect to the United Kingdom, to be significantly positive, thereby showing a less pronounced conservatism bias in Portugal and Germany.

RESULTS
The descriptive statistics are on Tables 1 and 2.

Table 1
Where: Sh.Price: Market share price (in Euros); BV: Book value per share (in Euros); NI: Earnings after extraordinary items per share (in Euros).

Table 2
With respect to Hypothesis 1, concerning the existence of balance sheet conservatism in Portugal, the market to book ratio is always greater than one in the five years under study.This result is not affected by the way in which the market to book ratio has been calculated, either the aggregate ratio or the simple mean ratio.These results are on Table 3.

Table 3
Regarding Hypothesis 2, on the existence of earnings conservatism in Portugal, our results show that there exists an asymmetry in good and bad news recognition in earnings.Bad news is recognised on a timelier basis (β 3 is significantly positive), and thus, there exists earnings conservatism (See Table 4).We also check that the total bad news effect (β 2 + β 3 ) is significantly positive, and that the intercept is significantly positive; both results being consistent with the existence of earnings conservatism.

Table 4
Where: X t : Earnings after extraordinary items per share deflated by share price at the beginning of the period; R t : Observed rate of return of the firm; D : Dummy variable.Takes value 0 if the rate of return is positive; t-statistics are White (1980)  Concerning Hypothesis 3, namely, the larger balance sheet conservatism bias of Portugal with respect to the United Kingdom, our results are very surprising.When we compare the market to book ratio through a median test (Mann-Whitney test) Portugal always shows a statistically significant smaller balance sheet conservatism bias than both the UK and Germany.On the other hand, and much as expected, Germany shows more pronounced balance sheet conservatism bias than the UK, but it is only statistically significant in 3 out of the 5 years under study (See Table 5).

Table 5
Z-statistics are based on Mann-Whitney test rank-sum comparison of median ratios.We exclude negative book value figures.Market to book value is calculated as the simple average of individual market to book value of all firms in the sample, for each country and each year.
being n 1 and n 2 the size of the two samples in comparison and R 1 the rank sum of the first sample.We replicate this analysis by using the whole sample (inclusive of firms with negative book value) and excluding from the first analysis also those firms with market to book ratio greater than 5, obtaining similar results.
When we use the valuation model in Joos and Lang (1994) and Joos (1997) the results (see Table 6) are not consistent with our hypothesis.The differential book value multiple in Portugal with respect to the United Kingdom, γ 2 , is significantly negative.This shows, contrary to our expectations, a larger balance sheet conservatism bias in the UK relative to Portugal.On the other hand, Germany, much as expected, shows a larger balance sheet conservatism bias than the UK, and thus, larger also than Portugal.This surprising result regarding Portugal could be explained by the fact that firms in the UK tend to have a more important amount of intangible assets, which makes the market to book ratio in the UK larger on average than in Portugal, but not due to conservatism, but rather due to the fact that investment in intangible assets is much lower in Portugal.Another economic reason for this result is that it is highly likely that the Portuguese stock market is not efficient enough to capture the value relevance of unrecorded goodwill (for example the value relevance on unrecorded intangible assets), with which market capitalization would not be a good proxy for the intrinsic value of the firm.Another fact that can contribute to this surprising result is the weaknesses of the valuation model used to compare balance sheet conservatism across countries.As several studies point out, this type of model can suffer from important problems of scale that may be causing bias to the results in some cases (See Barth & Clinch, 2000;Barth & Kallapur, 1996;Easton, 1998;Easton & Sommers, 2000).
Finally, concerning Hypothesis 4, on the larger earnings conservatism bias of the United Kingdom with respect to Germany and Portugal, our results (see Table 7) show that the differential bad news coefficient (β 3j ) of Germany and Portugal with respect to the United Kingdom is, as expected, negative, showing larger earnings conservatism bias in the UK with respect to Germany and Portugal, but this larger earnings conservatism bias of the United Kingdom is not statistically significant.

Table 7
Where: X t : Earnings after extraordinary items per share deflated by share price at the beginning of the period; CD j : Dummy variable.Takes value 1 if it is country j.Value reference country UK; R t : Observed rate of return of firm I; RD : Dummy variable.Takes value 0 if the rate of return is positive.
This result, although not totally consistent with Hypothesis 4, is similar that one obtained in other studies, such as Ball et al. (2000), where they do not even check the statistical significance of the differences between code-law based countries and the UK.Nevertheless, these results could be seriously biased by the fact that the explanatory variable in the reverse regression approach can suffer from a measurement error (Beaver, McAnally, & Stinson, 1997;Francis & Schipper, 1999), and also due to problems of cross-sectional dependence (Bernard, 1987).Another explanation for this result could be that, as LaPorta, Lopez-de-Silanes and Shleifer (1999) and LaPorta, Lopez-de-Silanes, Shleifer and Vishny (2000) point out, the difference in the composition of corporate ownership between codelaw based and common-law based countries no longer holds.However, from our point of view, even though the differences in that particular subject between code-law and common-law based countries could be disappearing, we do not think that the role of financial statements is the same nowadays in code-law and common-law based countries.Therefore, we firmly believe that the difference between a creditor oriented and a shareholder oriented accounting system still holds.

CONCLUSIONS AND IMPLICATIONS
The paper provides evidence on the existence of both balance sheet and earnings conservatism in accounting practices in Portugal: the market to book ratio is consistently greater than one and bad news is recognized in earnings on a timelier basis than good news.
RAC, 2 a. Edição Especial 2007 We also test the differential level of both types of conservatism in Portugal with respect to the two extreme examples of common-law and code-law based countries in Europe, namely the United Kingdom and Germany.The results of these analyses are not completely consistent with our hypotheses.We find that Portugal is less balance sheet conservative than the United Kingdom.This surprising result, contrary to our hypothesis, could be explained by two economic facts.Firstly, that British companies probably have on average more unrecorded intangible assets, and thus, on average, larger market to book ratios.And secondly, it is very likely that the Portuguese stock market is not efficient enough to recognize the value relevance of unrecorded goodwill, i. e., in the Portuguese case, market capitalization would not be a good proxy for the intrinsic value of the firm.
With respect to earnings conservatism, our results, although consistent with our hypothesis of a larger earnings conservatism bias in the UK, are not statistically significant.This is consistent with the results in other studies.Certain papers also point out that the results of earnings conservatism that use reverse regressions can be biased due to the existence of a measurement error in the explanatory variable, i. e., in the observed rate of return, and also to the existence of crosssectional dependence problems.This research assumes the non-existence of between country factors, such as macroeconomic differences between the countries, but what may be influencing our results are the so-called within country factors.This means our results could be influenced by a different sample composition in the countries under study.
The results have implications for accounting standard setting and which may be useful to both the European Commission and the IASB since they provide some insight into the properties of accounting figures in Portugal.Later on, and once that Portuguese firms begin using the new IASB rules, we will be able to analyze whether these properties have changed due to the implementation of IFRS (International Financial Reporting Standards) and whether this has contributed to a larger comparability of Portuguese accounting data with respect to other European countries.Additionally, further tests can be conducted to test whether the enforcement of such international rules has produced accounting harmonization concerning the balance sheet and earnings conservatism level across Europe, mainly in Portugal.Such further investigation can be contrasted with this study to analyze whether levels of conservatism stay the same as well as with other countries studies to assess if one can speak of a truly global accounting framework.

Table 6
Where: P t : Share price; NI: Earnings after extraordinary items per share; BV: Book value per share; DGER: Dummy variable that takes value one if it is a German firm; DPOR: Dummy variable that takes value one if it is a Portuguese firm; t-statistics are White (1980) heteroskedasticity consistent.