1. Introduction

In late 2004 the EU adopted Directive 2004/109/EC (‘Transparency Directive’) (TD) and with it introduced, in Article 6, a new quarterly statement known as the Interim Management Statement (IMS). For a typical EU country like the UK, this introduction increased the annual reporting frequency from two to four statements. Specifically, listed companies on EU-regulated markets have to issue, for any period starting after 20 January 2007, an IMS in the first quarter and an IMS in the third quarter, in addition to an interim report and an annual report. The Commission of the European Communities (2003, p. 3) argued that the introduction of IMSs was necessary to increase investor protection and to close the transparency gap between the USA and the EU especially as the Securities and Exchange Commission (1970) has required quarterly reports from US-listed companies since 1971. However, unlike the quarterly report in the USA, the IMS does not need to include a set of financial statements. Instead, an IMS meets the TD's requirements by simply giving a general description of the financial position and performance since the last interim or annual report and by explaining any material events and transactions that have since taken place. These descriptions and explanations can be entirely narrative if the firm so wishes and they are typically no longer than two pages in length. Thus, in effect, IMSs are relatively short trading updates, not financial reports per se.

More recently, the Commission of the European Communities (2011) announced its intention to make quarterly reporting voluntary again. Specifically, the Commission of the European Communities (2011, p. 5) argued in 2011 that abolishing IMSs should not have any negative impact on investor protection as Directive 2003/6/EC (‘Market Abuse Directive’) (MAD) already requires, in Articles 1 and 6, the immediate, ad hoc release of price-sensitive information. This, the Commission argues, implies that IMSs are redundant, as they are unlikely to contain any incremental information, that is, any new information over and above the information already required by the MAD. Mandatory IMSs were formally abolished on 22 October 2013 via Directive 2013/50/EU and member states will have to implement the withdrawal of the IMS requirement within 24 months.

The current paper empirically tests the Commission's argument that quarterly IMSs are unlikely to contain any incremental information. For that we hand-collect information on trading statements, including IMSs, in 2009 and 2010. These are the two years for which, at the time of data collection, IMS publication dates were available from PI Navigator, a global corporate finance and capital market database, and which, unlike 2007 and 2008, are less likely to suffer from teething problems often associated with the implementation of a new reporting regime. We then compare the actual number of trading statements, including IMSs, in 2009 and 2010 against the number of trading statements predicted for a hypothetical disclosure regime that operates under the MAD but without mandatory IMSs. Comparing actual against predicted values allows us to divide IMSs into two types of statements, that is, IMSs that are likely to be genuinely incremental trading statements (‘incremental IMS’) and IMSs that are likely to be substitutes for otherwise disclosed non-periodic trading statements (‘non-incremental IMS’).

We obtain estimates of the expected number of trading statements under the hypothetical disclosure regime in two different ways, but with similar results. First, we regress the number of trading statements in 2006, a post-MAD pre-IMS year, on a large number of firm characteristics which have been suggested as determinants in the prior literature on reporting frequency and voluntary trading statements (e.g. Kasznik and Lev 1995, Miller and Piotroski 2000, Cuijpers and Peek 2010) and we use the estimated regression coefficients to predict values for 2009 and 2010. Second, for a subsample of firms which are listed on the London Stock Exchange in both 2006 and 2009 or 2010, we use the actual number of trading statements in 2006 to predict values under the MAD-only regime for 2009 and 2010. We selected 2006 as the benchmark year as the MAD came into effect in the UK on 1 July 2005. Thus, 2006 is a year which operated under the MAD, but not under the TD.

Next we test the argument that mandatory IMSs are redundant by following Beaver (1968) and Landsman et al. (2012) and by calculating three-day announcement period abnormal return variability and abnormal trading volume associated with incremental and non-incremental IMSs. The calculation of these two metrics follows the notions in Beaver (1968) and Lev (1989) who deem information to be relevant to investors if (a) fluctuations in stock prices and (b) trading volumes can be attributed to that information and who argue that the greater the relevance of the information, the greater the associated price fluctuation and trading volume. The rationale is that the degree of price fluctuation and trading volume reflects the extent to which an investor reassesses the amount, timing and uncertainty of expected future cash flows, and the extent of this reassessment is argued to increase in the relevance of the information. For example, if mandatory IMSs are redundant, as argued by the EU, then incremental IMSs should exhibit no abnormal return variability and abnormal trading volume.

Our main findings can be summarised as follows. First, when we analyse the frequency of non-periodic trading statements pre- and post-IMS, we find that the introduction of IMSs coincides with a substantial reduction in the average per-firm number of non-periodic trading statements, especially in the first half of the financial year. This is consistent with a large substitution effect between periodic and non-periodic trading statements. In contrast, the substitution effect is smaller in the second half of the financial year, and, as a result, the majority of incremental IMSs are third-quarter IMSs, not first-quarter IMSs. Second, incremental third-quarter IMSs are associated with significant abnormal return variability and significant abnormal trading volume, consistent with the argument that abolishing mandatory IMSs will lead to some loss in relevant information. Third, when we compare abnormal return variability and abnormal trading volume between incremental and non-incremental IMSs, we observe consistently higher medians for non-incremental IMSs, and the observed differences are frequently significant, especially when we examine IMSs in the second half of the year, that is, third-quarter IMSs. If we interpret abnormal return variability and abnormal trading volume associated with non-incremental IMSs as proxying for the amount of relevant information that is disclosed even in the absence of a mandatory IMS regime, then the findings of higher medians for these statements suggest that the MAD, in its own right, is effective in ensuring the release of most relevant information, consistent with the Commission's arguments in 2011 and the formal withdrawal of mandatory IMSs in 2013.

In summary, we contribute to the literature in two ways. First, we provide evidence on the degree to which periodic and non-periodic disclosures interact with each other. In particular, our finding of a large substitution effect supports the idea in Gigler and Hemmer (1998) that a regulator's decision to increase reporting frequency can have a negative effect on a firm's other disclosures. Second, we provide evidence on the extent to which a disclosure regime based on mandatory ad hoc disclosures, rather than mandatory periodic disclosures, is sufficiently reliable to ensure the release of most relevant information. This evidence should be of interest to policy-makers, both inside and outside the EU, who wish, in the future, to weigh up the cost and benefits of mandatory periodic disclosures versus mandatory non-periodic disclosures.

The remainder of the paper is organised as follows. The next section reviews the literature on reporting frequency. Section 3 reviews the history of the TD. Section 4 describes the sample selection and provides evidence on the type of disclosures that are routinely made in UK IMSs. Section 5 presents findings from pre- and post-IMS trading statement frequencies, pre-IMS trading statement choice models, and IMS abnormal returns variabilities and abnormal trading volumes. Section 6 concludes.

2. Literature on reporting frequency

Most prior work on reporting frequency is empirical in nature and studies the incentives and economic consequences of increased reporting frequency, often in a US context. For example, Leftwich et al. (1981) document that, over the period 1937–1948, US firms’ choices on interim reporting frequency vary with the stock exchange listing, while Sivakumar and Waymire (1994) find no evidence that, over the period 1905–1910, industrial firms listed on the NYSE were more likely to report favourable information in voluntary interim reports. In terms of economic consequences, McNichols and Manegold (1983) show that the abnormal return variability around the annual earnings release date is reduced for the 34 AMEX firms that change their reporting frequency from annual to semi-annual or quarterly reporting in the 1960s. In a European context, several studies find that voluntary and mandatory interim reports spur significant price and trading volume reactions (e.g. Schadewitz 1996, Alves and Texeira Dos Sontos 2008), suggesting that interim reports in several European countries provide useful information to equity investors. However, IMSs are neither earnings announcements nor quarterly reports. Thus, it is unclear whether prior empirical findings on reporting frequency apply to the IMS regime.

Several recent studies investigate empirically how the introduction of voluntary or mandatory interim reporting affects stock price informativeness (Alford et al. 1993, Butler et al. 2007, Cuijpers and Peek 2010) or information asymmetry and the cost of equity (Fu et al. 2012). These studies account explicitly for the possible interaction effects between reporting frequency and other sources of public or private information. Specifically, it is possible that the beneficial effect of increased reporting frequency on price informativeness is moderated by the effect that the reporting frequency choice has on (a) investors’ private information acquisition activities and (b) the firm's willingness to make voluntary disclosures.

Higher reporting frequency can lead to more informative prices if it increases the total amount of information available to investors. However, as Verrecchia (2001) points out, this assumes that investors’ private information acquisition activities are exogenously given. Relaxing this assumption could mean that a higher reporting frequency coincides with a reduction in private information acquisition, which could partly or fully negate the effect of higher reporting frequency on the amount of available information to investors.¹

Whether higher reporting frequency increases the total amount of publicly available information is also unclear since higher reporting frequency can reduce the incentive for firms to make voluntary disclosures. In particular, Gigler and Hemmer (1998) show that an increase in reporting frequency may induce firms to delay their voluntary disclosure until the next mandatory reporting date. Specifically, the increase shortens the average period between a price-relevant information event and the next mandatory reporting date, thereby reducing the potential cost of delaying the disclosure. Examples of costs that might be reduced are reputational and litigation costs (Cuijpers and Peek 2010). For example, the reduced time span between mandatory reporting dates makes it more difficult for outsiders to argue that managers possessed undisclosed private information over a lengthy period of time.

While the theoretical predictions in Gigler and Hemmer (1998) are frequently used to motivate empirical work on reporting frequency (e.g. Fu et al. 2012), there is little direct evidence on whether the interaction effects suggested in Gigler and Hemmer (1998) are empirically important. The only direct evidence comes from Cuijpers and Peek (2010) who, in a supplementary test, show that UK quarterly reporters make voluntary news announcements on 3.97 days per year, compared to 4.22 days by UK semi-annual reporters, and the difference is marginally significant in a multivariate regression.

Below we argue that IMSs are trading statements. It is then natural to focus the analysis of the interaction effect on trading statements, not on all news announcements more generally.² Formally, we make the following prediction:³

3. EU Transparency Directive

After two consultation rounds in 2001 and 2002, the Commission of the European Communities (2003) presented on 26 March 2003 its proposal for a new Directive on the harmonisation of transparency requirements for issuers of securities on regulated markets. The idea was to ‘markedly improve the information made available to all investors about publicly traded companies’ as this was seen as ‘essential for the functioning of capital markets, enhancing their overall efficiency and liquidity’ and as ‘an appropriate response to developments in the US, including the Sarbanes-Oxley Act’ (Commission of European Communities 2003, pp. 2–3). Greater transparency was envisaged, as far as periodic financial reporting was concerned, through (a) mandating the timely publication of annual reports after the year-end, (b) the formulation of more stringent disclosure requirements for interim (i.e. half-yearly) financial reports, and most importantly (c) a new requirement, for issuers of shares, of ‘less-demanding’ quarterly financial information for the first and third quarters of the financial year. Demanding only ‘limited’ information for the first and third quarters was seen as a solution in the middle of two extreme positions. One extreme position was to demand full quarterly financial reports similar to the requirement in the USA. The other extreme was to continue as before and not to require any quarterly information at all (Commission of European Communities 2003, p. 3).

The TD was formally adopted on 17 December 2004. Some of the initial proposals had been watered down, presumably as a result of political negotiations between the Commission and Parliament/Council. For example, the final wording of the Directive no longer prescribes the disclosure of net turnover and profit or loss for quarterly statements. Instead, Article 6 of the TD simply requires, for the period between the beginning of the quarter and the publication date, (a) an explanation of material events and transactions and their impact on the financial position and (b) a general description of the financial position and performance.

The wording in Article 6 suggests that IMSs are very different from interim and annual reports and also very different from US-style quarterly reports. In effect, IMSs are lightly regulated trading statements with management retaining considerable control over form and content. For example, the issuer can choose which financial statement line item, if any, to comment on when discussing financial position and performance, and the management is free to present this information either in numerical or in narrative form. Also, there is no duty to indicate trends beyond the date the statement is published. Finally, management retains some control over the length of the reporting period as Article 6 only stipulates that it ‘shall be made in a period between ten weeks after the beginning and six weeks before the end of the relevant six-month period’.

The UK implemented the IMS provisions of the TD without any modifications, also known as the ‘copy-out’ approach. The new rules were implemented via the UK Listing Authority's Disclosure and Transparency Rules (DTR) (which are part of the Financial Conduct Authority Full Handbook) and became effective for periods beginning on or after 20 January 2007. DTRs also implemented the new shorter deadlines for the publication of interim and annual reports, of two and four months, respectively, but, overall, the Directive had only limited implications for UK reports as the typical UK issuer continued to employ IFRS for (condensed) interim and (full) annual reports.

In early 2010 the Commission set out to report on the operation of the TD, in accordance with Articles 6(3), 27(3) and 33. In particular, Article 6(3) of the TD required the Commission to ‘provide a report [ … ] on the transparency of quarterly financial reporting’ and to ‘include an impact assessment on areas where the Commission considers proposing amendments’. Based on the advice received from a number of external studies, regarding stakeholder perceptions and compliance costs, the Commission of the European Communities (2010) initiated a debate, on 27 May 2010, by asking whether there was scope for the transparency rules to be adapted, with a view to increasing the attractiveness of the regulated market for smaller companies.

Following a formal consultation round in summer 2010, the Commission of the European Communities (2011) published its formal proposal for an amendment of the TD on 25 October 2011. In it the Commission reiterated that an ‘[i]mprovement of the regulatory environment for small and medium-sized issuers and their access to capital are high political priorities for the Commission’ (Commission of the European Communities 2011, p. 3). In order to achieve this priority, the Commission proposed to abolish the obligation to publish IMSs for all companies listed on regulated markets as abolishing IMSs only for small and medium-sized issuers was argued to be confusing to investors and thus undesirable. The Commission of the European Communities (2011, p. 5) argued that abolishing IMSs ‘enables the small and medium-sized issuers to redirect their resources’ and that abolishing IMSs ‘should not have negative impact on investor protection’.

The Commission's proposal to abolish mandatory quarterly information and to reverse an important change introduced by the TD is based, in part, on a reassessment of the amount of periodic information that is deemed necessary to guarantee investor protection. While the Commission argued in 2003 that periodic reporting, including quarterly information, and ad hoc disclosures are different ways of informing the public, it emphasised in 2011 that quarterly information is not needed as the MAD already requires important information to be disclosed without delay.

The above discussion suggests that any formal test of the Commission's 2011 position needs to differentiate between IMSs that are likely to be substitutes for disclosures made under the MAD and IMSs that are genuinely incremental disclosures. We refer to the latter as incremental IMSs. Consistent with the Commission's argument that mandatory IMSs are not needed for investor protection, we formulate the following null hypothesis:

We operationalise the concept of relevant information for investors in terms of both abnormal return variability and abnormal trading volume. Examining both types of stock-market reaction side by side is consistent with the information content literature – including Beaver (1968), Landsman and Maydew (2002), DeFond et al. (2007) and Landsman et al. (2012) – which argues that tests of abnormal return variability and abnormal trading volume complement each other. Specifically, we follow Beaver (1968) in arguing that a change in price reflects an average investor's reassessment of future cash flows, and in this way it measures one aspect of information relevance, namely the extent to which investors change their beliefs in aggregate. Note that return variability measures price change by abstracting from the sign of the price change as positive and negative prices changes of the same size are interpreted as equally relevant and as positive and negative price changes can no longer cancel each other out (as is the case, for example, with abnormal returns). We follow Beaver (1968), Kim and Verrecchia (1991, 1997) and Landsman et al. (2012) and interpret trading volume as capturing another aspect of information relevance, namely as the extent to which an announcement generates diverse interpretations across investors. Specifically, it is argued that the greater the relevance of an announcement the more likely investors with different levels of private information and different abilities to interpret financial information will differ in their interpretations, and the more they will want to trade as a result of their dissimilar interpretations.⁴

For our third and final test, we calculate abnormal return variability and abnormal trading volume associated with non-incremental IMSs, that is, IMSs that our models predict to have replaced previously disclosed trading statements, and we use their abnormal return variability and abnormal trading volume as proxies for the amount of relevant information that is disclosed under the MAD but in the absence of a mandatory IMS regime. Specifically, if the MAD, in its own right, is effective in ensuring the release of most relevant information, as argued by the Commission in 2011, then one would expect that the typical market reaction associated with a non-incremental IMS is higher than the typical market reaction associated with an incremental IMS. The reason is that non-incremental IMSs are frequently released with the intention to communicate relevant insider information as required by the MAD. In contrast, a relatively larger number of incremental IMSs would be expected to be full of uninformative information as the firm is forced to issue a trading statement despite having little relevant information to reveal. Formally, we make the following prediction:

We test H3 by comparing abnormal return variability and abnormal trading volume between incremental and non-incremental IMSs. Finding evidence in favour of abnormal return variability and abnormal trading volume being higher for non-incremental IMSs would be consistent with the argument that most relevant information is released under the MAD and that mandatory IMSs are redundant.

4. Sample selection and content analysis

We begin our empirical analysis by collecting, for the years 2009 and 2010, trading statement release dates, including IMS release dates, for all non-financial firms included in the FTSE All Share Index in June 2009 and June 2010, respectively. We collect these dates from PI Navigator, a global corporate finance and capital market database with 15 million international company filings, including filings submitted via the London Stock Exchange's Regulatory News Service (RNS). We do not include in our sample period the years 2007 and 2008. The reason is that IMSs became effective only for periods beginning on or after 20 January 2007. Thus, many firms did not publish their first IMS before 2008. Given the potential for teething problems, including lower compliance rates, in the year of first-time implementation, we decided not to include 2008, either.⁵ This leaves 2009 and 2010 for inclusion in our sample period as the years that were available from PI Navigator at the time of data collection. Note that 2009 represents a year from the financial crisis period while 2010 was characterised by a rebound and improving sales and profit performance. Thus, the two years represent, in terms of performance and uncertainty, different economic environments.

In order to predict whether IMSs are incremental or non-incremental, we need information on trading statements from a benchmark period. Our selection of a benchmark year was guided by the necessity to find a year which operated under the MAD, but not under the IMS regime. With the MAD becoming effective on 1 July 2005 and mandatory IMSs applying to any financial year starting after 20 January 2007, the only year that meets this criterion is 2006. Thus, 2006 is employed as the post-MAD, pre-IMS benchmark year for both 2009 and 2010.⁶

In all three years, the sample selection starts with the FTSE All Share Index list. The number of firms in the index is 685, 615 and 614 in 2006, 2009 and 2010, respectively. These are the number of constituents in June 2006, June 2009 and June 2010. Subsequently, we delete observations for four reasons. First, we remove all financial firms from our sample. Second, we exclude a small number of firms without matching codes on Datastream. Third, we delete firms where the preliminary earnings announcement date is missing on PI Navigator. Preliminary earnings announcement dates are typically missing when the firm cancels its listing throughout the financial year. Fourth, we delete firms with annual accounting periods of less than 359 or more than 371 calendar days. This results in initial samples, for a test of H1, of 388, 341 and 326 observations in 2006, 2009 and 2010, respectively. Note that in all three years the majority of deletions – more than 200 per year – relate to financial firms. Removing financial firms is standard practice whenever accounting data are involved in estimation.⁷

Before we test our hypotheses, we provide some background information on the type of disclosures that are routinely made in UK IMSs. For that we randomly select 20 non-financial firms each from the three indices that make up the FTSE All Share Index, namely the FTSE 100, the FTSE 250 and the FTSE Small Capitalisation Index.⁸ For each group of firms – and for the total of all firms – we report in Table 1 the percentages of IMSs that include information on (a) financial performance, (b) financial position, and (c) material events and transactions. This information is collected through manual, meaning-orientated, content analysis (e.g. Krippendorff 1980, Weber 1990) and is broadly similar to the analysis in Deloitte and Touche (2007, 2008) (which was for the year of first-time IMS implementation). However, our focus is on what type of information is typically conveyed in an IMS, not on whether IMSs comply with the DTR rulebook.

Table 1 allows us to make a number of initial observations. First, 90% of all IMSs meet the TD's requirement to give a general description of the financial performance by making statements about the recent sales performance since the last interim or annual report. Also, references to recent sales performance are higher for large and mid-cap firms than for small-cap firms (93% and 96% versus 80%). Second, 45% of all IMSs provide an indication of the recent earnings performance. Third, in most cases backward-looking sales and earnings information relates to the group, not an individual segment. This is in line with the Directive's requirement. At the same time, statements about recent sales and earnings performance are often voluntarily supplemented with quantitative data. For example, 76% = 68% ÷ 90% of backward-looking sales statements are point estimates. Fourth, when IMSs voluntarily provide an outlook for a period beyond the IMS publication date, then references to ‘growth’, ‘progress’, ‘success’ or ‘outlook’ dominate references to ‘sales’ and ‘earnings’ (83% versus 25% and 32%). This disclosure behaviour is understandable as referring to vague, non-verifiable performance indicators minimises the risk that the outlook statement is proved to be ex post inaccurate. Fifth, around half of all IMSs refer to material events and transactions. Finally, the median IMS statement is not particularly long as a median of 757 words and 21 sentences corresponds roughly to a two-page trading statement.

In Table 2 we reproduce Carpetright plc's 2010 third-quarter IMS. Consistent with the analysis in Table 1, the discussion in Carpetright plc's IMS focuses, to a large extent, on sales and earnings, and sales and earnings news is often benchmarked against prior periods or analysts’ forecasts.

Overall, the impression we gain from Tables 1 and 2 is twofold. First, a description of the recent and future trading performance dominates the discussion in UK IMSs. Second, it is likely that UK IMSs are relevant to investors as sales and earnings news is precisely the type of information that assists in the calculation of intrinsic value changes (e.g. Palepu et al. 2010).

5. Hypotheses testing

We start our formal tests with H1. For that we need an operational definition of trading statements. In particular, we need a list of news headings typically associated with trading statements. To obtain this list, we downloaded, for around 100 pilot study firms, from PI Navigator, all RNS news announcements between two successive preliminary earnings announcement dates, and we manually assessed the contents of these announcements. We then judged trading statements to be confined to the following headings: ‘AGM Statement’, ‘Chairman Statement’, ‘Trading Update’, ‘Trading Statement’, ‘Pre-Close Update’, ‘Pre-Close Trading Statement’, ‘Pre-Close Trading Update’, and ‘Interim Management Statement’.⁹ While our manual assessment of relevant headings is independent of the classification supplied by PI Navigator, we note that choosing the PI category ‘Trading and Operating Updates’ returns the same RNS announcements as our manual list, with two exceptions, namely ‘AGM Statement’ and ‘Chairman Statement’, both of which are included in the PI category ‘Annual Results’. In particular, PI Navigator lists under the category ‘Trading and Operating Updates’ first-quarter and third-quarter IMSs, an indication that our classification of IMSs as trading statements is accepted by others.

Subsequently, we collect, from PI Navigator, all relevant trading statement release dates for all non-financial firms included in the FTSE All Share Index in 2006, 2009 and 2010, and we compare, in Table 3, the frequency of trading statements between the pre-IMS year, 2006, and the two post-IMS years, 2009 and 2010. This comparison is done for the first and the second halves of the financial year separately. Thus, we interpret IMSs as doubling the reporting frequency in each half of the financial year, and we examine the possible substitution effects separately for each six-month period. This is a natural focus given the notion in Gigler and Hemmer (1998) that information in trading statements, including IMSs, is fully subsumed in financial reports.

In order to facilitate a deeper understanding of the nature of the substitution effect, we record the frequency of trading statements released during the close period or in the 30 calendar days immediately prior to the close period, separately from trading statements made earlier in each reporting period.¹⁰ We also separately record trading statements made at the Annual General Meeting (AGM). Finally, we include in Table 3 a frequency count for quarterly reports. We do this for two reasons. First, quarterly reports are typically voluntary, even for firms with a dual listing overseas.¹¹ Second, firms which publish quarterly reports are exempted by the TD 2004/109/EC, Article 6(2), from publishing a separate IMS.

Table 3, Panel A, reports, for the full sample, average per-firm trading statement frequencies and compares the frequencies in 2009 and 2010 against the frequencies in 2006, the benchmark year. It also reports p-values from a two-sample t-test and a two-sample Wilcoxon rank-sum test, to test for differences in the mean and the distribution, respectively, but we note that parametric and non-parametric p-values yield quite similar results in terms of significance levels. Specifically, we find that the publication of IMSs in 2009 and 2010 coincides with a large and highly significant reduction in AGM trading statements and this observation is consistent with our prediction in H1. For example, while 68% of sample firms issued a separate trading statement at their AGM in 2006, only 13% and 11% continue to do so in 2009 and 2010. What is the story behind this large substitution effect? Obviously, most firms feel that there is no longer any benefit in a separate AGM trading statement, especially given that AGMs in the UK are typically scheduled to coincide with the 10-week window during which IMSs must be published. Thus, most firms decided to publish the first-quarter IMS on the same day as the AGM trading statement. Clearly, publishing a joint AGM/IMS trading statement is facilitated by the flexibility that firms are given in the TD in choosing their IMS reporting date.

As far as the first half of the financial year is concerned, we also observe that the number of (pre-) close period trading statements is significantly lower post-IMS and this reduction, together with the large reduction in separate AGM trading statements, leads to a large substitution effect which effectively negates the entire increase in trading statements that comes from the introduction of first-quarter IMSs. Specifically, the change in the average number of trading statements, including IMSs, in the first half of the financial year is 0.00 = 1.45 − 1.45 in 2009 and 0.05 = 1.50 − 1.45 in 2010, with p-values of .49 and .29 in 2009 and .17 and .40 in 2010.¹²

The picture is different in the second half of the financial year. Specifically, the absence of AGM trading statements in the second half of the year leaves only (pre-) close period statements and trading statements to be withdrawn. While the withdrawal of trading statements issued earlier in the second half is, consistent with H1, significant and large, at least in terms of percentage change, it appears that firms are somewhat more reluctant to withdraw second half (pre-) close period trading statements. One can only speculate why. Perhaps earnings guidance at the very end of the annual reporting cycle is indispensable as analysts and investors demand clarification about the current year's earnings and as it is only at that time that firms feel they can give accurate guidance. Whatever the precise reason the increase in average per-firm trading statements in the second half of the year is more substantial at 0.47 = 1.47 − 1.00 and 0.45 = 1.45 − 1.00.¹³

In Table 3, Panel B, we repeat the comparison of trading statement frequencies, but this time the sample composition is constant between the pre-IMS year and the post-IMS year, and we observe that the constant sample findings in Panel B are quite similar to the full sample findings in Panel A. This provides some assurance that the full sample results in Panel A are not likely to be driven by variations in the sample composition over time.

Finally, in Table 3, Panels C and D, we split the full sample into large and mid-cap firms, that is, firms that belong to the FTSE 100 or FTSE 250, and small-cap firms, and we observe that the substitution effect is generally greater for large and mid-cap firms than for small-cap firms. For example, we observe a larger reduction in the number of (pre-) close period trading statements for large and mid-cap firms than for small-cap firms, and this observation applies to both the first and the second halves of the year. This suggests that small-cap firms were the real losers of the TD in the sense that small-cap firms, more than large and mid-cap firms, were forced to issue additional trading statements that they had not wanted to issue otherwise. But this also means that it is the small-cap firms which are likely to benefit most from a withdrawal of the IMS regime. However, whether the associated cost saving is large is unclear, especially in the light of the evidence on the typical length of an IMS in Tables 1 and 2.

Before we test H2 and H3, we compare, in Table 4, abnormal return variability and abnormal trading volume associated with first- and third-quarter IMSs against abnormal return variability and abnormal trading volume associated with interim results and preliminary earnings announcements. For that we follow prior research, including Beaver (1968), Landsman and Maydew (2002), DeFond et al. (2007) and Landsman et al. (2012), and measure abnormal return variability as the ratio of the event window return variability to the non-event window return variability. Specifically, we start by calculating daily market-model adjusted returns, as:(1)

Are interim management statements redundant?

All authors

Thomas Schleicher & Martin Walker

https://doi.org/10.1080/00014788.2014.1002444

Published online:

03 March 2015

Figure 1. Timeline of mandatory periodic announcements and associated event and non-event periods. The figure illustrates the timeline of mandatory announcements in the UK under the combined MAD/IMS regime and illustrates the possibility of overlaps between event and non-event periods. The mandatory announcements include the preliminary earnings announcement (PE), the first-quarter IMS (IMS1), the interim result announcement (INT), and the third-quarter IMS (IMS3). The non-event periods include the 100 trading days, t − 60 to t − 10 and t + 10 to t + 60, relative to announcement release day t = 0. In the illustrative example above the non-event period associated with IMS1 (IMS3) excludes the trading days t − 10 to t + 10 associated with the PE (INT) announcement, while the PE (INT) non-event period excludes trading days t − 10 to t + 10 associated with IMS1 (IMS3).

Display full size

We follow DeFond et al. (2007) and Landsman et al. (2012) and define abnormal return variability, , as firm i’s average squared market-model residual, , divided by the variance of firm i’s market-model residuals during the non-event period, , that is, . Average squared market-model residual, , is calculated over three-day windows, t = −1,0,1, relative to release day, t = 0. is firm i’s daily market-model residual variance and is calculated from a maximum of 100 trading days during the non-event period. Because is skewed to the right, we follow Landsman et al. (2012) and take the natural logarithm, that is, our measure of abnormal return variability is defined as:(2)

Following DeFond et al. (2007) and Landsman et al. (2012) we define abnormal trading volume, , as the average trading volume during the event period, , scaled by the average trading volume during the non-event period, , and we follow Landsman et al. (2012) and take the natural logarithm, that is, our measure of abnormal trading volume is defined as:(3)

Daily trading volume, , is defined as firm i’s number of shares traded on day t divided by the number of shares outstanding on day t and is calculated from Datastream items UVO and NOSH. and are defined, respectively, as firm i’s average daily trading volume over the three-day event window, t = −1,0,1, and as firm i’s average daily trading volume over the non-event period, t − 60 to t − 10 and t + 10 to t + 60, relative to announcement day t = 0. As with abnormal return variability we exclude, where necessary, days t − 10 to t + 10 associated with an overlapping window from the IMS and the result announcement non-event period.

Table 4 Panels A and C show that in 2009 and 2010, median IMS abnormal return variability and median IMS abnormal trading volume are always positive and significant, or, in one case, positive and marginally significant, consistent with the idea that IMSs provide relevant information to investors. We also find that abnormal return variabilities and, to a lesser extent, abnormal trading volumes, associated with third-quarter IMSs tend to be higher than those associated with first-quarter IMSs, indicating that the information content is higher for IMSs released in the second half of the financial year. At the same time, we find that IMSs exhibit abnormal return variabilities and abnormal trading volumes that are lower than those associated with interim and annual result announcements. Perhaps this is not too surprising: While IMSs are typically short trading updates, interim and annual result announcements typically contain information on a much larger number of financial statement line items.

Following Cuijpers and Peek (2010) we also provide, in Table 4, Panels B and D, estimates of abnormal return variability and abnormal trading volume associated with interim and annual result announcements in 2006, and we observe that their magnitudes are similar to, and not significantly different from, the corresponding interim and annual result values in 2009 and 2010.¹⁵ This observation is consistent with the finding in Cuijpers and Peek (2010) that increasing reporting frequency leads to more relevant information being released through periodic disclosures, relative to non-periodic disclosures. This is true because significantly positive IMS abnormal return variabilities and significantly positive IMS abnormal trading volumes are not accompanied by a significant reduction in abnormal return variability and abnormal trading volume at interim and annual result announcements dates.

Overall, Table 4 suggests that, in the UK, IMSs are important vehicles to communicate relevant information to equity investors. In order to investigate whether this information is likely to be disclosed in a regime without IMSs, we need an estimate of whether or not IMSs are incremental. We obtain this estimate in two stages. In the first stage, we regress, in 2006, the total number of trading statements in the first and in the second halves of the financial year on a number of firm characteristics which have been suggested as determinants in prior literature on reporting frequency and voluntary trading statements (e.g. Kasznik and Lev 1995, Miller and Piotroski 2000, Cuijpers and Peek 2010) and we use the estimated coefficients to predict values for 2009 and 2010. In a second stage, we compare the actual number of trading statements, including IMSs, in 2009 and 2010 against the number of trading statements predicted for 2009 and 2010 under a regime that operates under the MAD, but without mandatory IMSs, that is, a regime as in 2006. We then treat IMSs as incremental if the number of actual trading statements in 2009 and 2010 is higher than the number of predicted trading statements by a margin of 0.75.

Possible determinants of a firm's decision to issue a trading statement, either voluntary or ad hoc, are listed in Table 5. The first three classes of variables, information environment, agency cost and American depositary receipt (ADR) status, are similar to the corresponding classes in Cuijpers and Peek (2010).¹⁶ We include loss status as a further determinant as Hayn (1995) finds that earnings information is less relevant to loss firms than to profit firms. Hence loss firms might feel less obliged to issue earnings guidance.

Change in operating profit is our proxy for the amount of relevant earnings information that managers are likely to hold privately.¹⁷ We include separate variables for positive and negative earnings changes as managers may have different incentives to pre-empt, via trading statements, the disclosure of positive and negative earnings surprises (e.g. Skinner 1994).

Special events, like restructuring, often have significant implications for sales and earnings. To the extent that special events-induced changes in sales and earnings are communicated via trading statements, the occurrence of special events should positively impact trading statement frequency. On the other hand, if special events are communicated through news announcements other than trading statements, then one might expect a negative association between specials events and trading statement frequency, as other news announcements render trading statements redundant. We use three proxies for special events, namely the size of (a) exceptional income, (b) exceptional charges, and (c) restructuring charges. Finally, we follow Kasznik and Lev (1995) and Miller and Piotroski (2000) and include industry sector dummies and dividend changes as additional determinants. For example, Kasznik and Lev (1995) show that firms in sectors with high regulation, like telecommunication and utilities, have a lower tendency to issue a fourth-quarter profit warning, while Miller and Piotroski (2000) find that dividend increases are associated with a lower incidence of forward-looking earnings statements.

Next, we use 2006 observations and regress the number of a firm's trading statements in the first and in the second halves of the year, respectively, on all variables in Table 5, and for each variable we present two coefficients in Table 6, namely one from a univariate regression, with only one variable at a time explaining trading statement frequency, and one from a multivariate regression, with all variables in Table 5 explaining trading statement frequency collectively. In Table 6 we estimate separate models for first- and second-half trading statement frequency and we do this for two reasons. First, Table 3 demonstrates that the average number of trading statements per firm varies considerably between the first and the second halves of the year, suggesting that different factors and incentives might be at work. Second, it is likely that a firm's decisions to issue (a) a first-half and (b) a second-half trading statement are correlated with each other meaning that pooling first- and second-half observations is likely to yield regression residuals which are no longer independent (a specification problem which would be difficult to rectify through the use of clustered standard errors given that each firm has only two observations).

Looking first at the univariate coefficients in Table 6 we find that several variables from Table 5 are significantly associated with a firm's decision to issue a trading statement. For example, analyst following and market value are positively associated with the number of first- and second-half trading statements, while loss status and R&D intensity are negatively associated with first- and second-half trading statements. The latter finding is inconsistent with the positive association in Cuijpers and Peek (2010) and suggests that R&D-intensive firms prefer not to talk about sales and earnings between interim and annual result reporting dates. In addition, the incentive to issue trading statements appears to vary across industry sectors, with firms in the consumer goods and consumer service industry typically issuing more trading statements than firms in other industries. Overall, it appears that the decision to issue trading statements is affected primarily by variables which change relatively little over time, like firm size, stock-market index and industry membership.

When we run multivariate regressions we notice that very few variables remain significantly associated with trading statement frequency. For example, in the first-half regression no variable remains significant at the 5% level. This means that, in our sample, the contribution of each individual variable, over and above the contribution of all other regression variables, is not significant. Nonetheless, for a test of H2 and H3, we still use the coefficients from the multivariate regression, and there are two reasons for this. First, in predicting which IMS is incremental and which IMS is non-incremental, we are concerned primarily with the overall explanatory power of the prediction model, as indicated in Table 6 by R² and F-value, rather than with the significance of individual regressors, as indicated by the p-value. Second, any attempt to eliminate individual (insignificant) regressors from the multivariate regression could be criticised as an example of over-fitting the data.

For a formal test of H2 and H3 we use the multivariate regression coefficients in Table 6 to predict the number of trading statements that would be issued in 2009 and 2010 in the absence of a mandatory IMS regime, and we treat IMSs as incremental if the actual number of trading statements, including IMSs, in 2009 and 2010, is higher than the predicted number by a margin of 0.75 though we notice that our findings remain qualitatively unchanged if we used margins of 0.50 or 1.00 instead.¹⁸ We then calculate abnormal return variability and abnormal trading volume associated with incremental and non-incremental IMSs. This is done, in Table 7, separately for first-quarter IMSs and third-quarter IMSs, IMS1 and IMS3.

In Table 7, Panel A, we observe that the number of incremental IMS3s is larger than the number of incremental IMS1s. This is consistent with the evidence in Table 3 that the introduction of IMSs has increased the frequency of trading statements more in the second half of the financial year than in the first half of the financial year.

More importantly, in Table 7, Panel A, median abnormal return variability and median abnormal trading volume associated with incremental IMS3s are positive and significant, with medians of 0.23 and 0.19 and p-values of .039 and .002. In contrast, median abnormal return variability and median abnormal trading volume associated with incremental IMS1s are lower at 0.08 and 0.11, and, with p-values of 1.000 and .694, are not significantly different from zero. On that basis, we formally reject the argument of no loss of relevant information, as reflected in the null of H2, in favour of the alternative that the withdrawal of mandatory IMS3s, but not the withdrawal of mandatory IMS1s, will involve the loss of some relevant information.

So far we have identified incremental IMSs by calculating the difference between the number of actual and predicted trading statements and by assuming that any additional statement is the IMS, not one of the other trading statements. While the overall picture in Table 3 is consistent with this assumption, we cannot rule out that, for individual firms, the additional statement is not the IMS.

Thus, as an alternative prediction model, we examine, for each firm with a listing in both the post-IMS year and the pre-IMS year, the number of trading statements in the pre-IMS year, and we define an IMS as incremental only if the number of announcements in each class of trading statement – that is, AGM trading statement, quarterly report, (early) trading statement, and (pre-) close period trading statement for IMS1, and quarterly report, (early) trading statement, and (pre-) close period trading statement for IMS3 – is at least as great in the post-IMS year as in the pre-IMS year. The median abnormal return variability and abnormal trading volume for this alternative definition of incremental IMSs is given in Table 7, Panel B.

We find that the results in Panel B are qualitatively similar to the results in Panel A. First, the number of incremental IMSs is much higher in the second half of the year than in the first half of the year. Second, median abnormal return variability and median abnormal trading volume associated with incremental IMS3s continue to be significantly positive, while median abnormal return variability and median abnormal trading volume of incremental IMS1s are insignificantly negative. Thus, we continue to reject H2 for incremental IMSs in the second half of the year, but not for incremental IMSs in the first half of the year.

If a MAD-regime without IMSs is sufficient in ensuring that most relevant information is released to equity markets, then one would expect that the typical market reaction associated with a trading statement in the MAD-only regime, that is, non-incremental IMSs, is higher than the typical market reaction associated with trading statements that are only issued in a combined MAD/IMS regime, that is, incremental IMSs. In Table 7 we test this prediction by comparing the abnormal return variability and abnormal trading volume between incremental and non-incremental IMSs. The corresponding one-tail test p-values for differences in median abnormal return variability and median abnormal trading volume are given in the last column of Table 7.

We find that median abnormal return variability and median abnormal trading volume are consistently higher for non-incremental IMSs than for incremental IMSs, and the observed differences are always significant when we compare non-incremental third-quarter IMSs against incremental third-quarter IMSs. In contrast, for first-quarter IMSs the differences between non-incremental IMS1s and incremental IMS1s are significant, or marginally significant, for the constant sample in Panel B, but not for the full sample in Panel A.¹⁹

Collectively, the evidence in Table 7 is supportive of the prediction in H3. In particular, the finding of consistently higher medians for non-incremental IMSs suggests that the MAD, in its own right, is effective in ensuring the release of most relevant information, consistent with the Commission's arguments in 2011 and the formal withdrawal of mandatory IMSs in 2013.

6. Conclusion

At the centre of debates about reporting frequency lie two key issues. First, to what extent does an increased reporting frequency lead to an increase in the total amount of corporate disclosures? Second, to what extent does any additional corporate disclosure lead to more relevant information being released early, for example, by bringing forward the disclosure of relevant information? The present paper attempts to answer both questions. It does this in the context of the newly mandated IMS whose formal withdrawal will have to be implemented by EU member states by late 2015.

Our empirical analysis proceeded in two stages. In the first stage, we collected information on trading statement frequency, including IMSs, in 2009 and 2010 and compared this frequency against trading statement frequency in 2006. This allowed us to evaluate how the introduction of mandatory IMSs has affected the frequency of non-periodic trading statements. Second, we used trading statement frequency in 2006 to estimate the number of trading statements that firms make in the absence of a mandatory IMS, and we used this number to predict which IMS is a genuinely incremental firm announcement, and not simply a substitute, and we calculated the event-period abnormal return variability and abnormal trading volume associated with these incremental and non-incremental IMSs.

Our findings suggest that, in the UK, the introduction of mandatory IMSs often coincided with the withdrawal of non-periodic trading statements, consistent with a large negative interaction effect between periodic and non-periodic disclosures. This substitution effect is more pronounced in the first half of the financial year when the existence of an AGM trading statement offers significant scope to substitute new disclosures for old disclosures. In contrast, the majority of non-periodic trading statements in the second half of the financial year come in the form of (pre-) close period trading updates which firms seem more reluctant to withdraw. Overall, our evidence suggests that the timing and nature of prior disclosures is an important determinant for the size of the substitution effect. To the extent that the timing and nature of prior disclosures vary across countries, one should also expect the substitution effect to vary. We believe that examining the substitution effect across other countries would be a natural extension of our analysis.

Consistent with the finding that mandatory IMSs have increased trading statement frequency more in the second half of the year than in the first half of the year, we find that third-quarter IMSs are more likely to be incremental than first-quarter IMSs. Crucially, incremental third-quarter IMSs exhibit significantly positive abnormal return variability and abnormal trading volume consistent with the idea that the withdrawal of IMSs will lead to some loss in investor protection. At the same time, we observe consistently higher abnormal return variabilities and abnormal trading volumes for non-incremental IMSs, relative to incremental IMSs, and this is consistent with the argument that a MAD-only regime, without mandatory IMSs, will ensure the release of most relevant information, a key argument in favour of the decision to formally withdraw IMSs by 2015.