سفارش تهاجمی به عنوان یک اندازه گیری برای ارزیابی سودمندی اطلاعات حسابداری

Motivated by the availability of high-frequency data on trading activity, this paper proposes the use of order aggressiveness as a metric to evaluate the usefulness of accounting information. I test, through an analysis of order aggressiveness, whether earnings announcements of firms listed on the Italian Stock Exchange limit order book have information content. I estimate an ordered probit relating order aggressiveness to unexpected earnings and to three market determinants of aggressiveness. Consistent with the theory on the choice between limit and market orders, I find that order aggressiveness increases with the absolute value of unexpected earnings. The results provide evidence on the extent to which the information contained in earnings is used by traders.

The vast body of literature on decision usefulness of accounting information examines the market response to the disclosure of financial statements items (recent critical surveys are contained in Fields et al., 2001 and Kothari, 2001). Most contributions, drawing their foundations in the seminal works of Ball and Brown, 1968 and Beaver, 1968, analyze the reaction of stock prices and trading volume. A group of works also compares price-based metrics and volume-based metrics as methods to test the market response (Rohrbach and Chandra, 1989, Bamber and Cheon, 1995, Cready and Ramanan, 1991, Cready and Ramanan, 1995 and Cready and Hurtt, 2002). The availability of high-frequency data on orders and transactions makes it possible to enrich the results of this branch of research. Accordingly, I propose the use of order aggressiveness as a complementary metric to evaluate the usefulness of accounting information. Ranking orders according to aggressiveness is a classification, first introduced by Biais, Hillion and Spatt (1995), used by a number of empirical studies examining the order flow. Stock prices and trading volume are determined by order submission strategies, which are described by the aggressiveness classification. Thus, order aggressiveness provides a metric of the market reaction to information more primitive than measures based on stock prices and trading volume. Further advantages exist in the method I propose with respect to the traditional metrics: the effect of unfilled orders is taken into account, an intraday dataset can be exploited, and results can be segmented by characteristics of orders (therefore by categories of traders). As pointed out by Lev (1989), the traditional approach to evaluate decision usefulness maintains that when individuals act as if they use information, then such information can be considered useful. Following this approach, I suggest that analyzing order submission strategies provides evidence on the extent to which accounting information is useful for market participants. Different actors can be motivated to understand whether accounting information is found to be useful through an analysis of order aggressiveness. As implied by many works in this literature (see, for example, Scott, 2003), accountants are interested in understanding how market participants perceive the information they release, and standard setters are interested in investigating if accounting methods are appropriate to convey information. Moreover, the observation of traders' reactions helps investors to revise their evaluation of the stocks and to better estimate trading costs. This paper studies the usefulness of earnings announcements of firms listed on the Italian Stock Exchange limit order book by examining the reaction of order aggressiveness to unexpected earnings. I estimate an ordered probit that relates order aggressiveness to the absolute value of unexpected earnings and to the main market determinants of aggressiveness identified by previous literature (the bid-ask spread, the depth of the order book, and price volatility). In line with the predictions of microstructure models on the choice between market and limit orders, I find that aggressiveness of both buy and sell orders increases with the absolute value of unexpected earnings. I interpret the existence of this relationship as evidence that traders actually use the information contained in earnings. Consistently, abnormal trading volume and abnormal volatility signal a market reaction to the information disclosure. Nevertheless, when examining abnormal returns, a clear indication that market behavior is affected by earnings announcements does not emerge. The analysis presents an empirical application of the metric I propose, and it aims to provide a methodological contribution to the decision usefulness research. The research design can be employed with data from the other exchanges to test the effect of different forms of information disclosures. It should be noted that examining order aggressiveness entails an informational advantage with respect to observing stock prices and trading volume, but it requires a greater amount of data. Potentially interesting areas of application of the metric I derive concern the reaction of different types of traders (e.g., buyers vs. sellers, institutionals vs. individuals) to the information release. The remainder of the paper is organized as follows: Section 2 summarizes the related research; Section 3 presents the motivation of the empirical analysis; Section 4 describes the dataset; Section 5 is devoted to the specification of the empirical model; Section 6 comments on the results; and Section 7 concludes the study.

Motivated by the availability of high-frequency data on orders and transactions, this work proposes the use of order aggressiveness as a metric to evaluate the usefulness of accounting information. I examine the effect of unexpected earnings on order aggressiveness for firms listed on the Italian Stock Exchange limit order book. I consider the annual earnings announcements from 2003 to 2005 for the 40 stocks belonging to the S&P/MIB index. I estimate an ordered probit that relates order aggressiveness to unexpected earnings and to the market determinants of aggressiveness identified by previous literature. The main finding is that order aggressiveness increases with the absolute value of unexpected earnings. This can be interpreted as evidence that part of the information released is actually used by market participants. I also compare these results to the results obtained by using traditional metrics of decision usefulness. The analysis of abnormal trading volume and abnormal volatility documents a significant market reaction to the information disclosure. By contrast, tests of decision usefulness based on abnormal returns do not generally allow the rejection of the hypothesis that the market does not react to the announcements. Order aggressiveness provides a metric of the market reaction to information more primitive than measures based on stock prices and trading volume. Moreover, it reflects a larger information set than stock prices and trading volume do, and it allows segmentation of the results by characteristics of orders. These features make order aggressiveness suitable as a complementary metric of decision usefulness. A limitation of the approach I propose is reflected by the great amount of data required to analyze order aggressiveness (in this example, there are 1,528,478 orders/observations). For example, processing high-frequency data for all the firms in COMPUSTAT, which represents the benchmark sample used by a number of studies in this literature, would not be feasible. This is why I limit my analysis to 40 firms over three years. Yet, the main contribution of this paper is methodological, and its objective is to present an application of the new metric. It is out of its scope to study the response to earnings announcements across a large sample of stocks or over a long time period. The method described to examine order aggressiveness in the Italian exchange can be applied to other stock markets. The Italian exchange is a pure order-driven market, as are, for example, Euronext (which groups the Belgian, French, Dutch, and Portuguese markets), the London TradElect, the Scandinavian and Baltic exchanges (that use the OMX platform), the Toronto, Shanghai, and Tokyo exchanges, and the Electronic Communication Networks (which interact with the NYSE and the NASDAQ). In quote-driven markets, orders received by market makers can be classified analogously. Previous works (e.g., Ellul et al., 2003 and Beber and Caglio, 2006) classify orders according to their aggressiveness in the hybrid trading structure (partly quote-driven and partly order-driven) of the NYSE. In a number of applications, decision usefulness research can benefit if a metric of market reaction based on traders' strategies is employed. It is crucial to consider the informational advantage associated with examining order aggressiveness with respect to observing stock prices and trading volume. The informational advantage increases as the number of orders as a proportion of the number of contracts grows. This ratio increases when price discovery is important and traders, before committing to contracts, submit orders in an attempt to infer information from the order book. At the extreme, order aggressiveness reflects a relatively greater information set than traditional metrics in batch auctions, where orders are grouped together and a single price is determined. Batch auctions are used in many exchanges, including the NYSE and the NASDAQ, to open or close the markets; in other exchanges, for example in a segment of Euronext, they are the sole means to trade illiquid stocks. A further area of application of the metric I derive concerns the reaction of different types of traders (e.g., buyers vs. sellers, institutionals vs. individuals) to the information release. This is related to a recent stream of works (Grinblatt and Keloharju, 2000, Bhattacharya, 2002, Cohen et al., 2002, Ekholm, 2006, Vieru et al., 2006 and Dey and Radhakrishna, 2007) that also consider behavioral biases as drivers of investor decisions after the news. When the research question is whether buyers and sellers react differently, observing prices and trading volume alone is not sufficient; studying the characteristics of orders submitted is a possibility to overcome this limit of traditional metrics of decision usefulness.