To differentiate between the effects of earnings announcements and their corresponding conferencecalls, we control for the tone of the press releases by separating the conferencecalls into two sections as in Matsumoto, Pronk, and Roelofsen (2011). Kimbrough (2005) and Matsumoto, Pronk, and Roelofsen (2011) argue that the prepared statements by management at the beginning of a call essentially reiterate the information in the carefully crafted press release. Thus, we use the tone of the introductory statements prepared by management as a proxy for the tone of the earnings announcements (INTRO TONE). Following the prepared remarks, the remainder of the time is opened up for discussion and management entertains questions from call participants. As such, Q&A TONE represents the linguistic tone for the question and answer session. The variables INTRO TONE and Q&A TONE are constructed exactly as the TONE variable described above, but are limited to the introductory and discussion portions of the conferencecalls, respectively. In addition to the combined (or overall) tone, we decompose Q&A TONE into MGR Q&A TONE and ANALYST Q&A TONE. We construct these two tone measures by tracking the source (manager or analyst) of each comment during the Q&A session. INTRO COUNT and Q&A COUNT represent the respective word counts for the two call
You must not allow conferencecalls to become a constant interruption to your work sched- ules. You have a responsibility to yourself to carefully consider whether a request to take part in a conference call is the best use of your time and what value you feel you can con- tribute. It is so easy to be diverted by the ringing of your phone or beeping of a PDA that you often forget to view this interruption in the same way you would a meeting request. Such calls must be seen as a cost-effective and efficient use of people’s time, both inter- nally and externally. Your organization and any others that you need to have conferencecalls with must be educated in how and when these calls are a good use of everyone’s time and expertise.
As regards forecast dispersion, we generally fail to detect a statistically significant change during conference call quarters. We find only weak evidence of an increase in fore- cast dispersion during conference call quarters for the longer measurement period. To gain a deeper insight into the the relation between forecast dispersion and conferencecalls, we separate forecast dispersion into its underlying theoretical components, uncertainty and information asymmetry, where forecast dispersion is an increasing function of both (Bar- ron, Stanford and Yu, 2009). We find that overall uncertainty declines after conferencecalls, while there is a strong increase in information asymmetry among analysts. Thus, the absence of a detectable change in the forecast dispersion after conferencecalls suggests that the decrease in uncertainty is offset by a proportionate increase in information asym- metry. This finding is consistent with theoretical models (e.g., Kim and Verrecchia, 1994) and empirical evidence (e.g., Barron et al., 2002) demonstrating that anticipated public disclosure can lead to increased information asymmetry, because analysts generate new private information in addition to the public information released during such disclosure events.
Second, I contribute to the literature that examines whether conferencecalls provide material information to conference call participants. Prior research finds significant trading activity at the time of the call (Frankel et al. 1999; Bushee et al. 2003; Bushee et al. 2004, Lansford et al. 2009), improvements in analyst forecast accuracy following the call (Bowen et al. 2002), more timely incorporation of earnings news into prices for firms initiating conferencecalls (Kimbrough 2005), and a reduction in information asymmetry for firms holding regular quarterly calls (Brown et al. 2004). Matsumoto et al. (2011) find incremental information content for both the presentation and the Q&A sessions of the call relative to the press release. In this paper, I examine the variation within firms that provide conferencecalls and find that some firms provide less information than others. In particular, the results of my analyst forecast accuracy and timeliness tests suggest that conferencecalls are less informative when firms prepare scripted responses to analysts’ questions in advance of the call. This evidence complements research that suggests that greater disclosure does not always result in better information for market participants. For example, Lawrence (2013) argues that irrelevant information disclosure can potentially confuse and misguide individual investors and limit their processing of material financial information (see also Skinner 2003). My results suggest that scripted conferencecalls are less informative even for more sophisticated investors (i.e., analysts).
Conferencecalls die naar aanleiding van een kwartaalbericht worden gehouden, zijn de laatste jaren steeds populairder geworden, maar er is weinig bewijs voor het incrementele nut dat een conference call heeft boven het kwartaalbericht. In hoofdstuk 2 van dit proefschrift wordt één potentiële factor die in additioneel nut kan resulteren onderzocht, de mogelijkheid om gedurende het vraag-en-antwoord deel van de conference call nieuwe informatie aan het management te ontlokken. Op basis van een steekproef van ongeveer tienduizend transcripties van conferencecalls, onderzoek ik de relatie tussen het deel van de conference call waarin het management een presentatie geeft en het discussiedeel van de conference call. Ik vind dat de discussieperiode van de conference call langer wordt naar mate het management in de presentatie abnormaal weinig toelichting geeft. Daarnaast is de relatie tussen weinig toelichting in de presentatie en langere discussies sterker naarmate meer analisten de onderneming volgen. Ook vind ik bewijs dat (in het algemeen) langere calls meer informatief zijn voor de markt; langere calls zijn gerelateerd aan grotere opvolgende aanpassingen in de voorspellingen van analisten en een verbetering in de juistheid van de voorspellingen van analisten. Ten slotte onderzoek ik of bepaalde typen toelichtingen in de presentatie, leiden tot kortere discussie. Ik vind dat meer financieel getinte toelichtingen in de presentatie leiden tot een kortere discussie, terwijl meer toekomstgerichte toelichtingen tot een langere discussie leiden. Bij elkaar genomen, suggereren deze resultaten dat analisten tijdens de conference call een actieve rol spelen in het onthullen van informatie, waardoor een rijkere informatieomgeving ontstaat dan anders het geval zou zijn.
On a separate, but related point, China can be a pain, not just in terms of cost but also in terms of usability. Many China dial-in numbers only actually work from a limited set of service providers within China, which can cause a whole set of issues for guests joining conferencecalls,
2016). We predict that the market negatively reacts to pessimistic managerial mood from bad weather during the call date. The results support our prediction: controlling for net tone and quantitative information in the call, we find that bad weather decreases the market reaction (cumulative abnormal return from one day before to one day after) by 0.21% which translates into an average $20 million market value reduction in firm market capitalization. Although fluctuations in cloud cover are likely to capture only a fraction of variation in managerial mood, this result suggests that associated mood changes play an economically material role in market reaction beyond that which is captured by textual analysis methods in the literature. This result holds when (1) controlling for investor mood; (2) examining propensity-score matched pairs; (3) controlling for clustering of our sample in some specific states. Previous research demonstrates that the tone of disclosure after an earnings announcement also influences PEAD (Feldman et al. 2010). As such, we also look at the persistence of weather effects on conferencecalls by
The idea behind the test is that if analysts do ask questions that are privately valuable to them in completing their private valuations, or updating their private models on the firm (as we find suggestive evidence for with respect to their individual subsequent forecast accuracies following conference call participation), we might expect to see this reflected in their reports subsequent to the calls. We find evidence consistent with this idea. In particular, the text of analysts’ reports show significantly more similarity to the text of the questions they themselves just asked in the earnings call, compared to: i.) those of all other analysts covering the same firm, and ii.) those that the same analysts published prior to the call. An important caveat to these results is that inclusion in the call could be proxying for better access in general. Thus, the analysts who are asking their questions during the call may also be those that are receiving differential access to management. It could then be the case that they press deeper on the same issues in the private meetings that they inquire about in the conference call (e.g., same store sales growth in Europe). We also present a number of anecdotal examples of text from questions asked by these analysts in the call in Appendix Table A11 that appear to support this idea as well. Thus while only suggestive, these additional findings do provide more depth and color on one potential channel through which these bullish analysts gain a relative advantage on accuracy (i.e., through access to asking privately valuable questions both through conferencecalls—on which we have evidence—and perhaps also through access outside of these calls).
In comparison, there is little academic evidence on non-U.S. firms’ use of conferencecalls. Bassemir et al. (2012) examine the conferencecalls held by German firms listed on the Prime Standard Index of the Deutsche Börse. The stock exchange mandates that, as one of the disclosure requirements for being part of the index, firms must conduct at least one conference call per year. Bassemir et al. (2012) find that firms conduct on average two (mostly closed) calls per year, even though they are required to report earnings on a quarterly basis. Meanwhile, using a set of Thai firms, Liang et al. (2012) find foreign ownership to be an important motive behind firms in emerging markets holding a conference call. Firms may decide to hold conferencecalls based on the institutional features of their home countries and/or the main stock exchange on which they are listed. However, to our knowledge, there is no other cross-country study investigating conferencecalls.
correlated omitted variable, I follow Ke and Yu (2005) and add to equation (1) a variable that equals 1 if the analyst’s most recent outstanding quarterly earnings forecast is below the consensus forecast, and zero otherwise. Results (not reported) show that adding this variable has a negligible effect on the magnitude and significance of the coefficients reported in Table 4 and inferences remain unchanged. However, the pessimism logit coefficient is positive and significant (0.139, p<.001), implying that pessimistic analysts are more likely to participate on conferencecalls compared to other analysts. These results are consistent with Ke and Yu (2005), although the effects of forecast pessimism on the odds of participating are an order of magnitude lower than the effects of strong buy or strong sell recommendations.
We obtain conference call transcripts, spanning the period of January 2001 to June 2007, from Thomson StreetEvents, a division of the Thomson Reuters news service and database vendor. We first restrict our sample to transcripts that are identified in the StreetEvents database as earnings related, and to those for which we are able to extract a reliable call start time, city, and company name and/or firm ticker information. We further restrict the sample to transcripts in which each of the management address, analyst question, and management answer portions of the call exceed 50 words, and to firms that are publicly-traded and headquartered in the United States. We focus on conferencecalls that we are able to confirm to be related to earnings announcements, which we define as falling into a window of [0, 2] days relative to the t=0 earnings announcement day (as defined by either Compustat or I/B/E/S), and for which we are able to extract location time zone. We require firms’ annual and quarterly financial data to be available from Compustat, their stock data to be available from Center for Research in Security Prices (CRSP) and Trade and Quote (TAQ), and their analyst forecast and manager guidance data to be available from I/B/E/S. We exclude the conferencecalls of firm-quarters characterized as having a negative common book value of equity. We require some uniformity in call participants’ presumed body clocks and thus focus on calls that are initiated in Eastern or Central time zones. In our calculations of residual tone (discussed in detail in Section 3.2), we make use of the entire sample of available Eastern and Central time zone call transcripts (i.e., spanning all call start times). For our primary time-of-day tests, however, we restrict the sample to calls initiated during the window of 08:00 to 15:59 Eastern Time (i.e., calls initiated prior to the close of trading). The imposition of all of these constraints yields a sample of 18,408 calls initiated by 1,865 distinct firms. In some robustness checks, we also include calls initiated in Pacific Time locations. Details related to the impact of each of the sample inclusion criteria on the final determination of the sample are summarized in Table 1.
4. Discussion focussed on the posting of the minutes of Bureau conferencecalls to the UNECE website. The Chair advised that the aim was to produce a set of minutes with limited or no personal attributions and once written approval had been received from Bureau members the minutes could be posted.
we adjust the event window metrics by subtracting from each measure the respective average for the same window on the same trading day during the prior four weeks. In other words, the raw trading volume (volatility/range) during the conference call occurring on a Tuesday between 11:30 am EST and 12:30 pm EST will be adjusted by subtracting the mean of the trading volume (volatility/range) within the same one hour window on the four prior Tuesdays: days -7, -14, -21 and -28 from the conference call date. For the peer firms we exclude non-event days with earnings announcement releases. Thus, abnormal volume, volatility, and range metrics represent unusual levels of trading for a given firm-window: Positive (negative) values imply greater (smaller) trading activity during conferencecalls or earnings announcement intraday windows than during comparable non-event windows for a given firm. Untabulated descriptive statistics confirm that, as expected, the means and medians of all metrics for conference call holding firms are significantly different from zero on both the earnings announcement and the conferencecalls windows. The mean of the abnormal metrics for peer firms are small and positive, suggesting that they experience a slightly higher than normal level of trading activity over the earnings announcement and conference call windows. 21 Replicating the structure of our main results we estimate models (1a) and (1b) for the three non-directional metrics in Table 7. Because the variables are normalized relative to non-event period trading patterns, the coefficients β 1
2. Challenge —How can Districts, Al-Anon Information Services, and other Al-Anon Committees use conferencecalls to derive similar benefits (reduced travel costs and less time away from home for the Group Representatives and other Trusted Servants) from conferencecalls as the Areas?
sentences using the Python NLTK program. The identification of FLD sentences combines techniques from the accounting literature (Matsumoto et al., 2011; Muslu et al., 2014) and computational linguistics (Bird et al., 2009). Python NLTK provides functions to classify forward-looking and non-forward-looking sentences. However, forward- looking identification using NLTK contains measurement error because it is not specifically designed for financial reporting language. For example, the words ‘expect’ and ‘anticipate’ are routinely used to deliver management guidance and forecasts, but NLTK does not classify sentences containing these words as forward-looking in some instances. Therefore, this study follows Muslu et al. (2014) and Matsumoto et al. (2011) to develop wordlists to identify FLD in conferencecalls. Matsumoto et al. (2011) provide a wordlist to identify FLD in conferencecalls; Muslu et al. (2014) provide a more comprehensive FLD identification scheme based on 10-K filings. Since written documents are different from oral communication, the wordlist in this study combines and modifies the identification schemes in these two studies into a comprehensive FLD identification scheme specifically for conferencecalls. The classification scheme classifies a sentence as forward-looking if it: (1) contains words/phrases that indicate future time periods (e.g. “future”, “next quarter”, “next year”, etc.); (2) contains verbs or their conjugations that indicate future expectations, plans or actions (e.g. “anticipate”, “aim”, etc.); (3) contains a reference to a year after the year of the call; (4) contains other words/phrases that are typically used in management guidance (e.g. “guidance”, “projection”, etc.); or (5) is classified as forward-looking by Python NLTK. Further details are provided in Appendix 2.1.
While the primary focus of our paper is whether the informational cues in managerial linguistic complexity on conferencecalls provide a directional prediction about analyst forecast revisions, we also examine whether such cues improve accuracy and reduce dispersion in analyst forecasts. A greater amount of informative linguistic complexity on the call allows analysts to more precisely assess the persistence of the earnings shock, which should lead to improvements in forecast accuracy and reductions in forecast dispersion relative to calls with less informative disclosure. For linguistic complexity that represents obfuscation, it is possible that all analysts view obfuscation as a similarly negative signal about future earnings news, and hence become more accurate in their forecasts. However, prior work suggests that obfuscation is successful in creating uncertainty about the true nature of the news (Bushee, et al. 2018). If managerial obfuscation on the call leads analysts to put less weight on the new public information in the call and rely more on private information or on stale information, then the accuracy of their forecasts should decrease and dispersion should increase.
The learners’ need to socially interact with others was one of the most recurrent patterns identified throughout the data collection and analysis stages of this study. It was found that this social interaction need was evident in a high number of occasions all through the students’ surveys, focus groups, and the teacher’s reflective instrument. More specifically, it was found that learners seemed to notice a relevant connection between language and social interaction practices. In this respect, Fitch and Sanders (2005) identify a bilateral relationship between language and social interaction: “Language use is examined with reference to the social interactions that comprise it and social interactions are examined with reference to the language use that forms and organizes them” (p. 4). From this statement, it could be inferred that learners in this study seemed to notice the relevance of using the target language with the main purpose of socially interacting with their peers through the use of Skype™ conferencecalls. The following excerpts, collected from the same learner in different data collection instruments and corroborated by an entry taken from the researcher’s reflective instrument, show clear examples of the above discussed need for social interaction through the use of the target language: