Research Methodology and Data Collection

1987 described in Table 1, our initial sample consists of 4,483 management forecast disclosures issued by management of 287 companies listed in the SET. ================================= Insert Table 1 here. =================================

4. Empirical Results

Table 1 provides sample construction for descriptive statistics. We require data on the firm-quarter level to analyze characteristics of forecast firms. 360 We remove management forecasts of each firm which are redundant forecasts in each quarter. In doing so, we obtain 1,368 firm-quarters in our sample. We also remove 98 firm-quarters without returns data available in the DATASTREAM database and exclude 79 firm-quarters without earnings data available in the DATASTREAM database. Finally, we obtain 1,191 firm-quarters from 263 firms. Table 2 separately provides descriptive statistics of all Thai listed firms, forecast firms, and non-forecast firms. In addition to means and standard deviations, Table 2 also summarizes various percentile values for price -deflated unexpected earnings UE, earnings variations SD_NI, return variations SD_RET, and market capitalization MV. ================================= Insert Table 2 here. ================================= Mean of UE of forecast firms is positive UE = 0.294 while that of non-forecast firms is negative UE = -13.140. However, mean of UE for forecast firms is insignificantly different from mean of UE for non-forecast firms t statistic = 0.390. This is inconsistent with the notion that larger earnings surprise firms are more likely to provide management forecast than are smaller earnings surprise firms [Ajinkya and Gift 1984, Kasznik and Lev 1995, and Supattarakul 2003]. Mean of SD_NI for forecast firms is significantly greater than mean of SD_NI for non- forecast firms t statistic = 7.583 suggesting that higher earnings variation firms are more likely 360 Prior studies on firm‘s forecast choice also use firm-quarter level in their analysis [Kasznik and Lev 1995 and Supattarakul 2003]. 1988 to provide management forecasts than are lower earnings variation firms. This evidence is inconsistent with prior studies which find that forecast firms have less earnings variability than non-forecast firms [Imhoff 1978, Cox 1985, and Waymire 1985]. The plausible reason is that high earnings volatility firms are likely to reduce their risk by providing more relevant information to align market expectation. Mean of SD_RET for forecast firms is insignificantly greater than mean of SD_RET for non-forecast firms t statistic = 1.420. This evidence is inconsistent with prior studies which find that higher return variation firms are more likely to issue management forecasts than are lower return variation firms [Supattarakul 2003, and Chen 2003]. Finally, Mean of MV for forecast firms is significantly greater than mean of MV for non-forecast firms t statistic = 8.245. This is consistent with the notion that larger firms are more likely to issue management forecasts than smaller firms [Imhoff 1978, Cox 1985, Kasznik and Lev 1995 and Supattarakul 2003]. 4.1 How do Thai listed companies disclose their management forecasts? Most prior studies of management forecasts are limited to US firms. This study explores management forecast disclosure practices in Thailand. Specifically, we explore management forecast disclosure practice by addressing the following questions: 1 how many firms issue management forecasts?; 2 how frequently do they issue their forecasts?; 3 when do they issue their forecasts e.g., before or after the end of an accounting period?; 4 in what horizon do they employ for their forecasts e.g., quarterly or annual forecasts?; 5 what type of information is disclosed e.g., revenue or earnings forecasts?; and 6 in what form do they issue their forecasts e.g., quantitative or qualitative forms?. Answers to these questions provide empirical evidence of management forecast disclosure practice in Thailand. Results in table 3 suggest that 68 of Thai listed companies voluntarily issue their management forecasts to the public at least once during our sample periods. On average, each firm issues approximately 8.46 forecasts per year. However, most firms issue only one forecast