| Main Variables | |
| RET | = | annual stock return which is measured by subtracting the closing price on April 1st of the next year from the closing price on April 1st of the current year, and then dividing the result by the closing price on April 1st of the current year for the audited client (Chen et al., 2013) . | |
| NI | = | a proxy variable for earnings which is measured by continuing operations segment net income divided by equity market value of the current year (Chen et al., 2013) . The study by Easton & Harris (1991) indicated that earnings divided by the beginning-of-year stock price can be employed to assess the relationship between earnings and returns. Additionally, employing this measure as a substitute variable for unexpected earnings can mitigate measurement errors in current period earnings fluctuations. The coefficient of NI is denoted as earnings response coefficient (ERC). If net income increases, the stock price return can also relatively increase, signifying that stock prices can reflect the company’s value, resulting in a positive ERC. Conversely, if stock prices fail to reflect the company’s value, the ERC becomes negative. | |
| POST | = | a binary variable for whether the departure of audit-partners from Big4 to small-sized audit firm. It is set to 1 after the event occurs and 0 otherwise. | |
| SFIRM | = | a binary variable for small-sized audit firm. It is set to 1 if the audit client is audited by small-sized audit firm (Diwan audit firm), and 0 if the audit client is audited by Big4 (EY audit firm). | |
| Control Variables |
| SIZE | = | the company size which is measured by the natural logarithm of total assets (Uang et al., 2011; Yen & Chen, 2011; Chen et al., 2013) . Freeman (1987) argued that larger-sized companies have more precise estimates of unexpected earnings, implying a positive correlation between earnings and stock returns. Another study suggested that larger companies may employ accounting choices to reduce political costs, consequently affecting earnings quality, implying that ERC tends to be lower (Lee & Chen, 2012) . |
| LEV | = | the debt ratio which is measured by dividing the total liabilities by the total assets at the end of the year (Lee et al., 2007; Yen & Chen, 2011; Lee & Chen, 2012; Chen et al., 2013) . Some literature suggested that higher leverage ratios may make it more difficult for companies to meet debt covenant conditions. To avoid violating debt agreements, companies might engage in earnings manipulation, leading to lower ERC (DeFond & Jiambalvo, 1994) . |
| BETA | = | the system risk which is measured by capital asset pricing model using daily stock returns from the previous year (Chen et al., 2013) . According to the CAPM model, as the risk increases, investors will demand higher expected returns, leading to lower future dividend discounted values. Consequently, the responsiveness of earnings becomes lower (Collins & Kothari, 1989) . Empirical studies on the Taiwanese securities market had demonstrated a negative relationship between BETA and ERC (Lee et al., 1989) . |
| GROWTH | = | the growth ratio which is measured by dividing equity value by book value of stockholder equity. Companies with high growth rates imply greater future cash flows, leading investors to expect higher future returns, thereby resulting in a higher ERC (Collins & Kothari, 1989) . |
| AGE | = | the years of listing on SEC or OTC (Ding & Jia, 2012) . Lee & Chen (2012) suggested that companies with longer tenure in the market have reduced opportunities for information asymmetry. Additionally, Ghosh et al. (2005) found that the earnings response coefficient of a company tends to increase with the number of years it has been listed. |
| EP | = | the earnings persistence which is measured by the reciprocal of the price-to-earnings ratio (Lee & Chen, 2012) . Higher price-to-earnings ratios indicate lower earnings persistence, implying that market price surpasses the company’s earnings per share. Consequently, in situations where earnings levels are undervalued, the ERC tends to be smaller (Beaver & Morse, 1978) . |
| IND | = | industry dummy variables. |