Independent variables

category. We also want to know how the other variables affect the classification of financial assets. We also add variables that relate to the classification of financial assets according to the previous research about the earning management or window dressing.

NFT% stands for net securities investment income divided by the sum of financial assets at fair value through profits and loss (for trading portfolio) and available-for-sale financial assets. This variable provides a relative measure amount of the insurance companies to engage the gains trading during the period. We expect that the companies with lower ROE or ROA will do more active gains trading. Gains trading will help insurance companies with loss on the marketable securities to window dress the financial report and keep the operating performance well. This implementation of SFAS NO.34 and revision give the companies more freedom to decide the classification of financial investment securities by themselves. We also want to exam whether the insurance companies with different ROE and ROA will be more active to engage in the gains trading after the implementation of SFAS NO.34.

Note the higher value of NFT% represents more active gains trading.

3.2 Independent variables

We define the independent variables X as below ： ROA 、 ROE 、 leverage(debt/assets)、current ratio(current assets/current liability)、ln(assets)

The expected influence of independent variables to the dependent variables are shown in Table 1.

Table 1 The expectation of influence Y

‧

We use net income in the end year divided by the total asset in the end year.

Return on asset ratio shows how profitable a company is relative to its total assets.

This variable provides a measure of the current earning level and, this will provide the incentive of how a company will need to engage in the gains trading. Entities with lower ROA, rather than entities with higher ROA will engage in the gains trading to improve their earnings. Firms with relatively low ROA may be inclined to classify the financial assets into the available-for-sale category because they need to reduce the volatility of ROA. Jordan et al. (2011a, 2011b) uses the insurance industry as sample to research, and they find that ROA significantly influences the insurance companies to engage in gains trading. The ROA is negatively relative to the gains trading for insurance company. Therefore, we expect the ROA will have a negative relation with the available-for-sale variable and the ratio of net income to financial assets at fair value through profits and loss (for trading portfolio).

ROE

ROE measures the rate of return on the shareholders’ equity of the common stock owner. This variable shows how well a company uses investment fund to generate earning growth. ROE is best used in comparing different companies in the same industry. We know that the company with higher ROE means that it may have higher growth rate if the company reinvests in the future. Entities with relatively low ROE may give the managers more operating pressure to reduce the volatility of net income from the investment securities. On the other hand, the insurance companies investors and shareholders put more attention on the ROE. We know that higher ROE represents better operating performance of managers, and the companies also have better stock return (Clubb, Naffi, 2007). We assume the entities with lower ROE will classify more investment securities to the available-for-sale category instead of the for-trading category. And the managers will have more intention to engage in the gains trading than the managers of the insurance companies with higher ROE. The insurance company managers may hide the loss from the investment securities by reclassifying from for-trading category to available-for-sale category. This will makeup the ROE number and could also reduce the impact to stock price.

‧

The ratio represents the year-end liability to year-end total asset. Jordan et al.(2011), Christie(1990) suggest that leverage is related to the earning management for entities. Entities with higher leverage pose higher risk to creditors and investors, and therefore, they usually have the incentive to manipulate the earnings to lessen the investors and creditors’ perception of firms’ risk. But still some literatures like Jordan et al. (2011) suggest that the entities with a large amount of debt may have less need to manage its earnings through gains trading because their earnings are already boosted by the excess return resulting from the positive financial leverage. We expect the companies with a large amount of debt will intend to classify the financial asset into the available-for-sale category to reduce the volatility of net income and to lessen the investors’ perception of firms’ risk.

Current ratio

Current ratio represents the ratio of current assets to current liability. This ratio measures the liquidity demand for a company. The entities with higher current ratio will have higher demand to repay the short-term liability. That is, the entities will have fewer incentives to classify the financial assets into the available-for-sale category. On the other hand, the entities with lower current ratio, may classify the financial assets into the available-for-sale category instead of the for-trading financial asset category. As 黃 劭 彥 （ 2011 ） notes, they find the companies with higher current ratio will be inclined to make the gains trading. That is, the current ratio will be positively related to the gains trading. We expect that the variable will have positive effects on the ratio of net income to financial assets at fair value through profits and loss (for trading portfolio).

ln(assets)

This variable mainly measures the size of entities; we use the natural log of the insurance company assets. The log of assets was used rather than the absolute assets size because the latter one is usually not normal distributed. Logging is the most usual means that we used as a variable and will not sacrifice the explanation power of the asset size. There are many literatures trying to find the relationship between earning management and the entities’ assets. Guenther (1994), Jordan et al.

‧

(2011), and 黃劭彥 et al. (2011) indicate that the size may affect the level of earning management. Companies with different asset scale will face different operation pressure. For example, the managers of large companies may face more intense pressure than those of small companies. To achieve the goal of earning, the manager may be more inclined to engage in gains trading. According to 黃劭彥 et al. (2011), they find that the asset size exactly has a significant effect for the company to classify the financial asset into different categories. They note that the more assets a company has, the more likely the company will classify more financial assets into available-for-sale category. This explains companies with large asset scale have higher tolerance for the volatility of shareholders’ equity, so they are inclined to hold more available-for-sale financial assets.

3.3 Methodology