*3.2. Variables*

#### 3.2.1. Dependent Variable

When studying the relationship between firms' cash holdings and performance, to measure firms' performance, Shinada (2012) used the return on asset (denoted by ROA). Iftikhar (2017) and Vijayakumaran and Atchyuthan (2017) also used the ROA to measure company's performance. Agreeing with these above studies, we used the book value to calculate firms' performance. The measurement of firms' performance was defined as below:

$$\text{ROA} = \frac{\text{Profit before tax and interest}}{\text{Total assets}} \tag{1}$$

#### 3.2.2. Threshold and Explanatory Variables

There are two categories of explanatory variables in our panel data and threshold regression model. One is the threshold variable, which is the key variable used to assess the optimal cash holding ratio of a firm and to capture the threshold effect of cash holding on company's performance.

The threshold variable is a variable. When the threshold variable is bigger or smaller than the threshold value (γ), the samples can be divided into two groups, which can be expressed in different slopes β<sup>1</sup> and β2. The explanatory variable is a variable, reflecting its impact on the dependent variable. In the threshold regression model, explanatory variable impacts are not fixed but depend on the threshold value of the threshold variable.

In this study, the measurement of threshold and independent variables through the cash holdings ratio (denoted by CASH) was performed. Following previous studies (Azmat 2014; Martínez-Sola et al. 2013; Nguyen et al. 2016; Opler et al. 1999; Vijayakumaran and Atchyuthan 2017), the cash holdings ratio was calculated as cash and cash equivalents divided by total assets. We used the book value to calculate the cash holdings ratio. The measurement of firms' cash holdings ratio was defined as below:

$$\text{CASH} = \frac{\text{Cash and cash equivalents}}{\text{Total assets}} \tag{2}$$
