liquidity and profitability of businesses (Deloof, 2003). A conservative working capital policy would try to minimize financial distress by maintaining more than sufficient working capital investment, whereas an aggressive working capital policy would try to maximize profitability by retaining only the minimum working capital investment (Kieschnick et al., 2013). Conservative management of inventory would entail reducing the ordering cost, as well as eliminating the risk of stock-outs and cost of forgoing purchase discounts (Corsten & Gruen, 2004). Allowing the customers to pay late might increase sales and improve customer relationship (Summers & Wilson, 2003), although it might also increase the risk of bad debts. However, aggressive inventory management would involve reducing holdings costs and the risk of obsolescence. A longer payable period might help to improve short-term liquidity, but simultaneously, it might also endanger the relationship with suppliers.

Working capital management has remained a problem for many manufacturing firms across the globe. However, the Malaysian economy showed strong signs of recovery in 2010 from the global economic crisis (Watanabe et al., 2013)2013. Because of the increased competition to enhance performance among firms in developing nations, challenges for Malaysian firms are increasing. A market research report by the PwC (2018) on the state of working capital investment of listed Malaysian firms across 14 sectors comprising 424 firms, revealed that up to RM 110 billion cash was tied up in their working capital. This implied that some of this could be released by improving the working capital performance.

Generally, studies such as those by Mohamad and Saad (2010), Ng et al. (2017), Zariyawati et al. (2017), Alarussi and Alhaderi (2018), Al-Mawsheki et al. (2019) and Sim et al. (2019) have investigated the effect of working capital management on the profitability of Malaysian firms. However, this present study differs from the aforementioned studies in several aspects. First, it has divided working capital into two primary groups – operating and non-operating working capital management. Operating working capital includes current asset and current liability accounts that change spontaneously with the firm’s operating scale. These accounts include inventory, receivables, and payables. Non-operating working capital includes current asset and liability accounts that do not change spontaneously with the firm’s operating scale, such as loans receivable, idle equipment, unused land or outdated machinery. These should include all other current asset and liability accounts (Le, 2019). As the two groups of working capital are of different natures, and more importantly, their influence on firm performance might be different, they are measured separately. The measure of non-operating working capital used is other net working capital (ONWC), derived by taking the sum of all non-operating current assets and then deducting from it the sum of all non-operating current liabilities (expressed as a percentage of total assets).

Second, this present study has further conceived using two alternative measures of operating working capital – disaggregate and aggregate

– separately. Three disaggregate measures of operating working capital are used in the present study, each of which was derived using each of the three operating working capital component accounts, i.e., inventory, receivables, and payables (Zariyawati et al., 2017). The measures for these three operating working capital components are inventory conversion period (ICP), average collection period (ACP), and average payment period (APP), respectively. As an alternative, the study has also used the aggregate measure of operating working capital, which is the cash conversion cycle (CCC), and this is measured as the summation of ICP and ACP minus APP (Vural et al., 2012). The use of these disaggregate and aggregate measures of operating working capital has allowed the study to examine which of the two measures could explain firm performance better. Third, the study has also used the following two measures of firm performance, return on asset (ROA) and Tobin’s Q (TQ). In this context, ROA is a short-term accounting-based measure, whereas TQ is a market-based, and arguably, a long-term measure of a firm’s performance (Masadeh et al., 2015; Smirlock et al., 1984).

Although accounting-based data has been useful in capturing firm performance (Giner & Reverte, 2006), investors and economists have been more concerned with market-based performance measure (Ullmann, 1985; Woo et al., 1992). Market-based measure reflect the market (investors’) perception of the expected future performance of the firms (Dubofsky & Varadarajan, 1987; Wisner & Eakins, 1994), and is less influenced by company-specific financial reporting rules and managerial manipulation (McGuire et al., 1988). In an attempt to resolve these issues, various studies have found TQ to be a more accurate measure of firm performance, compared to the use of the accounting-based firm performance measures (Wolfe & Sauaia, 2005). According to Barney (2007), TQ has various advantages over the accounting-based firm performance measures since TQ does not rely only on profits, which are framed by accounting rules. Moreover, TQ, it has been argued, is seen as a long-term firm performance measure (Dwivedi & Jain, 2005) and is based on current information that reflects the present value of future cash flows (Ganguli & Agrawal, 2009; Wahla & Hussain, 2012). A TQ value higher than 1 would mean that the firm has created more value by using invested resources (Li et al., 2004), has better investment opportunities (Lang et al., 1989), and has a higher growth potential (Brainard & Tobin, 1968). While the value lower than 1 indicates that the firm’s market value is less than the book value of the assets, which means that the market value of the firm’s assets, based on investors’ assessment, is less than the investment made (at cost) in the firm’s assets.

Even though the data of accounting return has been criticized as not sufficiently representing genuine economic value, not the least, because of its possibility of being manipulated by managers (Johnson & Kaplan, 1987), it has still been widely accepted and applied in many contexts (Maines & Wahlen, 2006) by various users in decision making (Giner & Reverte, 1999; Truica & Trandafir, 2009). Therefore, this present study has used both the accounting-based and market-based measures of firm performance and has defined them as short-term performance (ROA) and long-term performance (TQ) respectively (Dwivedi & Jain, 2005; Rafindadi & Bello, 2019). The rest of the article has been organized into five sections as follows: Section 2 provides a brief review of previous studies. Section 3 describes the data source and the research methodology. In Section 4, the results of analysis are discussed. Finally, Section 5 provides the conclusion.

capital and firm performance in developed and other developing countries, there have been relatively, few studies carried out in Malaysia. For example, Mohamad and Saad (2010) examined the influence of working capital management on profitability for 172 firms listed on Bursa Malaysia from 2003 to 2007. Their study employed TQ, return on invested capital and ROA as the proxies for firm profitability. Correlation and multiple regression analyses indicated that working capital measures, namely cash conversion cycle, current asset to current liabilities ratio, current asset to total asset ratio, current liabilities to total asset ratio, and total debt to total asset ratio significantly influenced the three measures of profitability.

Zariyawati et al. (2017) investigated the relationships between cash conversion cycle, inventory turnover in days, average collection period, average payment period and firm profitability for 30 large and 70 small firms listed on Bursa Malaysia from 2009 to 2013. ROA was employed as a proxy of firm profitability. The Hausman specification resulted in the adoption of the random effect method, which suggested that large firms could increase their profitability by increasing the average payment period, while small firms could increase their profitability by reducing the average collection period and inventory days. Alarussi and Alhaderi (2018) examined the relationship between working capital and profitability of 120 Malaysian listed firms from 2012 to 2014. They selected return on equity (ROE) and earning per share (EPS) to measure firm profitability. Their study provided evidence that working capital components differently influenced different profitability measures. In the regression results, they found that working capital significantly and positively influenced EPS, but not ROE. Recently, Wong et al. (2019) investigated the impact of working capital on SME’s profitability from 2005 to 2012. Their findings revealed that Malaysian SMEs created value by adopting efficient working capital practices.

Researchers have also employed different measures of working capital and their efforts have provided mixed results. For example, studies by Akbar (2014), Gul et al. (2013), Makori and Jagongo (2013), and Rehman and Anjum (2013), showed that cash conversion, inventory turnover ratio and average payment period significantly influenced firm profitability. In contrast, some studies showed no relationship between the mentioned measures and firm profitability (see Arshad & Gondal, 2013 and Manzoor, 2013). Previous studies such as the work by Kusuma and Bachtiar (2018), Rehman and Anjum (2013) and Wuryani (2015) used the aggregate measure of net working capital ratio that included inventory, accounts receivable, and accounts payables along with its component measures of ICP, ACP and APP in the same model. It is argued here that this practice was inappropriate because then working capital would have been measured multiple times leading to multicollinearity arising from the inter-dependent ‘independent’ variables. Therefore, the present study has avoided this problem by ensuring that no two related variables or measures of working capital are included in the same model. Crucially, the previous studies have not identified these following areas of concern: (1) whether the disaggregate measures of operating working capital are, or the aggregate measure of operating working capital is, more useful in explaining firm performance, (2) whether non-operating working capital would influence firm performance differently than operating working capital, and (3) whether working capital would affect short-term performance (ROA), long-term performance (TQ), or both. This present study has tried to fill in the aforementioned knowledge gaps.

The analysis of the data was carried out in two stages. The first stage explored the descriptive statistics for all variables, and these included their mean, minimum, maximum, standard deviation, skewness, and kurtosis. In the second stage, the relationship between the working capital measures and firm performance was examined using linear regression with the panel-corrected standard errors method.

As the literature review has revealed, there are different measures of working capital and firm performance. The present study has used four empirical models in order to examine whether the disaggregate measures of operating working capital (ICP, ACP, and APP) affect, or the aggregate measure (CCC) of operating working capital affects, a firm’s performance more, both in the short-term (ROA) and long-term (TQ). The regression models are specified as follows:

Although the Generalized Least Squares (GLS) method (Parks, 1967) is considered theoretically superior to the OLS, it is generally only applicable to time series cross-sectional data (Beck & Katz, 1995). A practical and better alternative is the Feasible GLS (FGLS). However, FGLS tends to underestimate the precise variability of the estimator, if the time period is less than the cross-sectional units (Beck & Katz, 1995; Jönsson, 2005). To avoid these problems, the Panel Corrected Standard Errors (PCSE) method, which yields robust covariances has been suggested in the analysis of time-series cross-sectional data (Beck & Katz, 1995). When computing the standard errors and the variance– covariance estimates, PCSE removes the heteroscedastic and cross- sectional dependence problems across panels (Moundigbaye et al., 2018).

Table 8 presents the results of the linear regression using panel corrected standard errors method for all the four models. Results reveal that both the disaggregate and aggregate measures of operating working capital were relevant and affected both the short-term as well as long-term firm performance. Models 1 and 2 show that ICP, APP, and CCC negatively influenced ROA. This means, first of all, the lower the ICP, i.e., the sooner inventory was sold, the higher was the ROA, i.e., the higher the firm performance. This result supported the findings in the studies by Arunkumar and Ramanan (2013) and Makori and Jagongo (2013). However, this result contradicted the findings in studies by Arshad and Gondal (2013), Gul et al. (2013), and Jakpar et al. (2017) which found a positive impact of ICP on firm performance. Second, the lower the APP, i.e., the sooner suppliers were paid, the higher the ROA, i.e., the higher the firm performance.

This result supported the findings of Makarani and Bineshian (2013) and Manzoor (2013). However, this finding contradicted the findings of studies by Gul et al. (2013) and Kusuma and Bachtiar (2018). Third, ACP was not significant, indicating that the collection of receivables did not appear to influence short-term firm performance. This result is in contrast to the findings in Jakpar et al. (2017). Fourth, the lower the CCC, i.e., the shorter the period between a firm paying its suppliers and collecting from its customers, the higher the firm performance. This result supported the findings in Mohamad and Saad (2010), but contradicted the findings in Jakpar et al. (2017). Fifth, about non- operating working capital, the higher the ONWC, i.e., the greater a firm’s proportion of other (excluding inventory, receivables, and payables) net working capital over total assets, the higher the firm’s performance. There have been no comparable findings from previous studies regarding this issue.

Models 3 and 4 show that all the three disaggregate measures (ICP, ACP, and APP) and the aggregate measure (CCC) of operating working capital were significant in influencing TQ. The coefficients of ICP, ACP, APP, and CCC were all negative, implying that the quicker conversion of inventory to sales, prompter collection of receivables, faster payment of payables, and the shorter the period between paying suppliers and collecting from customers, each of these resulted in higher long-term firm performance. The findings of the negative influence of ICP and ACP on firm performance were supported by the findings in Bhatia and Srivastava (2016). However, these findings contradicted the findings of the study by Nurein et al. (2015). The finding of the negative influence of APP on firm performance was at odds with the studies by Bhatia and Srivastava (2016) and Nurein et al. (2015), who found a positive impact on firm performance (TQ). The negative influence of CCC on firm performance was supported in studies by Bhatia and Srivastava (2016), Nurein et al. (2015) and Singh et al. (2017). However, contradictory findings were obtained in a study by Vural et al. (2012), which found a positive and significant impact on firm performance (TQ).

As with the previous models, ONWC also positively influenced TQ in Model 3 and Model 4. This variable has not been used widely in previous studies. The intent in the present study was to model the ONWC as a separate variable, and not consolidate it with the other working capital accounts. This was done to see if non-operating working capital would have a different influence on firm performance, compared to operating working capital. Interestingly, ONWC, the measure of non-operating working capital, positively influenced firm performance, whereas measures of operating working capital negatively influenced firm performance.

Although less inventory, receivables (only for long-term firm performance), and even payables seemed to increase firm performance, in contrast, more other working capital accounts appeared to increase firm performance. This also holds an important implication for research into the influence of the working capital on firm performance – all working capital accounts should not be consolidated into an aggregate sum, e.g., net working capital expressed as a percentage of total assets, as this will lump all the operating working capital accounts (inventory, receivables, and payables) with the non-operating working capital accounts. Had this been done, the nuanced influences of different working capital components on firm performance found here would have remained uncovered. Less operating working capital (inventory, receivables, and payables), but more non-operating working capital, seemed to increase firm performance.

Additional results show the impact of two control variables, namely firm size and debt ratio, on firm performance. Firm size has been found to be positive and significant in all four models, while debt ratio was positive and significant in the third and fourth models. These findings seemed to imply that firm size influenced both the short-term

(ROA) and long-term firm performance (TQ). In other words, larger firms showed better performance, whereas debt ratio only influenced long-term firm performance (TQ). This meant that higher financial leverage (use of debt) increased long-term firm performance.

Another notable finding was the slight difference in the adjusted R-square among the four models, in particular, between Model 1 and Model 2, and also between Model 3 and Model 4. The adjusted R-square was greater for Model 1 (than Model 2), and Model 3 (than Model 4), indicating that the use of the three disaggregate measures of operating working capital was slightly better than the use of the single aggregate measure of operating working capital in explaining firm performance, both for the short-term and long-term.