Investigating the Nonlinear Relationship between Working Capital and Profi tability: a Case of Pakistan Textile Firms Raziskava nelinearnega razmerja med obratnim kapitalom in donosnostjo: primer pakistanskih tekstilnih podjetij

This study analyzes the impact of working capital (WCR) on operating profi t of Pakistan textile fi rms from 2009 to 2016. The nonlinear relationship has been found between working capital and operating profi t, which indicates that an optimal level of working capital exists in the textile fi rms of Pakistan. The fi rms are seeking the optimal working capital, where WCR (4.78%) of sales in generalized method of movement is used. Further, the study reveals that in the fi rms which maintain the positive working capital, it has a significant negative infl uence on the profi tability, while in the fi rms with negative working capital, it has a significant positive infl uence on their profi tability. The study also ascertains that cash holding level is an important factor for effi cient working capital management.


Introduction
Pakistan's economy is the 24 th largest economy in the world in terms of purchase power parity and the 42 nd largest economy in terms of gross domestic product (GDP). Since 2014 Pakistan's economy has sustained the 4.5% growth rate and ranked among the top growing economies of South Asia. Pakistan's textile sector is the second largest sector of the economy aft er agricultural. Pakistan is the 4 th largest textile producer in Asia and fulfi lls 8% of global textile demand products. It is the 8 th largest textile exporter in Asia, and has ranked among the top 10 textiles producing countries in the world. Th e textile sector in Pakistan has maintained 12 billion US dollars investment in the period 2011 to 2012. It contributes Ashfaq Habib 1 , Xiaoxia Huang 2 1 Donlinks School of Economics and Management, University of Science and Technology Beijing 2 Donglinks School of Economics and Management, University of Science and Technology Beijing approximately 9.5% to the country GDP and obtains 40% of total banking credits for the industrial sector. It also provides employment to an average 45% of total manufacturing labor force and maintains 54% share in the national exports [1][2][3]. Th e textile industry in Pakistan has also been aff ected by the global fi nancial crisis 2008. Th e average growth rate 8.22% was found from 2001 to 2007 and -0.70% in the period 2008 to 09. Th e industry maintained a sluggish growth rate of an average 0.59% from 2010 to 2016. Th e Pakistan's textile share in international export decreased from 13.5 billion to 12.5 billion US$ in the years 2013 to 2015 [4]. Th e textile industry in Pakistan faces the number of challenges including the high-energy crisis, intensifying oil prices, lack of R & D in fi ber & cotton and the dearth of modern technology in the textile sector. Th e industry has also been aff ected by lower demand for its products in EU and US markets in the last four years. Further, the production cost has increased by the depreciation of the local currency in terms of US dollar and the increased rate of infl ation in the economy. Moreover, the political and economic uncertainty in the country has also negatively aff ected the textile industry in Pakistan [5,6]. Th is study focuses on examining the operational performance of the textile industry in Pakistan. Th e working capital (WCR) can be used to measure the operational performance of business [7]. Pakistan's textile industry holds a lower level of working capital 2.33% (even negative) of gross sales and the large amount of trade credits 5.12% of sales in our selected sample. Th e industry also maintains a large stock of inventory 7.34% of sales in our study period. Conversely, the textile sector in Pakistan maintains a higher percentage of accounts receivable 3.34% of sales. Overall, it implies that the textile industry in Pakistan has tried to use working capital to run the day-to-day operations of the business. In such situation, working capital is important for textile fi rms to survive in weak fi nancial conditions and augmented economics uncertainty. Similarly, the liquidity management (cash and cash equivalent) cannot be ignored for textile fi rms in good times and even so in unfavorable economic circumstances. Shortage of liquidity can cause bankruptcy and excess liquidity harmful to a fi rm's profi tability [8]. Working capital includes current assets and current liabilities. Th e current assets are cash, inventories, accounts receivables and marketable securities [7].
Th ese assets are used to make the payment of shortterm liabilities. Short-term liabilities are trade debts, short-term debts, and outstanding liabilities [9]. A higher working capital provides an opportunity to increase sales by maintaining a large stock of inventories to control the input price fl uctuations and distributions. It also facilitates to extend trade credits and avails discount to make early payments. Moreover, with higher working capital, a fi rm can adopt the liberal trade credit policies about potential customers to boost up the sales revenue. But it requires the additional fi nancing which increases the interest expenses and bankruptcy risk [10]. Further, the excess cash balances in working capital restrict fi rms to invest in the value enhancing projects and increase the opportunity cost of the business [11]. On the other hand, a lower level of working capital demands to hold a small amount of cash, to maintain a lower level of inventories, and to adopt the restrictive trade credit policy. It helps to reduce the interest expenses and the opportunity cost of debt. But it can also be caused by the liquidity crisis, shortage of inventories and reduction in sales revenue. Firms may also fail to make payment in time and increase the bankruptcy cost of the business [12]. According to Mun et al. [13], a nonlinear U-shape relationship exists between working capital and profi tability. Th e managers are required to trade-off among the working capital components to maximize the fi rm's profi tability. Th e trade-off among the current assets and current liabilities may lead to the effi cient working capital management. Th e effi cient working capital management requires that a fi rm can pay the short-term fi nancial obligations and avoids over-investment in the current assets [14]. Hence, a fi rm can increase the profi t by effi cient monitoring of cash, inventories, accounts receivable and accounts payable. Cash is the most liquid element of working capital. It directly infl uences the operational ability of a fi rm to meet its short-term fi nancial claims [15]. Keynes [16] explains that a fi rm can maintain liquidity in the form of cash to meet the operational expenses which exceed its regular income. Th e fi rms also maintain cash as precautionary motives to prevent themselves against the risk of liquidity shortfall. However, it is not always benefi cial to hold too much cash. Cash is the least profi table asset, and investment in cash is costly [17]. Th e managers may invest in less liquid assets to increase the profi tability of a fi rm. Myers [18] proposes a cash trade-off theory that equalizes cost and benefi ts of cash holding. Hence, the optimal level of cash can be obtained by trade-off among diff erent components of the working capital. Th erefore, the aim of this study is to analyze the Ushape relationship between working capital and operating profi t of the textile fi rms in Pakistan. Th is research also focuses on determining the optimal level of working capital that performs an important role to design the effi cient level of working capital. It supports to overcome the limitations and diffi culties faced by the textile industry in Pakistan to manage the business operations smoothly. Finally, we also examine the cash holding level that performs a pivotal role in the relationship between profi tability and working capital.

Working capital
Working capital is used to measure the short-term operational performance of the business. Th e operational risk and returns are generated from the procurement of inventories from vendors and collection of accounts receivable from customers. Cash conversion cycle (CCC) has been accepted to measure the risk and rewards associated with operations of business [19]. CCC refl ects the length of time a fi rm needs to convert inventories, accounts receivable, and accounts payable into cash. Traditionally, CCC has been used as a proxy of working capital to investigate the infl uence of working capital on operating profi t [14,[20][21][22][23]. In fact, CCC refl ects the operational side (inventories, accounts receivable, and accounts payable) of a fi rm's working capital. In this manner, previous research studies used CCC to measure the operational performance of the business. Th e relationship between CCC and the fi rm's profi tability was analyzed in seven industries from 1974 to 1993 [24]. Th e results suggested that a lower level of CCC was associated with higher profi tability in several industries. Th e negative association between CCC and operating profi t was also found in Saudi Arabia's listed fi rms [20]. Similarly, the CCC also developed the negative association with operating profi t in Japanese fi rms 1999-2004 [22]. Th e negative association of CCC with operating profi t indicates that fi rms are maintaining an effi cient level of working capital to run the routine operation of the business. At lower level of CCC a fi rm is rotating inventory more quickly, maintaining a higher percentage of accounts receivable and delaying payments to vendors as much as possible. Th e eff ect of working capital on operating profi t was investigated in small and medium-size fi rms of Finland [9]. Th e results are consistent with the prior study of Eljelly and Abuzar [20] in terms of association between operating profi t and working capital. Likewise, the negative relationship has been found between CCC and profi tability in Finnish listed fi rms in diff erent business cycles [11]. Here, CCC indicates that Finish fi rms are maintaining a higher rate of accounts receivable and managing inventories more effi ciently to run the business operations smoothly, while insignifi cant relationship was found between working capital and profi tability in Belgian listed fi rms 1992-1996 [19]. However, a lower level of CCC may also deteriorate the fi rm's profi tability [25]. If the level of inventory is reduced too much, it can cause the interruption in the manufacturing process, can increase the carrying cost of raw material, and cannot avail the adequate amount of discount from suppliers. Wang and Yung [26] identify that if a fi rm maintains a lower level of inventory, the sales may also be reduced due to the shortage of product in the market. Further, the restricted credit policies can destroy the relationship with potential customers. Similarly, paying trade credits aft er an extended period of time and demanding more credits from suppliers may damage the fi rm's reputation [27]. According to Baos-Caballero et al. [12], the inverse relationship exists between CCC and profi tability. Th ey argued that a fi rm should increase the volume of working capital by increasing the investment in inventories and accounts receivable to boost up sales when CCC is too short. However, the eff ect of CCC is turned negative on a fi rm's performance aft er a certain point when CCC is too large. Hence, the managers have to fi nd an optimal level of inventories, accounts receivable, and accounts payable to maximize the fi rm's profi tability. Aktas et al. [11] also suggest that a nonlinear relationship exists between working capital and profi tability in American manufacturing fi rms. Th ey argue that US fi rms can enhance the operational performance by increasing investment in under-invested working capital or by reducing the over-invested working capital. Similarly, some other studies have also found the nonlinear relationship between working capital and profi tability [13,28]. Th eir fi ndings are consistent with the studies of [11] and [12]. Aft er a detailed literature review, it has been observed that these studies only capture the operational side of working capital and ignore the fi nancial side (cash, marketable securities, outstanding liabilities and short-term debt) which plays an essential role in determining the relationship between working capital and profi tability. It has also been found that the eff ect of working capital on profi tability has changed due to the change in the characteristics and environment of each industry. Th e managers are required to trade-off among the diff erent components of working capital to enhance the fi rm's profi tability. Th is trade-off in working capital may create a nonlinear relationship between profi tability and working capital. So, overall, it has been found that working capital may develop a nonlinear U-shape relationship with profi tability. Th erefore, we propose the hypotheses: Hypothesis 1: • Th e nonlinear relationship exists between working capital and operating profi t in the textile fi rms in Pakistan. Hypothesis 2: • If the textile fi rms maintain the positive working capital, the negative relationship will be developed between working capital and operating profi t. Hypothesis 3: • If the textile fi rms maintain the negative working capital, the positive relationship will be developed between working capital and operating profi t.

Cash holding
Th e primary motives of cash holding are transaction motive, precautionary motive, tax motive and agency motive. Th e objective of transaction motive is to reduce the liquidating cost of assets instead of holding cash at the time of emergency [29]. In precautionary motive, companies maintain suffi cient amount of cash than their desired level to meet the unexpected situations in future. Th e large tax paying companies manage cash for payment of tax obligations [30]. In agency motive, managers prefer to hold cash rather than distribute dividends among equity holders [31]. Working capital should be maintained with consistent profi tability and sound liquidity for quick customer's response [32]. Managers need to fi nd an optimal level of liquid assets to run operations smoothly. Th e smooth operational strategy in working capital and capital structure characterize the level of cash holding [33]. Corporate liquidity is considered an economical source of fi nancing to fi nance the growth opportunity as compared to generate the funds from the external market. Th e internal source of fi nancing is reduced by the fi nancial distress cost. But the free cash fl ows may increase the agency cost of shareholders. Th e fi rms can trade-off between the cost and benefi ts of internal fi nancing to determine the optimal cash holding level [12]. Th e value of cash holding was analyzed in Chinese listed fi rms from 1993 to 2007. Th ey argued that a fi rm could earn the higher profi t and growth rate by holding a large amount of cash. Th e cash holding level also developed the inverse relationship with networking capital [34]. Th e fi rms that maintain the positive working capital and positive cash holding are expecting a higher uncertainty in future and poor ability to generate cash fl ows from operations. Th e positive working capital and positive cash holding increase the fi rm's opportunity costs, which negatively infl uences the fi rm's profi tability [12]. On the other hand, if a fi rm has the capacity to generate effi cient cash fl ows from the operation, it increases the non-cash assets in working capital. It reduces the opportunity cost of cash holding, which positively infl uences the fi rm's profi tability [13]. Th erefore, the relationship between working capital and profi tability is aff ected by the level of cash holding. Hence, the cash performs a moderating role between profi tability and working capital.
Hypothesis 4: • If the positive working capital exists in the textile fi rms in Pakistan, the negative relationship between operating profi t and working capital will be infl uenced by the level of cash holding. Hypothesis 5: • If the negative working capital exists in the textile fi rms in Pakistan, the positive relationship between operating profi t and working capital will be infl uenced by the level of cash holding.

Sample
Th is study uses the fi nancial data of the Pakistan listed textile fi rms from 2009 to 2016. Th e data has been extracted from the balance sheet analysis and published fi nancial statements of textile fi rms compiled

Variables
Th is study uses the Return on assets (ROA) as the dependent variable for a proxy of a fi rm's operating profi t. Th e working capital and its components (cash, inventory, accounts receivable and accounts payable) are used as independent variables. To fi nd a unique relationship between profi tability and working capital, the whole sample is divided into positive and negative working capital groups. Th e reason behind this is to examine the traditional view of working capital that a higher level of working capital enhances the value of the fi rm, while an opposite view maintains that a higher level of working capital negatively infl uences the fi rm's performance. Hence, it is not suitable to examine the whole sample mutually without considering the level of working capital.
Working capital consists of cash assets (cash + cash equivalent -short-term debts) and non-cash assets (inventory + accounts receivable -accounts payable). Th e cash holding rate of cash assets (CHR = cash + cash equivalent -short-term debts/sales) is used as a measure of cash holding level. While, the cash conversion rate of non-cash assets (CCR = inventory + accounts receivable -accounts payable/ sales) is used as a measure of cash conversion cycle. Th e interaction term (CHR*WCR) is created to fi nd the moderating role of cash. Th e dummy variable is created 1 for positive CHR and 0 for negative CHR. Th e fi rm's size, leverage, growth, and gross domestic product are also used as control variables in this research. Table 1 shows the variables interpretation, and their acronyms used in this study. Th e redundant variable test is applied to check the reliability of explanatory variables in the regression equations. All variables are found statistically significant during the redundant test. Similarly, the multicollinearity is checked through tolerance (Toler) and variance infl ationary factor (VIF). Th e value of Toler and VIF for working capital rate (WCR = 0.58 -Toler, 1.12 -VIF), fi rm size (FS = 0.52 -Toler, 1.07 -VIF), sales growth (Growth = 0.96 -Toler, 1.02 -VIF), leverage (LEV = 0.81 -Toler, 1.12 -VIF), and gross domestic product (GDP = 0.81 -Toler, 1.10 -VIF) are greater than 0.5 and less than 2 respectively. Th e value of Toler and VIF shows

Regression analysis
Th e study uses the panel data regression statistics to examine the operational performance of the textile fi rms in Pakistan. First, the panel ordinary least squares (OLS) regression method is applied to Equation 1 to explore the nonlinear relationship. Further, the panel fi xed eff ect (FE) and panel generalized method of movement (GMM) are also applied to assure the results are robust. Th e FE method is used to remove serially correlated errors in Equation 2. Th e Hausman test is applied to choose fi xed or random eff ect method. Th e Cross section random chi-square statistics (7.86) under Hausman test eff ect is significant, therefore the fi xed eff ect method is used instead of the random eff ect method. Further, the GMM is applied to control the endogeneity problem in Equation 3. Th e endogeneity problem occurs when one or more explanatory variables are correlated with errors term (μ i,t ). It can be the result of omitted variables or measurement errors. Th e proper instrument variables are used to deal with the endogeneity problem that correlate with indogenous independent variables but do not correlate with errors. Arellano and Bond (1991) argue that lag-regression and lag-diff erence of all independent variables can be used as instrument variables to deal the endogeneity problem. For the fi nal analysis, the following methods are used to investigate the relationship between ROA and WCR.

2-panel fi xed eff ect regression statistics (FE)
3-panel generalized method of movement regression statistics (GMM) where ROA i, t-1 is an instrument used to remove endogeneity. Standard: WCR i,t , FS i,t , Growth i,t , LEV i,t , GDP i,t 4 Results and discussion  indicates that positive CHR fi rms hold more liquid assets than negative CHR fi rms and face a lower bankruptcy risk. Similarly, CCR (-0.05) and (-0.01) in both groups are signifi cantly diff erent to each other. Th e positive CHR group has earned higher ROA (0.08), WCR (0.31), Growth (0.10) and LEV (0.32), but lower CCR (-0.05) than the negative CHR group in Table 3.

Descriptive statistics
In the negative WCR group, sub-group negative CHR    Table 4 shows the results of OLS, FE, and GMM regression techniques to fi nd the nonlinear relationship between working capital and profi tability. In the whole sample, WCR develops a signifi cant positive relationship with ROA in OLS (0.06), FE (0.04), and GMM (0.03) by using equations 1-3. Th e WCR 2 term is used to fi nd the nonlinear relationship between working capital and profi tability. Th e WCR 2 develops the negative association with ROA in OLS (-0.03), FE (-0.03), and GMM (-0.01) in the whole sample. It shows that working capital develops a nonlinear Ushape relationship with operating profi t in Pakistan's textile industry as proposed in our fi rst hypothesis (1). Using the partial derivation in terms of WCR, we have calculated the optimal level of working capital to enhance the fi rm's profi tability. Th e textile fi rms can reach an optimal level of WCR in terms of sales in OLS (8.5%), FE (6.23%) and GMM (4.78%).

Analyses results
In the whole sample, the control variables leverage Th e fi ndings are consistent with the previous studies of [12,13] in the analysis of USA and UK fi rms respectively. Further, the whole working capital is divided into positive working capital group and negative working capital group to fi nd the unique pattern of association between working capital and profi tability. In Table 5, in the positive WCR group, WCR develops a signifi cant negative relationship with ROA in OLS (-0.03), FE (-0.02), and GMM (-0.01) as proposed in hypothesis 2. However, in the negative WCR group, WCR is signifi cant positively associated with ROA in OLS (0.05), FE (0.03), and GMM (0.02) as proposed in hypothesis 3. Th e results indicate that the magnitude of coeffi cients of negative WCR under the negative WCR group is higher than that of the coeffi cients of positive WCR under the positive WCR group. It implies that negative working capital is more effi ciently responsible for increase of the profi tability than the positive WCR.
Based on the results of Arellano-Bond serial correlation test in GMM as shown in Table 5, we verifi ed that the instruments used to estimate the GMM were valid and correctly specifi ed. Hence, we cannot reject the null hypothesis that the endogeneity problem does not exist at the 2nd order for any of our regression methods. Th us, the instruments variables do not correlate with the errors in GMM and give the ro-bust results about the endogeneity problem, which is very diffi cult to control in OLS and FE. Th erefore, the results presented under GMM in Table 5 are more reliable and robust than the OLS and FE.
To check whether the cash is playing a moderating role in the relationship between profi tability and  working capital, we have developed the interaction term of (WCR*CHR) in OLS, FE, and GMM as presented in Table 6. Th e CHR is a dummy variable that is 1 for positive CHR and 0 for negative CHR.
In the positive WCR group, the interaction term (WCR*CHR) has created the signifi cant negative relationship with ROA in OLS (-0.05), FE (-0.04), and GMM (-0.02). Th e results show that if a fi rm has positive working capital and increases the cash level, it will enhance the negative relationship between working capital and profi tability as explained in hypothesis 4. On the other hand, in the negative working capital group, the interaction term (WCR*CHR) has created the signifi cant positive relationship with ROA in OLS (0.03), FE, (0.03), and GMM (0.02) as explained in hypothesis 5. It indicates that if a fi rm has negative working capital and increases the cash level, it will enhance the positive relationship between working capital and profi tability. Overall results show that cash plays a moderating role to determine the relationship between working capital and profi tability.

Conclusion
Th is study analyzes how working capital aff ects the operating profi t of textile fi rms in Pakistan. Th e research has revealed that a signifi cant nonlinear relationship exists between working capital and operating profi t (Hypothesis 1). Th e study has also found that a positive working capital has negatively infl uenced the operating profi t (Hypothesis 2). However, the negative working capital has infl uenced positively (Hypothesis 3). Th ese fi ndings are similar to the studies of Mun et al. [13] and Baos-Caballero et al. [12], where they suggest that the optimal level of working capital depends upon the fi rm's performance. Th is study also determines an optimal level of working capital 4.78% of sales in GMM model. Th e GMM estimator permitted us to control the non-serial correlated issues and to fi nd the most robust results. Further, the moderating role of cash has been examined in the relationship between working capital and operating profi t. In the positive WCR group, the interaction term (WCR*CHR) has created the signifi cant negative relationship with profi tability (Hypothesis 4), while in the negative working capital group, the interaction term (WCR*CHR) has developed the signifi cant positive association with profi tability (Hypothesis 4) It implies that cash plays a moderating role in the relationship between working capital and profi tability. Hence the cash holding level is an important factor for designing the efficient working capital management.

Research implication
Th is study suggests some theoretical and managerial implications for the textile industry in Pakistan. Th eoretically, the study suggests that a nonlinear relationship exists between working capital and profitability in textile fi rms. Th e study has also found that cash plays an important role to determine the relationship between working capital and profi tability. Th e results show that cash holding level can be used as a measure of the fi rm's internal cash-generating capability. Th e study has found the signifi cant negative interactive eff ect of cash in positive working capital and signifi cant positive interactive eff ect of cash in negative working capital fi rms. It indicates a fi rm can design and maintain an effi cient level of working capital by trade-off cash assets with non-cash assets.
In respect of managerial implication, this research suggests that when the textile fi rms have the positive working capital, it is better to ease the cash conversion cycle either by paying suppliers earlier or by increasing investment in inventories. Further, when a fi rm is effi ciently generating the cash fl ows from operations, it does not need to hold a large amount of cash as long as it needs to maintain suffi cient amount of inventories and accounts receivable. On the other hand, when a fi rm does not have a sufficient amount of working capital, it needs to secure assets immediately by decreasing expenses, deferring the investment, or acquiring loans to avoid the negative eff ect of liquidity shortfall. Such types of textile fi rms in Pakistan should increase the volume of working capital until it reaches 4.78% of sales.