“A reassessment of the relationship between working capital management and firm performance: evidence from non-financial companies in Nigeria”

This paper reassesses the relationship between working capital management (WCM) and firm performance in the Nigerian context. The study is motivated by the limited insights available on the impacts of WCM on firm performance in the country. To date, most studies from Nigeria have been largely descriptive and focused on a small sample size that is non-representative of the population. In addition, there are limited rigorous statistical analyses involved in such studies. This paper addresses the methodological limitations apparent in prior literature and provides a better understanding of the relationship between WCM and firm performance, revealing how firms can manage their operations more profitably. The paper adopts a panel data regression analysis on a sample of 75 non-financial firms listed on the Nigerian Stock Exchange from 2007 to 2015. The results of the analyses showed that WCM variables have an inconsistent rela- tionship with the measures of performance adopted, which were return on assets and Tobin’s Q. Specifically, accounts receivable management and inventory management were negatively associated with the return on assets, while accounts payable management, cash conversion cycle and cash conversion efficiency were positively associated with return on assets. Additionally, accounts receivable management and inventory management were positively associated with Tobin’s Q, whereas accounts payable man- agement, cash conversion cycle and cash conversion efficiency were negatively associated with Tobin’s Q. These results were found to be robust using quantile regression. The results of the quantile regression showed inconsistency across the various quantiles used (0.10, 0.25, 0.50 and 0.75). These findings have two important implications. The first is that WCM variables influence the performance of firms. The second is that the mixed findings partly indicate that firms and managers must understand and for- mulate WCM policies that reflect their peculiar conditions.


INTRODUCTION
In a world of resource scarcity and limited access to operating capital, firm performance has become a critical issue. Working Capital Management (WCM) makes a fundamental contribution to the performance of firms by providing adequate liquidity in the form of free cash flow to finance the operational activities of firms (Deloof, 2003;Eljelly, 2004) and enhances shareholders' wealth (Deloof, 2003;Filbeck & Krueger, 2005; Afrifa & Padachi, 2016). The importance of WCM is evidenced by the considerable amount of empirical research conducted on the relationship between WCM and firm performance (e.g., Deloof, 2003 Osundina, 2014;Aileman & Folashade, 2014) demonstrates that WCM is important and key to providing management and operational efficiency that can improve the liquidity and profitability of firms. However, these studies are constrained by methodological limitations, hence, undermining the implications of their findings on the relationship between WCM and firm performance in Nigeria. Some of these methodological limitations include an inadequate sample size, a poor sample selection procedure and the appropriateness of the statistical analysis.
The methodological shortcomings of the extant studies on WCM and firm performance from Nigeria motivated this empirical study to provide a more robust understanding of the relationship between WCM and firm performance in Nigeria. Also, motivation was drawn from the fact that Nigeria as a country has displayed high vulnerability to investable and operating capital due to the impact and exposure to the worldwide financial crisis of 2007-2008. Consequently, low financial development characterizes Nigeria. This is typified in the declining and inactive capital market operations and currency depreciation that has created instability in exchange rates that have negatively affected the economy (Akinlo, 2012). Subsequently, most firms in Nigeria have faced a myriad of challenges ranging from a scarcity of foreign exchange, to infrastructure deficits, to high banking charges and to lack of raw materials. Using a sample of 75 firms over the period of 2007-2015 and adopting the panel data regression analysis that corrects for potential unobserved variables that may be correlated with the variables, the results of this study offer insights into the practice of WCM for managers and firms for improving their cash flow and performance. Therefore, the paper contributes to WCM literature in many ways. First, the paper contributes to WCM literature by providing evidence from a large sample size. Second, the study used two alternative measures of firm performance, i.e. accounting (ROA) and market measures (Tobin's Q). A third contribution is that the paper demonstrates the importance of cash conversion efficiency as an important variable in explaining WCM. Finally, this paper advanced the quantitative technique by employing the panel data and quantile regression to determine the relationship between WCM and firm performance. Thus, the findings from such methods will also help managers improve the quality of their financing decisions to enhance their financial and management performance.
The rest of this paper is organized as follows. Section 1 discusses the literature review on the methodological weakness of previous WCM studies in Nigeria and the development of the proposed hypotheses. Section 2 presents the methodological approach adopted in this study. In Section 3, the results of analysis are presented with a general discussion of the findings of this study. The last Section concludes the paper.

Methodological weakness of WCM studies in Nigeria
WCM is essential to the success of all business sectors; however, the majority of the studies conducted in Nigeria are faced with problems resulting from an inadequate sample size and a short time period and are concentrated in the manufacturing sector. For example, Festus (2012) focused on determining how WCM could be used to resolve profitability and distress issues arising from ebusiness organizations in Nigeria using a sample of five non-financial firms from 2005 to 2007. The study found that WCM significantly influenced the success of the businesses. Barine (2012) studied WCM and the profitability of 22 banks and non-banking firms for the 2010 financial year. Findings showed that firms in Nigeria rely heavily on external financing thereby making them vulnerable to any financial crunch that comes along. The study encouraged an adequate WCM policy because such is critical in enhancing free cash flow for a firm. Takon and Atseye (2015) evaluated the effects of WCM on the profitability of 46 firms in Nigeria. The study found a significant relationship between measures of WCM and the return on assets of businesses. They concluded that the high cost of acquiring funds and the unstable economic conditions have a great negative influence on firm performance in Nigeria. Meanwhile, Kurawa and Garba (2014)

Hypotheses development
Considerable research exists on the relationship between WCM and firm performance from developed and other developing countries (e.g., Abuzayed, 2012). Deloof (2003) used 1,009 Belgian non-financial firms between 1992 and 1996 to determine the relationship between WCM and corporate profitability. Using correlation and regression analysis, he found a negative and significant relationship between the gross operating income of Belgian firms and working capital measures.
The results also concluded that the manner in which working capital is managed will determine its impacts on firm profitability. Therefore, managers could bring additional value to a firm and its shareholders by appropriately managing the working capital components. Eljelly (2004) investigated the liquidity and profitability trade-off of companies in Saudi Arabia between 1996 and 2000. Using correlation and regression analysis, the study revealed a significant and negative relationship between profitability and liquidity measured by current ratio. The study gives credence to the notion that cash conversion and cash gap are more appropriate measures of liquidity than is current ratio. Firm size was found to be a significant factor, and analysis revealed that firms with a higher current ratio and a longer cash conversion cycle exhibit a higher negative relationship with profitability. Size is important to a firm and brings with it several benefits. In line with the findings of Padachi (2006) found the same results using a sample of 58 small manufacturing firms from 1998 to 2003 in Mauritius to determine trends in WCM and its influence on firm performance. Padachi (2006) found a significant relationship between WCM (cash conversion cycle, inventory, receivables and payable) and profitability measured by return on assets. Shah and Sana (2006) found a negative relationship between gross operating income in the Pakistani oil-andgas sector and the inventory period, sales growth, accounts receivable and the cash conversion period. Accounts payable had a positive relationship; however, the negative relationship found between sales and profitability might be associated with the sensitivity or peculiarities of the sector studied. Meanwhile, Mathuva (2010) studied the influence of components of WCM on the profitability of 30 firms listed on Nairobi Stock Exchange for the period 1993 to 2008. He found a significant and negative relationship between the account collection period, cash conversion and firm profitability. The results also revealed a significant and positive relationship of the inventory period and the payment period on profitability. Falope  H1a-e: There is a significant relationship between the WCM variables of accounts receivable management, accounts payable management, inventory management, cash conversion cycle, cash conversion efficiency and the ROA of firms in Nigeria.
H2a-e: There is a significant relationship between the WCM variables of accounts receivable management, accounts payable management, inventory management, cash conversion cycle, cash conversion efficiency and the Tobin's Q of firms in Nigeria.

Variable measurements
The variables used in this study are described in Table 2   Replaced data using averaging method - 14 21 Useful data (total sample) 75 dasticity and auto/serial correlation and guarantee that the results of this study are free from any estimation bias, the VCE robust and cluster approach was adopted in both models as Baum (2006) suggested. The models in this study were all estimated using the STATA 13 statistical software.

Descriptive statistics and correlation
Descriptive statistics for the variables in their natural metric are presented in Table 3, while the transformed variables are presented in Table   4 to facilitate interpretation and understanding. Several items are of note. First, substantial variance existed between the accounting and market measures of performance adopted. Second, the descriptive statistics are consistent with other WCM studies (e.g., Mathuva, 2010). Third, the data for this study were normally distributed as the skewness and kurtosis ranged from -0.06 to 1.8 and 1.7 to 9.4, respectively. This shows that the data were within the expected range for a normal data. This is because the skewness and kurtosis fell below the threshold value of +/-3 and +/-10, respectively, as Kline (2011) suggested. To bring the data to a closer range, ARM and APM were logged.
Their new values and effect are shown in Table 4. Thus, the subsequent analysis will be based on the    Notes: Variables were winsorized at 3% to mitigate the effect of outliers in this study, while *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. logged data. The transformed data in Table 4 show that ARM now had a mean value of 3.49, while the APM mean was now 3.63. Their skewness and kurtosis are -0.31 and -0.53 and 2.91 and 3.00, respectively. Table 5 presents the correlation, and no correlation coefficients between a pair of variables in this study exceeded the threshold of 0.80 that Field (2005) suggested to indicate a problem of multicollinearity. Thus, the conclusion can be made that the choice of these variables would not result in misspecification; this was also confirmed by the results of the variance inflation factor (VIF), which show a value 1.7 (though not tabulated), but less than the threshold of 10 to suggest no serious problem of multicollinearity according to Field (2005). Table 6 presents the results estimating the relationship between WCM variables and firm performance. The results are presented separately, wherein dependent variables proxied by ROA and TQ are reported in column 1 and column 2, respectively. The results were obtained using the panel data regression (fixed effect model) with the VCE robust and cluster estimate to control for heteroskedasticity and auto/serial correlation. The results presented in Table 6 show that ARM was negative but insignificantly associated with ROA (β = -0.0014091, p > 0.10). The negative relationship between ARM and ROA implies that shorter account receivable periods were associated with ROA. Thus, a decrease in the ARM periods by one percent increases ROA by 0.0014091. It can be inferred from this result that early collection of debt from customers increases the performance through the supply of cash flow that meets both the operational and financing activities of firms. Contrary to Hypothesis 1a, the negative relationship between ARM and ROA was not significant and, therefore, does not support the hypothesis. With regards to TQ, the results showed that a positive and significant relationship exists between ARM and ROE (β = 0.1514978, p < 0.10), indicating support for Hypothesis 2a. The positive relationship means that an increase in ARM will lead to an increased TQ of firms. This suggests that a percentage increase in ARM was associated with a 0.1514978 increase in TQ. The result with respect to ROA supports the assumption of WCM, which states that a shorter account collection period is beneficial but provides no statistical evidence to support the results found as the relationship was statistically insignificant. However, this finding was consistent with Deloof (2003), and Lazaridis and Tryfornidis (2006), whereas the later relationship (ARM and TQ) contradicts the previous studies undertaken.

Regression analysis results
In Hypotheses 1b and 2b, a significant relationship was predicted between APM and firm performance precisely measured by ROA and TQ. The results presented in Table 6 (model 1) reveals support for Hypothesis 1b, as the result indicates that APM was positive and significantly related to ROA (β = 0.0080847, p < 0.10). This implies that extending payment periods to suppliers was associated with a higher ROA. Thus, a percentage increase in APM increases ROA by 0.0080847. This result reveals that delaying a payment gives firms the opportunity to overcome financing constraints by using cash that would have been paid to suppliers for operational activities. This result is consistent with the findings of Mathuva (2010), Azam and Haider (2011), which emphasize extending payment periods enables firms to take absolute advantage of such cash. In model 2, APM was found to be negative and insignificantly associated with TQ (β = -0.0602374, p > 0.10), implying that early payments to suppliers have advantages that lead to increased performances. Therefore, a percentage decrease in APM will lead to an increase in TQ by 0.0602374. This result is similar to the findings of Deloof (2003). Deloof argued that only unprofitable firms wait longer to pay debts, whereas profitable firms pay early and enjoy discounts and many other benefits. However, the result is not substantively supported, as Hypothesis 2b is not supported. Like the results of ARM, mixed support was present for the hypothesized influence of APM on firm performance. Supporting Hypothesis 1c, inventory management (β = -0.0002009, p < 0.10), the coefficient was found to be negative and significantly associated with ROA. The coefficient indicates that a one-day decrease in the INVM period was associated with a 0.0002009 increase in ROA. This result is also consistent with WCM Theory, the Pecking Order Theory and the findings of Deloof, (2003) and Lazaridis and Tryfornidis (2006). In contrast, the relationship between INVM and TQ was positive and insignificant (β = 0.0029773, p > 0.10), suggesting that higher profitability in terms of TQ is dependent on a longer inventory conversion period or a larger inventory. Increasing Next, the impact of CCC on firm performance was examined (Hypotheses 1d and 2d). Model 1 reveals that CCC was positive and insignificant (β = 0.0000805, p > 0.10); this indicates that a longer CCC is associated with higher ROA. Thus, an increase in CCC by one day was associated with a 0.0000805 increase in ROA of firms. Thus, because the relationship was statistically not significant, Hypothesis 1d was not supported. Whereas in Model 2, CCC was negative and significantly related TQ (β = -0.0014703, p < 0.10), providing strong support for Hypothesis 2d. This depicts that, when a firm shortens its CCC by one day, a profit of 0.0014703 will accrue. This result supports the WCM Theory and the Pecking Order Theory as well. Similarly, the result is consistent with the findings of Deloof (2003). Finally, Hypotheses 1e and 2e predicted that CCE would have a signifi-cant impact on firm performance measured by ROA and TQ. In Model 1, CCE was positive and significantly associated with ROA (β = 0.0281825, p < 0.10), indicating support for Hypothesis 1e. The result is consistent with expectations and suggests that the performance of firms is dependent on the efficient method adopted in managing their production and cash cycle. Specifically, the coefficient means that a one percent increase in CCE was associated with a 0.0281825 profit in the form of ROA. For Model 2, CCC was found to be negative and insignificantly associated with TQ (β = -0.0649369, p > 0.10), which does not support Hypothesis 2e.
Meaning that efficiency in some instances does not translate into higher profits for firms. For the control variables, FSz was positive and insignificantly associated with ROA (β = 0.0087515, p > 0.10) but negative and significantly associated with TQ (β = -1.012187, p < 0.01). These findings suggest that the size of a firm brings advantages that enhance the profitability of firms in some situations, whereas in other situations, it is inconsequential. SGt was positive and significantly associated with ROA (β = 0.0569127, p < 0.001) and TQ (β = 0.03248868, p < 0.05). What these findings suggest is that firms are more likely to increase their profits when their sales increase. FDR was negative and significantly associated with ROA (β = -0.0504328, p < 0.10) but was positive and insignificantly associated with TQ (β = 0.2924548, p > 0.10).  Notes: The first regression result for Model 1 is presented in the column labelled ROA, where return on assets was used as the dependent variable; while the second regression result for Model 2 is presented in the column labelled TQ where Tobin's Q was used as the dependent variable. Variable results begin with their coefficients and t-statistics are in parentheses, while *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Results were obtained using the FE model with robust cluster estimates.

DISCUSSION OF RESULTS
This paper reassesses the relationship between WCM variables and firm performance to address the methodological limitations evident in prior WCM studies in Nigeria. The paper explores one of the largest samples ever used to study listed nonfinancial firms in Nigeria. Overall, the findings of this study reveal that optimization of investment in WCM enhances profitability and the market value of firms. Furthermore, the relationship between WCM variables and firm performance was mixed and inconsistent. For example, ARM was negative and insignificantly related to ROA but positive and significantly related to TQ. The negative result between ARM and ROA highlights the importance of realizing accounts receivable early from customers, and it is consistent with the findings of Deloof (2003) and Lazaridis and Tryfornidis (2006). Whereas, the positive result between ARM and TQ broadly leads to the consideration of extending accounts receivable or credit periods to customer when opportunities for higher sales are envisaged. This finding also extends past studies that found extending receivable periods as essential for improving firm performance (e.g., Sharma & Kumar, 2011). The results of ARM provide a nice link that can help firms attract additional financing. It indicates that the way in which ARM impacts firm performance may depend on evaluating substantially whether allowing shorter accounts receivable periods increases a firm's performance than do longer accounts receivable periods. It is thus possible that the negative result is associated with higher performance for firms that have large market acceptability, while the positive result may be driven by firms seeking to penetrate the market and deplete their stock of finished. This is possible, as noted by Sharma and Kumar (2011), that in India, competition reduced the rate of patronage and that firms had to offer good packages to earn the continuing patronage of customers. Hence, firms need to understand both the needs and impact of their decisions, so that they can make a viable policy. The results provide more understanding of managing WCM in contrast to prior studies, because they provide a more dynamic view of WCM. According to the results between ARM and TQ, the p-values reveal that firm performance is maximized by granting a longer credit period to customers in Nigeria because it is statistically sig-nificant. This result is consistent with the findings of Sharma and Kumar (2011).
Another insightful result of this study is that APM was positive and significantly related to ROA but was negative and insignificantly related to TQ. Indeed, APM offers a direct way to increase the performance of firms. The mixed results again suggest that APM is highly specific to context. The positive relationship between APM and ROA implies that extending payments is a tactical decision to provide free cash flow for financing the operational activities of firms. It supports the notion that a longer APM is associated with higher profitability and is consistent with the findings of Falope and Ajilore (2009). While the negative relationship between APM and TQ underscores the importance of evaluating the cost and benefits of early payment. Hence, whilst extending APM or paying late deprives firms of the leverage to bargain for better pricing and reduces their reputation and opportunity to earn discounts, at the end suppliers may view such practices as a sign of insolvency. This may have the implication of depriving firms access to produce and reach out to their customers with the products or services of the suppliers without making immediate payments for such.
In this context, one useful and transferable lesson from the mixed findings is the importance of analyzing and navigating the shoals between early payment and late payment to determine which could reduce the risk of insolvency and influence firm performance, noting that both extending and early payments have benefits and consequences. Thus, this mixed result advances prior studies by highlighting not only the need to extend payment periods to suppliers as most prior studies argue (Mathuva, 2010;Azam & Haider, 2011) but also to recognize and make sense of opportunities to pay early when discounts and other economic benefits are offered to enhance firm performance. In this study, the p-value between APM and ROA is statistically significant and shows that extending payment to suppliers increases firm performance in Nigeria. This is consistent with the findings of Mathuva (2010), and Azam and Haider (2011).
This study also found a mixed result for the INVM model. INVM was negative and significantly re-lated to ROA but was positive and insignificantly related to TQ. The discrepancies in the result of INVM again suggest that firms need to weigh the costs and benefits associated with holding large inventory when making a choice. This is important for providing uninterrupted production and minimizing costs associated with holding a large inventory. Precisely, such answers the question of the level of inventory that a firm should hold. Unfortunately, the inability of previous studies to highlight the importance evaluating the benefits and costs between holding a large inventory and small inventory prevents greater utility being made of prior studies and could account for the failure of firms in Nigeria. Nevertheless, the p-values of the results show that the negative relationship between INVM and ROA is statistically significant, indicating that minimizing inventory level was associated with higher performance for Nigerian firms. The result reflects the Nigerian condition, suggesting that under conditions of high inflation and unfavorable macroeconomic conditions as Nigeria faces now, the benefits of holding optimal (small) inventory levels that guarantee uninterrupted production do outweigh the potential of large inventory under this condition. This is because once inflation reduces and the economy improves, prices will be adjusted, and holders of large inventory will be faced with adverse shocks. The result is consistent with Deloof (2003) findings.
For CCC and CCE, the results show that both were positively associated with ROA but negatively associated with TQ. The CCC was insignificantly related to ROA but was significantly related to TQ whereas CCE was significantly related to ROA but was insignificantly related to TQ. Unlike prior studies that emphasized that a negative CCC is associated with higher performance (Deloof, 2003;Murugesu, 2013;El-Maude & Shuaib, 2016) or positive as the case with Abuzayed (2012) and Nijam (2016), the result of this study advances prior studies by emphasizing that CCC impacts are firm-driven. Accordingly, the conversion cycle of large firms differs from that of small firms.
In the light of this, for example, a road construction company may have a longer CCC because of the nature of its activities than a manufacturing company that produces sugar. The differences in firm operations are important and manifested in the mixed findings. Noting this demonstrates one of the complex issues this study clears to overcome managerial and policy problems that arise when firms adopt recommendations from studies that do not note differences in their operations. Regarding the CCE, limited evidence exits on its association with firm performance. However, the results of this study show that firm performance is significantly associated with higher CCE.

ROBUSTNESS TEST
The results of the quantile regressions are presented in Tables 7 and 8 for ROA and TQ models, respectively. The results indicate that firm performance measured by ROA and TQ differ consider-ably between the quantiles. For example, the result presented in Table 7 shows that there is heterogeneity over the different quantiles on the relationship between APM, APM, INVM, CCC and ROA. At the 0.10 and 0.50 quantiles, the coefficients of ARM were negative but insignificant and are consistent with the Fixed Effect result. Nevertheless, at the 0.75 quantile, the coefficient of ARM was negative and significant at 1%. Contrarily, at the 0.25 quantile, the coefficient of ARM was positive but insignificant. APM has positive and insignificant coefficients at the 0. 10 Table 6 differ considerably from the Quantile Regression reported in Table 7. In large part, the results reflect the dynamism of WCM variables, and have important implications for understanding the performance of firms in terms of ROA.  Table 6. At the lower bounds of the 0.10 and 0.25 quantiles, the TQ at the 0.10, 0.50 and 0.75 quantiles because the coefficients were negative and statistically insignificant. With regards to the control variables, FSz, SGt and FDR, the results obtained were different across the different quantiles except the coefficient of FDR that is consistent across all quantiles and supports the findings reported in Table 6. In effect, the results presented in Table 8 show that the relationship between WCM and TQ is somewhat mixed across all the quantiles. Hence, a major conclusion to be drawn from the quantile regression is that firms must strive for a greater flexibility when managing WCM as an effective contributor of cash flow that enhances firm performance. This is because WCM is largely dependent on several other factors that change frequently. Such may include firm's operations, customers perceptions, competitions, environmental factors and many others.

CONCLUSION
The findings of this study provide practical insights into the management of working capital by firms, specifically in Nigeria. Two important findings emerged: First, the mixed results highlight the point that WCM variables need to be understood and managed in the context of a firm's peculiar conditions to provide the cash-flow for financing operational activities of firms and increase their performance. Therefore, a knowledge of the business environment, customers, suppliers and market conditions are essential to achieving this goal. Second, the existing literature on WCM in Nigeria is insufficient to guide policy-making by firms in Nigeria. This is because the sample sizes of most studies were small and predominantly taken from one sector, yet, their results were generalised. The implication of this is that policies may be formulated and implemented based on such recommendations whereas these firms are not part of the sample studied. This may lead to a policy mismatch and could have detrimental effects on the performance of firms in Nigeria. Additionally, the findings of this study were derived from the application of rigorous analytical tools and the use of a larger sample size that is representative of non-financial firms in Nigeria (see appendix A), thereby extending the significance of its results beyond the study's universe. Therefore, this paper is deemed important not just to non-financial firms in Nigeria, but to similar firms in the developing world and beyond.
Theoretically, this study advances WCM knowledge by addressing the methodological limitations evident in WCM literature in Nigeria. Broadly, the study integrates the WCM literature by substantiating the mixed results in prior studies. The findings of this study imply that WCM needs to be understood in the context of a firm's specific condition to increase performance. In this way, the study contributes to WCM literature by emphasising the focus on the importance of recognising differences in operational activities of firms. Moreover, previous studies overlook the issue of heterogeneity and its effect. Using the quantile regression at the 0.10, 0.25, 0.50 and 0.75 quantiles, this study shows a presence of heterogeneity across these various quantiles for the relationship between WCM and firm performance in Nigeria. Therefore, this study contributes to WCM literature in terms of methodological approach and showing that WCM is dynamic even at short interval.
Future research may re-examine the impact of WCM on the performance of firms in the financial sector using an expanded sample size because this study only considered non-financial firms. Another research avenue is to extend this study by determining the sensitivity of the findings in this study through other methodological approaches. Finally, a new framework that incorporates the effect of operational activities of firms on WCM variables needs to be examined.