Quantitative Data Analysis and Logistic/Supply Chain Management Decision-making - Research Paper Example

Quantitative Data Analysis and Supply Chain Management Decision Making: Application of Regression Analysis in Sustaining Zero Inventory Policy Institutional Affiliation: Executive Summary This essay presents an analysis and evaluation of the application of statistical techniques in deciding the levels of stock to purchase in order to sustain a zero inventory policy in a firm. In particular, the paper explores the applicability of chi square tests, Analysis of Variance and regression analysis in maintaining zero inventory. To arrive at a decision on the most suited model, review of literature pertaining applicability, strengths and weaknesses of each type of test with regard to the role of forecasting for a supply chain department has been done. The results indicate that regression analysis is a stronger forecasting tool, compared to both chi square and Analysis of Variance tests. The study recommends the use of regression analysis for determining optimal amounts of inventory in order to effect the zero inventory policy. The results discussed are based on empirical analysis of results. The shortcomings of this method of analysis have been discussed. Introduction Supply chain management is a field that has been expanding both in applicability and research over the last three decades (Boon-itt, 2011). Besides taking complexity in the techniques used to optimize results for respective departmental managers, researchers have also succeeded in integrating theories and practices used in many other fields to successfully solve complex situational and operational problems (Civelek, 2014; Done, 2011). These new “intermarriages” of supply chain theories and methods with scientific techniques present managers not only with solutions to their day-to-day issues, but they also provide insights into new fields of exploration for the ever evolving field of supply chain management (Modi & Mabert, 2010). For instance, computing and information technology have been accredited for bringing in efficiency, transparency and reliability in the field of supply chain management through introduction of customized software to track operations and transactions (Paiva, Phonlor & D’avila, 2008). Engineering techniques have eased means of communication and increased efficiency of managerial units through provision of customized hardware that respond to managerial needs. Equally, mathematical and statistical techniques provide both predictability and accountability in the management of supply chains. These two attributes are important for decision making in a firm (Qi, Boyer & Zhao, 2009). This paper presents an examination of how to implement a zero inventory policy using statistical techniques. In particular, the paper presents regression analysis as a tool for assessing the volumes of each product required to effectively eliminate likelihoods of stock outs in both the inventory and supplies management. To achieve this target, I have explored three statistical techniques (the chi-square analysis, analysis of variance, and regression analysis) and made contextual analysis in terms of their suitability and adaptability in the situations of determining whether to adopt a zero inventory policy. The main limitations of using regression analysis for the specified fields of the policy determination process and ways to overcome the limitations have also been discussed. Choosing an Appropriate Statistical Technique Selecting an appropriate technique to use when presented with a situation that needs decision making on the side of the management is based on the type of data available, the exact situation, and effectiveness of the technique in solving the observed problem (Seric, Rozga & Luetic, 2014). Some common problems that managers have to contend with include having to decide which product is faring better than the other, the level of efficiency of a department, when to purchase new stock and the amount of quantity to purchase, how much stock to pile in the stores so as to avoid embarrassing stock-out phases among several other decisions (Singh & Wahid, 2014; Toyin, 2012). In this analysis, I have particularly addressed the issues of predicting the amount of stock to purchase in order to sustain zero inventory based on existing data. Chi Square Analysis The chi square test is suitable for comparing two or more groups when the outcome variable is categorical (Done, 2011). When the data presented successfully determines the units of an item needed to entirely meet the demand/ supply needs of a particular product, the test helps in determining whether this data can help in resolving an issue of how much of another commodity is needed to satisfy these forces. This is done through the chi square test of independence. The test is a powerful tool for examining the amount of stock that the firm should purchase or produce. This is based purely on the existence of associations among the products (variables) under evaluation. The main drawback of the technique is that it is much limited to the determination of associations between sets of variables in terms of independence of variables in order to help in decision making. When an association cannot be established, the technique is not helpful in predicting the levels of commodities to keep in the firm/ department (Agus, 2011). In this respect, a versatile tool that can evaluate associations in more than one way is better suited than the chi square test. Analysis of Variance ANOVA, as the test is commonly referred to, is a tool for determining whether there exist significant differences between/ among variables in a model. The test is most suitable when examining whether the volumes/ units of items have significant differences between them, in terms of their respective means as provided by the F-statistic and its corresponding p-value (Humphries, Towriss & Wilding, 2007). One strength of the test is that it is able to split up specific variables according to their deemed characteristics, and provide a means to draw suitable conclusions from them. For instance, the test is capable of separating gender as a variable in terms of whether a group is male or female, and present internal differences within the particular variable. This attribute is shared by the t-tests, which take various forms that address specific needs of the researcher (Aggarwal, Aggarwal, Sharma & Khurana, 2010; Humphries, Towriss & Wilding, 2007). This presents ANOVA as a relatively weaker tool compared to its equivalent, the t-tests. One more strength of the test is that it provides a basis for evaluating experimental data. However, ANOVA is based on various strict rules, including the assumption of normality of the error term, whose distribution is shared by the response variable. This rule is not always met, and it requires complex treatment to overcome the challenge or adoption of an equivalent non-parametric test. ANOVA is not always helpful in a supply chain situation, since its applicability is based on the need to examine differences among means of commodities, while the main concern is accurate prediction of commodity amounts with regards to their associations with other commodities (El-Tannir, 2014). Again, a test that would respond to this more common need is essential in the prediction of optimal stock levels that enhance zero inventory. Regression Analysis Regression analysis is useful in determining relationships between variables in an equation. The technique employs several approaches to analyze data, based on suitability in terms of adherence to various assumptions. The test possesses one characteristic that recoups for the shortcomings of both ANOVA and chi square tests. This is reflected in: 1) The ability to test associations in different ways, including linear, logistic, non-linear among others. 2) The possibility to address issues of failure to meet assumptions appropriately and easily using a single test; for instance, regression analysis in MS Excel. As observed, regression analysis provides equations that help in determining the forecasts of productivity or purchases based on interrelated factors in the department. This makes the test the most versatile, adaptive, and flexible choice for application in the determination of supply volumes and inventory levels for a business. Regression Analysis in Determining Levels of Purchases for the Zero Inventory Policy Zero inventory is a policy that requires a firm to retain only the necessary amount of stock applicable for the present operational processes in the firm, thereby reducing the costs associated with maintenance of inventory (Shah, 2012). When successfully adopted, the firm is capable of maximizing cash flow, and making a wider range of important decisions. Determination of the volumes of stock to purchase is based on set production/ manufacturing levels, which are a reflection of uptake rates in the market. This important consideration has several implications on the firm as a whole. If production/ purchase level fails to meet the needs of its customer base, the firm will be effectively under-producing, implying failure to meet market needs. This has the implication of giving competitors advantage through increased sales volumes, and draining consumer confidence in the long run (Ding, 2014; Shil, 2010). More so, such occurrences result from unpredictable consumption patterns, which result from failure to determine the correct interrelations between demand and supply. However, using regression analysis of existing data, it is possible to determine the exact amounts of each commodity required to meet the needs of another process, effectively sustaining continuity of the production chain through a coordinated and predictable supply chain in such a way that no inventories are retained unnecessarily. Four variables have been selected from the available data set in order to make an informed decision as to whether the firm can reduce the amount of stock to just the amounts that will be sold through the month. In this case, we make the assumption that the stocks need to last through the month. For this to be realized, the stock that is yet to be sold should be fully disposable by the time the next order arrives. Furthermore, the company will want to make one order to last through the month in order to minimize ordering costs. With these assumptions made, the task is to make a regression model on how the accompanying independent variables (stock on hand, stock ordered, and wait time) can be fixed to only the latest figures on monthly sales per product (dependent variable). The following results were obtained. Regression Statistics Multiple R 0.881703 R Square 0.7774 Adjusted R Square 0.775259 Standard Error 101.7988 Observations 316 ANOVA   df SS MS F Significance F Regression 3 11291676 3763892 363.205 2E-101 Residual 312 3233255 10363 Total 315 14524931         Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0% Intercept -8.89071 9.909656 -0.89718 0.370317 -28.3889 10.60749 -28.3889 10.60749 Stock on hand latest (Units) 0.230995 0.016138 14.31391 4.48E-36 0.199242 0.262748 0.199242 0.262748 Quantity ordered waiting receipt 0.225732 0.012334 18.3017 2.44E-51 0.201464 0.250001 0.201464 0.250001 Expected wait (number of days) 3.434132 1.419068 2.41999 0.016092 0.641978 6.226286 0.641978 6.226286 Interesting statistics in the regression output are the adjusted R-squared, the F-value and its accompanying p-value, and the coefficients of the independent variables and the constant. According to the adjusted R square, the chosen variables accounted for 77.5% of the overall change in the number of units sold in the previous month. This figure is relatively high, indicating that the choice of independent variables is justified for the response. The F-value (363.205) is significant at the 5% level of significance (p < 0.001), implying that the coefficients of the independent variables are greater than zero. This also implies that the model depicted by the coefficients is significant. The model obtained from the coefficients is: Number of items sold in one month = β0 + β1(Stock on hand) + β2(Quantity ordered waiting receipt) + β3(Expected wait time). Using the figures in the table, the exact model is: Number of items sold in one month = -8.891 + 0.231(Stock on hand) + 0.226(Quantity ordered waiting receipt) + 3.434(Expected wait time). In this case, the βs represent the coefficients of the independent variables. For instance, β1 is the coefficient of the stock in the premises at present. The coefficient has a large role in our analysis of the situation. For example, β1 indicates the number of items by which the stock presently held changes with every unit change in the number of items sold in a month. With the three independent variables, we can estimate the amount of change that would occur in the total number of items sold per month, provided the other factors of production are kept constant. Meanwhile, the equation provides a means to evaluate efficiency and transparency through predictability in the sales expected from a level of stock purchases. For instance, the management is able to determine when underproduction occurs, and follow up on the faulty departments, which is essentially among the primary roles of supply chain managers (Storey, Emberson, Godsell & Harrison, 2006). With these deterministic models produced through regression analysis, the management can be able to meet any challenges posed by demand or supply of the materials at any phase of the production process. For example, if the management determines accurately that demand will fall at a particular time, they will reduce the orders placed for the items to reflect this anticipated change. This approach eliminates massive wastage caused through purchasing of surplus stock compared to the available volumes of another (Kannabiran & Sundar, 2011; Kushwaha, 2012). Clearly, regression analysis is an important test in determining the balance between purchases and sales rates. Using the model, we now estimate the number of items that should be purchased in order to ensure that the firm has zero inventory in accordance with the assumptions indicated above. We take an example where we wish to evaluate a product whose wait time is 30 days, had 200 units sold last month, and does not have remnants at the moment. 200 = -8.891 + 0.231(0) + 0.226(unknown) + 3.434(30) = 468 units. This is the average number of items that the firm should purchase at the moment in order to sustain the zero inventory policy. Indeed, several factors determine the levels of demand for a firm’s products. Among these is quality, timeliness of delivery, adequacy of supply among others (Vieira et al, 2013; Sukati, Hamid, Baharun & Tat, 2009). However, the firm has many considerations to make, including viability of target markets (Tesfay, Sakita & Mawrides, 2014). For the above firm to maintain zero inventory, data on each of the four variables must be accurately recorded. The importance of this relational equation is demonstrated by its ability to correctly predict what unitary impacts a shift in any of the factors would have on the other factors (El-Tannir, 2014; Miguel, 2011). In order to attain the zero inventory, it is necessary to accurately predict expected sales volumes and wait time. This demonstrates why regression analysis is such an important test for supply chain managers. Limitations of Regression Analysis as a Tool for Determining the Zero Inventory Policy and How to Address Them Despite being such an acclaimed tool of analysis, regression analysis has several major drawbacks that affect its application in various situations. One major limitation is the inability of the results to reveal causal mechanisms in a data set (Shil, 2010; Civelek, 2014). It is important to note that obtaining strong relationships through either regression or correlational analysis does not imply causality. In practice, causality is concerned with whether the variables evaluated are the actual causes of the increase or decrease in the selected response variable (Shil, 2010; Aggarwal et al, 2010). The fact that relationships obtained through regression analysis cannot be generalized to imply the dependent variables were the automatic causes of change in the dependent variable partly limit the credibility of results so obtained. Researchers have tried many techniques to address this inadequacy of regression analysis; including introduction of external variables to explain the variations in the variables. Though not always considered external variables, mediating and moderating variables present a clearer view of how much influence an external variable that is conceptually related to both the response and independent variable could have on the relationship (El-Tannir, 2014). Another limitation of regression analysis emanates from the fact that in the case that the data fails any of the assumptions laid out, the results obtained will not fulfil the requirements of the test. This situation will imply that the results so obtained are useless to the investigator (Shil, 2010). However, the problem can be addressed by employing techniques that reduce the data to meet the set assumptions. One such process is normalization of data. The method relies on historical data, which places start-ups at a disadvantage since they take some time to be able to accumulate meaningful data for the purposes of evaluation. For these firms, it is almost impossible to not only implement a zero inventory policy but also any meaningful inventory policy in the initial stages. Such firms are encouraged to make short term projections of their situation in order to overcome the disadvantage. Conclusion Regression analysis has various strengths over chi square tests and ANOVA when making decisions in supply chain management, and particularly in predicting optimal levels of stock for the model of choice adopted by a firm. These advantages make the test the most prime tool for decision making among supply chain managers. However, it cannot be applied in all situations that call for statistical analysis, as every test is best suited to generate helpful results based on situational criteria. The fact that the main target of supply chain managers is integrating smoothness in the supplies operations of departments and the firm as a whole makes deterministic analysis the more preferred choice for this group of managers. However, as demonstrated, the test has several limitations, which, when addressed individually, allow analysts to obtain helpful information from data. References Aggarwal, K. K., Aggarwal, R. B., Sharma, R. K. & Khurana, A. (2010). Relevance of knowledge towards measurement of human resources on investment decisions in Sri Lanka. International Journal of Research in Commerce and Management. 1(5). ISSN: 0976-2183. Agus, A. (2011). Supply chain management, supply chain flexibility and business performance. Journal of Global Strategic Management. 9: 134-145. Boon-itt, S. (2011). Achieving product quality performance: The roles of supply chain integration and information technology. International Journal of Innovation Management and Technology. 2(5): 373-376. Civelek, I. (2014). A commodity production model with operational flexibility of investing optional capacity on offshore platforms. International Journal of Supply Chain Management. 3(3): 1-6. Ding, Y. (2014). The main affecting factors of the B2B e-commerce supply chain integration and performance. International Journal of U- and E- Service, Science and Technology. 7(1): 145-158. Done, A. (2011). Developing supply chain maturity (Working Paper No. WP-898). Retrieved from the University of Navarra website: http://www.iese.edu/research/pdfs/DI-0898-E.pdf. El-Tannir, A. A. (2014). Bullwhip effect variance ratio approximations for aggregated retail orders in supply chains. International Journal of Supply Chain Management. 3(3): 43-48. Humphries, A. Towriss, J. & Wilding, R. (2007). A taxonomy of highly interdependent, supply chain relationships: The use of cluster analysis. Journal of Logistics Management. 18(3): 385-401. Kannabiran, G. & Sundar, S. (2011). A study on impact of IT-leveraged supply chain operational benefits on competitive marketing performance. Great Lakes Herald. 5(2): 37-45. Kushwaha, G. S. (2012). Operational performance through supply chain management practices. International Journal of Business and Social Science. 3(2): 222-232. Miguel, P. L. & Brito, L. A. (2011). Supply chain management measurement and its influence on operational performance. Journal of Operations and Supply Chain Management. 4(2): 56-70. Modi, S. B. & Mabert, V. A. (2010). Exploring the relationship between efficient supply chain management and firm innovation: An archival search and analysis. Journal of Supply Chain Management. 46(4): 81-94. Paiva, E. L., Phonlor, P. & D’avila, L. C. (2008). Buyer-supplier relationship and service performance: An operations perspective analysis. Journal of Operations and Supply Chain Management. 1(2): 77-88. Qi, Y., Boyer, K. K. & Zhao, X. (2009). Supply chain strategy, product characteristics, and performance impact: Evidence from Chinese manufacturers. Decision Sciences. 40(4): 667-696. Seric, N., Rozga, A. & Luetic, A. (2014). Relationship between business intelligence and supply chain management for marketing decisions. Universal Journal of Industrial and Business Management. 2(2): 31-35. Doi: 10.13189/ ujibm.2014.020202. Shah, P. (2012). Zero inventory: A myth? Logistics Week. Retrieved from http://www.dsims.org.in/web/popular_media/Log_India_Zero_Inventory_A%20Myth_Piyush_Shah_July_2012.pdf. (Accessed 19th October, 2014). Shil, N. C. (2010). Customized supplier selection methodology: An application of multiple regression analysis. Supply Chain Forum – An International Journal. 11(2): 58-70. Singh, G. & Wahid, A. (2014). Supply chain risk management: A review. International Journal of Supply Chain Management. 3(3): 59-67. Storey, J., Emberson, C., Godsell, J. & Harrison, A. (2006). Supply chain management: Theory, practice and future challenges. International Journal of Operations and Production Management. 26(7): 754-774. Doi: 10.1108/01443570610672220. [Emerald]. Sukati, I., Hamid, A. B. A., Baharun, R. & Tat, H. H. (2009). A study of supply chain management practices: An empirical investigation on consumer goods industry in Malaysia. International Journal of Business and Social Science. 2(17): 166-176. Tesfay, Y. Y., Sakita, B. M. & Mawrides, E. K. (2014). Developing a new market strategy from supply chain management perspective: A case of Zotun in Brazil. International Journal of Supply Chain Management. 3(2): 94-110. Toyin, A. I. (2012). Supply chain management (SCM) practices in Nigeria today: Impact on SCM performance. European Journal of Business and Social Sciences. 1(6): 107-115. ISSN: 2235-767X. Vieira, L. M., Paiva, E. L., Finger, A. B. & Teixeira, R. (2013). Trust and supplier-buyer relationships: An empirical analysis. Brazilian Administration Review. 10(3). [EBSCO]. Read More