Empirical investigations of the applicability of sampling techniques for inventory valuation by computer simulation

Abstract

Industrial engineering techniques such as work simplification and methods improvement have become increasingly important as industries, under the stimulus of intense competition, strive to improve operational efficiency and reduce costs. These techniques are equally essential for the successful functioning of service organizations. Among these, statistical techniques have contributed significantly to work simplification. Over the past few decades, statistical methods have been systematically applied across various fields, including business, industry, and scientific research. Statistical sampling techniques have long been employed in industries to improve product quality and reduce costs. They have also been successfully applied to internal accounting procedures and management control problems. However, the use of sampling techniques for inventory valuation has only recently attracted attention, as they offer reliable estimates at reduced time and cost. Organizations, both manufacturing and service-oriented, often maintain large inventories to ensure material support. In such cases, inventory valuation based on complete enumeration of all items becomes prohibitively expensive and time-consuming. Consequently, the need for work simplification in inventory valuation has been strongly felt by banks, auditors, and other stakeholders. Statistical estimation procedures, which provide results within allowable error limits and with a known degree of risk by examining only a representative portion of the population, offer a promising solution. Although some studies on the application of statistical sampling techniques for inventory valuation have been reported in the literature, they suffer from limitations from both scientific and practical perspectives. The present study aims to scientifically investigate the applicability of sampling techniques for inventory valuation by empirically examining the behavior of these procedures under different environmental conditions (i.e., inventory populations with varying characteristics) using computer simulation.

Scope of the Study The problem is introduced in Chapter I, along with a brief review of earlier works in India and abroad. Chapter II and Chapter IV deal with the empirical investigation of the applicability of Stratified Random Sampling and Optimum Allocation Sampling Plans for inventory valuation. Computer simulation methodology was adopted to repeatedly test these sampling procedures on two different inventory populations to generate sampling error distributions. The empirical results indicate that the performance of these sampling procedures is unsatisfactory under very high accuracy specifications and for inventory populations with high variability, regardless of the size of the preliminary random sample. Furthermore, the results clearly show that population characteristics significantly affect the behavior of the sampling procedure. The applicability of these procedures can be improved by specifying a lower allowable error or a higher confidence level to achieve the desired precision. Most statistical sampling plans require the population variance to determine the required sample size objectively. In practice, this variance is estimated from a preliminary random sample of suitable size. However, there is no established theory or empirical evidence suggesting an optimal size for such preliminary samples for practical purposes. Chapter III addresses this gap by empirically investigating the sufficiency of different preliminary sample sizes, ranging from 20 to 100, to determine a practical recommendation. The results suggest that, although population variance directly influences the required sample size, a preliminary sample of 60 items is generally sufficient for practical applications. The author hopes that these empirical studies will assist practitioners in effectively applying sampling techniques for inventory valuation, thereby reducing time, cost, and effort while maintaining acceptable levels of accuracy and reliability.