Click for next page ( 50


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 49
V. THE EFFECT OF ORGANIZATIONAL STRUCTURE ON INVENTORY MANAGEMENT PERFORMANCE 5.1 ORGANIZATIONAL STRUCTURE AND INVENTORY MANAGEMENT DECISIONS The objective of this analysis is to examine the effects of a public transit agency's inventory management organizational structure and inventory management decisions upon the agency's inventory performance. In addition, selected agency and fleet characteristics were included in the analysis to determine if these factors, combined with organizational factors, have an effect on inventory performance. The organizational and inventory management decision factors, agency and fleet characteristics, and inventory performance indicators used in this analysis are summarized below in sections 5.1.1, 5.1.2, and 5.1.3. Section 5.2 of this chapter describes the approach and statistical techniques that were used in analyzing the effects of the factors and characteristics on inventory performance. Section 5.3 describes the organizational structure and inventory management decision factors in more detail and presents the results of the analysis of their effects on inventory performance indicators. Section 5.4 presents the general relationships between significant groups of factors and the performance indicators resulting from regression analysis. Section 5.5 summarizes the conclusions drawn from the analysis. 5.~.~ Organizational Structure and Inventory Management Decision Factors For the purpose of this analysis, a transit agency's organizational structure Includes the policies and management decisions that define how inventory is to be managed, as well as the personnel reporting relationships, responsibilities, and staffing levels. In particular, the effects of the following factors were examined in this analysis based on responses from the survey conducted during this study: . Staffing levels for managing inventory, both within and outside of the inventory management organization Organizational structure type (based on the five organizational structure types defined in Chapter m) and specific attributes, such as the responsible department and level, the existence of a fil11-time organizational head, the separation of the inventory planning function, etc. Setting inventory management goals for stocking levels and sentence Methods used to replenish inventory matenal, including safety stock levels Purchasing practices for inventor material 49

OCR for page 49
Storehouse network configuration and coverage Published procedures and catalog Physical inventory/cycle counting Dequencies Repair and stocking of re-usable components Use of systems and technology These factors and the specific attributes that were analyzed are described in detail in section 5.3 of this Chapter. S.~.2 Transit Agency and Fleet Characteristics As concluded in Chapter IV, transit agency and fleet characteristics alone have no significant effect on inventor perfonnance indicators, except for some differences between bus and rail Inventory. However, the following agency and feet characteristics were included in the regression analyses to determine if these characteristics have any effects in combination with specific inventory management structures and decisions: number of vehicles In Beet (bus, rail, and total) dollar value of inventor (bus, rail, and total) agency operating cost number of vehicle models (bus, rail, and total) inventor purchase dollars (bus, rally average annual miles (bus, rail) average vehicle age in years (bus, rail, and total) average vehicle age in percent of expected life expended (bus, rail, total) annual passenger miles (bus, rail) percent of vehicle manufactured In the United States (bus, rail, and total) (See Chapter ~ for more information on these factors.) 5.~.3 Inventory Performance Indicators The foldowing performance indicators, defined in Chapter IV of this report, are used to define inventory performance in this analysis: 2. 4. 5. 6. Bus Inventory Dollars per Vehicle Rail Inventory Dollars per Vehicle Annual Bus Inventory Turnover Annual Rail Inventory Turnover Bus Percent Demand Filled (Fill Rate) Rail Percent Demand Filled (Fill Rate) 50

OCR for page 49
Percent of Items Stocked Out per Week Average Days to Fib Bus Inventory Backorders Average Days to Fill Rail Inventory Backorders 1. 8. 9. 10. Percent of Items Out of Balance ~ I. Percent Obsolete Bus Inventory 12. Percent Obsolete Rail Inventory 13. Total Inventory Dollars per Person 14. Inventory Personnel Cost per Inventory Dollar IS. Total Inventory Transactions per Person 5.2 MET HODOLOGY FOR ANALYZING ORGANIZATIONAL EFFECTS The analysis summarized In this chapter examined the eject of each organizational structure and inventory management decision factor on the inventory performance indicators. The responses to the survey conducted during this study were used as data for this analysis. The inventory perfonnance indicators were derived Dom survey Section V data. Organizational and Inventory management decision factors were derived Dom sections m, IV, and V! of the survey. Transit agency and feet characteristics were denved Dom survey sections ~ and H. 5.2.! Preparing Survey Data for Analysis ~ perfonriing the above statistical analyses, the data was prepared as follows: "Yes/No" responses were coded. Factors whose categories were "yes" or "no" were quantified as "I" and "0" for inclusion in correlations and regressions. For example, survey responses as to whether an agency uses blar~cet purchase orders to replenish inventory were coded as ~ or 0 to represent "yes" or "no". This technique was also applied to each of the five orgaruzational types so that organizational type could be used In quantitative analyses. For example, if an agency's inventory management organization was type 3, then a vanable representing organizational type 3 was set to "I", and four vanables representing organizational types I, 2, 4, and 5 were set to "0". Missing responses were eliminated. Many survey respondents did not answer ad questions. As a result, some of the values of the orgaruzational and management decision factors were blank. The entire observation was excluded from analysis when the unanswered question was a variable In the analysis. For example, if a survey respondent did not specifier whether blanket orders were used In purchasing inventory material, all of the respondents' answers were omitted from analyses involving the use of blanket orders. Eliminating the entire observation ensured that the value of the performance indicator being examined related only to the values of the organ national and management decisions In the analysis. The observation was eliminated only for those analyses involving the missing response. Bus and rail responses were applied separately. Rail variables were excluded for bus inventory perfonnance indicators and vice versa. For example, the number of buses was not included in the regression analysis for rail inventory dollars per vehicle. Variables for the total fleet were used for 51

OCR for page 49
analyzing performance indicators representing overall inventory performance, but two yes/no variables were included to identify whether the property "has buses" and "has rail". For example, the total number of vehicles, as well as variables indicating whether the agency has bus and/or rail, were included in analyzing overall inventory accuracy (percent of items out of balance). Identical responses were eliminated. Variables were excluded when all ofthe survey responses were the same for the selected performance indicator. For example, all rail properties used a reorder point method to replenish inventory (response = "yes"). Therefore this variable is actually a constant in our survey and cannot be used to analyze the effect on inventory performance. As a result, this variable was excluded from the analysis of rail inventory performance indicators. 5.2.2 StatisticalAnalysis Methods Three primary statistical methods were used for this analysis, based on whether the organizational factors were numerical (e.g. number of storehouses) or divided into categories (e.g. reporting to maintenance, finance, administration, etc.). Correlation and regression were used for quantitative data and l-tests were used for category data. Section 4.3.2. 1 (Chapter IV) of this report describes correlations and section 4.3.2.2 describes t-tests. Regression analysis was used to examine the relationship between a performance indicator and a combination of organizational and inventory management decision factors. Regression analysis is a statistical process that defines a formula for predicting the value of a performance indicator using the values of organizational and management decision factors. For example, a regression equation can be derived to predict the value of annual inventory turnover using the number of buses, the number of individuals assigned to inventory management, and whether the authority uses an automated inventory system. Regression analysis will result in the equation that best utilizes the values in the survey to predict inventory turnover. Note that regression analysis does not imply that the number of buses, number of inventory management individuals, and the automated system causes inventory turnover to go up or down. The analysis simply defines the relationship between the variables. Two regression analysis methods were used in this study: (1) multiple linear regression and (2) step-wise linear regression. Multiple linear regression produces a formula for a line that uses Redefined multiple variables to predict the independent variable. All variables are included in the regression equation. Step-wise linear regression builds a linear equation by selecting variables one at a time from a Redefined set of variables. This technique selects the best variables for predicting the independent variable and does not necessarily include all variables in the equation. Both of these regression methods develop equations using the "least-squares" method. The least-squares method is a mathematical technique that produces a linear equation that minimizing the sum of the squares of the distance of survey data points from the regression line. The resulting regression equation is of the form: y = C + ClXl + C2X2 + C3X3 ... + CiXi Where: y is the independent variable (e.g. annual inventory turnover) 52

OCR for page 49
c is the constant for the equation (or the y-~ntercept of the line) x1 through ~ are the dependent variables (e.g. number of buses) c' through ~ are the coefficients for the dependent variables i is the total number of dependent variables In the equation Performing regression analysis involved the following steps: Eliminate redundant variables from the analysis (multi colineanty). In developing the regression equation, it is important to eliminate dependent variables that are highly correlated with each other. These variables wiD distort the analysis and could lead to incorrect conclusions. For example, a transit agency's total number of buses and the total bus passenger miles per year are highly correlated. These two variables represent redundant data and could distort the analysis, such as resulting in coefficients that mostly offset each other. Since these two variables represent almost the same data pattern, one of the variables is eliminated Dom the regression analysis. The technique used to identify highly correlated variables, or multi-colinearity, was to develop a correlation Manx. This matrix showed the correlation between each variables In the survey with each other variable. As a general rule of thumb, when vanables showed a correlation of over .70 (or less than -.70), one ofthe variables was eliminated from the regression analysis. Select the dependent vanables that best predict the independent vanable. Once multi~olinearity was addressed, the next task was to select the set of vanables Mom the 45 - 50 remaining that best predict the Independent vanable. Step-w~se regression was used for this task. Ste~w~se regression selects variables one at a tune to include or exclude Dom the regression equation until the best set of variables is found. The significance level can be set as a enters for including dependent variables. The higher the significance level, the fewer variables wid meet the criteria and be included in the equation. Step-w~se linear regression was run on the remaining survey data variables for each inventory perfonnance Indicator. The significance level for including variables In the equation was varied to observe the resulting effect on the variables included in the regression equation, the amount of variance in the independent variable explained by the equation, and the overall confidence level of the prediction. For example, a minimum 99/O confidence level may result In an equation with few variables or may not produce an equation at all if no variables exceed the 99% minimum rate. Lowering the minimum rate to 95/O may allow additional variables to be included and therefore increase the percent of variance in the independent variable that is explained by the equation. However, if the minimum rate is lowered too much, the confidence In the prediction resulting Tom the equation decreases, even though the equation explains a higher percent of the vanance. Furthermore, there is lower confidence that the dependent variables included in the equation actually have an effect. 53

OCR for page 49
Regressions were run until all variables with a 90% or greater confidence were included in the regression equation. ~ other words, there is at most a 10% probability the dependent variables that the regression selected Dom the survey data as significant, are really due to change and should have a coefficient of zero. The dependent vanables Dom this step-w~se regression analysis having a confidence of 90% or greater were selected for the next step. 3. Denve the regression equation. A multiple linear regression analysis was performed on the dependent variables identified in the previous step. This analysis resulted in a regression equation, a coefficient of determination (r23, an analysis of vanance, and other statistics. 4. Determine how well the equation predicts the independent variable. The coefficient of determination is the percent of variance in the independent variable explained by the regression equation, based on the survey data. This coefficient is a measure of how well the regression equation predicts the independent variable. For example, if a regression is run for annual inventory turnover with an rid of .85, the recession equation win explain 85% of the - var~ance In annua Inventory turnover. Determine the confidence level of the prediction. The F-statistic and the F-distnbution are used to determine the confidence level of the prediction resulting from the equation. The confidence level measures the confidence that the prediction resulting Dom the survey data's regression equation is not equal to zero (i.e., the null hypothesis). This is the confidence that all ofthe coefficients in the recession equation and the constant are not equal to zero. For example, if the confidence level is 92%, one can infer with 92% confidence that the predicted value of an inventory performance indicator is not equal to zero. 6. Determine the sigruficance of each dependent variable. Just as a confidence level can be calculated for the entire regression equation, a confidence level can also be calculated for each coefficient of each dependent variable. This confidence level is a measure of the probability that each dependent variable's coefficient is not equal to zero. S.3 THE EFFECT OF ORGANIZATION & INVENTORY MANAGEMENT DECISIONS This section summarizes findings regarding the effect of each organizational and inventory management factor on inventory management indicators. Only effects with a confidence level of 85% or higher for t-tests, and a correlation coefficient of .70 or higher (-.70 or lower) are identified as significant. Data groups must have at least three responses to be included in l-tests 54

OCR for page 49
and five for correlations. The following sections describe each organizational and inventory management factor in detail and present the findings associated with the factor. 5.3.! Invento~yStaff~ng Levels The effect of inventory management staffing levels was examined separately for (~) inventory organization personnel - total number of individuals within the inventory management organization, and (2) total inventory management personnel - total number of individuals within the transit agency with inventory management responsibility, regardless of whether they are within the inventory management organization. Inventory staffing levels were analyzed relative to the total number of vehicles in the transit agency and the agency's total inventory dollars. This allowed the effects of relative staffing levels to be compared across agencies of difference sizes. Based on this analysis, there is no correlation between relative staffing levels and any of the inventory perfonnance indicators. Furthermore, there is a high correlation between inventory organization employees and total employees involved with inventory management (r = .997). Therefore, it is not necessary to include both measures as an inventory performance indicator. Since the total number of employees involved with inventory provides a more comprehensive view of inventory management activities, it is used in inventory performance indicators "per person" (such as transactions per person or dollars per person). 5.3.2 Organizational Structure Type Five organizational structure types were defined in detail in Chapter III, ranging from no formal inventory management Unction to a dedicated inventory management group at the department level. In addition to examining the effects of these organizational structure types, selected organizational attributes were analyzed separately, such as the level to which the inventory organization reports, the departments to which the organization reports, the separation of the inventory planning functions, etc. The ejects of the organizational types and each attribute is presented separately. Organizational Type 1 - no formal inventory management filnction, inventory responsibility located in the maintenance department Organizational Type 2 -- formal inventory management function located in a department other than maintenance reporting to the sub-department level Organizational Type 3 - formal inventory management function located in the maintenance department 55

OCR for page 49
Organizational Type 4 ~ Organizational Type 5 - formal inventory management function located in a department other than maintenance reporting to the department level a dedicated department level inventory management group All rail transit properties that responded to the survey have organization Type 5, except for two which have Type 4 and one which has Type 3. Therefore, there were not enough rail property responses in organization types other than Type 5 to perform l-tests for each organization type. As a result, l-tests for rail inventory performance indicators were performed based on two categories: those that use organization Type 5 and those that do not use organization Type 5. formalization Type. The findings for the effects of organization type are: Agencies with Type ~ organization have less bus inventory dollars per vehicle than those with Type 2 and Type 5. Those with Type 4 also have less bus inventory dollars per vehicle than Type 5 organizations. Type ~ -- $3,175 Type 2 -- $7,612 (~36% confidence) Type ~ -- $3,175 Type 5 -- $6,408 (96% confidences Type 4 -- $4,403 Type 5 -- $6,408 (~% confidence) . Agencies with organization Type 5 have higher annual bus inventory turnover than those with Types 2, 3, or 4. Type 5 -- 2.61 Type 2 -- I.22 (91 /O confidence) Type 3 -- I.61 (85% confidence) Type 4 -- 1.21 (94% confidence) Agencies with organization Type 5 have higher inventory fill rates than those with Types 1, 3, or 4. Type 5 -- 95.~% Type 2 -- 134.0% (95% confidence) Type 3 -- 136.2% (95% confidence) Type 4 -- 137.3% (89% confidence) Agencies with organization Type 4 have a lower stockout percentage than those with Type 5 organizations, 0.14% compared to 1.64% (90% confidence). Agencies with organization Type 3 take more days to fill a backorder for bus inventory than those with organization 2 or 5. Agencies without organization Type 5 also take more days to fill a backorder for rail inventow than those with a Type 5 organization. 56

OCR for page 49
Days to fill bus inventor backorder: Type 3 -- 21.2 days Days to fill rail inventory backorder: Not Type 5 -- 40.7 days . . . Type 2 -- 5.5 days Type 5 -- 7.8 days (99% confidence) (99/O confidence) Type 5 -- 10.0 days (89% confidence) Agencies with organization Type 5 have a higher percent of obsolete bus inventory than those with organization 1 or 3. Agencies with organization Type 5 also have a higher percent of obsolete raid inventory than those without a Type 5 organization. Percent obsolete bus inventory: Type 5 -- 12.9% Percent obsolete rail inventory: Type 5 -- 8.5% Type 1 -- 6.2% Type 3 -- 5.8% (96% confidence) (98% confidence) Not Type 5 -- 1.3% (96% confidence) Agencies with no rail and organization Type 3 manage more inventory dollars per person than those with Type 1, 2, or 5. In addition, those with Type 4 manage more than those with Type I. Type 3 -- $135,664 Type 4 -- $106,869 Type 1 -- $57,51 1 Type 2 -- $84,570 Type 5 -- $88,569 Type 1 -- $57,51 1 (99% confidence) (94/0 confidence) (94/0 confidence) (92% confidence) Agencies with organization Types 1 or 2 process fewer inventory transactions per person per week than those with Types 3, 4, or S. Type 1 -- 33.2 Type2--61.2 Type 3 -- 130.9 Type 4 -- 183.7 Type 5 -- 206.8 Type 3 -- 130.9 Type 4 -- 183.7 Type 5 -- 206.8 57 (99% confidence) (97% confidence) (99% confidence) (96% confidence) (94% confidence) (99% confidence)

OCR for page 49
Note: Although all differences between organizational types were not significant at or above the 85% confidence level, as organization type progresses from 1 to 5, the number of inventory transactions processed per person per week increases. Full-time head of group-whether the top inventory management person was dedicated fur-time to inventory management responsibilities or had additional non-inventory related responsibilities. Agencies with a filll time head of the inventory management group: have higher annual bus inventory turnover-2.02 vs. 1.37 take less days to fill bus inventory backorders-13.2 vs. 20.8 process more transactions/week per person ~ 171.8 vs. 97.8 manage more inventory dollars/person-$198,730 vs. $95,462 However, these agencies also: . (93% confidence) (8S% confidence) (96% confidence) (99% confidence) have a higher percentage of obsolete bus inventory - 9.4% vs. 4.3% (99% confidence) have a higher percentage of items out of balance - 7.3% vs. 4.4% (92% confidence) Separate planning function-whether the inventory management organization included personnel dedicated to inventory planning functions with no responsibility for storekeeping or other inventory functions. Agencies with a separate planning Unction: have higher annual rail inventor turnover-0.79 vs. 0.50 (85% confidence) have higher Inventor ! rates-94.0% vs. 90.5/0 (88% confidence) manage more inventory dolIars/person -- $244,233 vs. $131,296 (97% confidence) However, these agencies also: . have higher bus Inventory dolIars/vehicle-$7,106 vs. $4,484 (99% confidence) have a higher percentage of obsolete bus inventory-12.5% vs. 5.6% (98% confidence) have a higher percentage of obsolete rail inventory-9.2% vs. 2.3% (90% confidence) Inclusion of purchasing-whether personnel responsible for executing the purchase of inventory matenal were Included In the same organization as inventory management personnel. Agencies that include purchasing in the same organization as inventory management: have lower bus inventory dolDars/vehicle ~ $4,230 vs. $5,424 (94% confidence) have a lower percentage of stockouts-0.33% vs. 2.32% (95% confidence) Centralized versus decentralized -- whether the inventory organization was centrally managed or whether multiple inventory organizations existed within the transit agency. 58

OCR for page 49
Centralized inventor management organizations manage more inventory dollars per ~~~~ ~- (96% confidence). person than decentralized organizations-5159,070 vs. 511(),4Z7 Department reporting to-this attribute was examined two ways: (~) whether or not the inventory organization reports directly to the maintenance department (yes or no), and (2) the specific department to which the inventory management organization reports (e.g. maintenance, operations, a~ninistration, finance, etc.~. Agencies that report to the maintenance department have the following findings versus those Cat do not report to maintenance: lower bus inventory dolIa~/vehicle-$4,535 vs. $5,756 lower percentage of obsolete bus inventory-6.0% vs. 9.~% lower inventory personnel costh~nvento~y dodar-$0.25 vs. $0.38 However, these agencies also: take more days to fib bus inventory backorders-20.9 VS. 9.0 manage fewer Inventor dolIars/Derson-$124 457 vs. $193 593 process fewer transactionsJweek per person-~ 14.0 VS. 165.5 , ~_ , (86% confidence) (92% confidence) (94% confidence) (99% confidence) (94% confidence) (SS/O confidence) En addition, the foDow~ng findings relate to the specific departments responsible for inventory management: . . Agencies with inventory management reporting to the finance department have a higher bus inventory fill rate than those reporting to maintenance-96.0% versus 86.0% (99% confidence). Agencies with inventory management reporting to the finance department process fewer transactions per week per person than those reporting to all other departments: Reporting to finance: 73.3 reporting to maintenance - I 14.0 (82% confidence) reporting to purchasing-223.5 (88% confidence) reporting to materials-1S5.6 (90% confidence) reporting to administration - 164.1 (90/O confidence) Level reporting to-the level within the transit agency to which the inventory organization reports (e.g. board, executive, department, or sub-department). . Agencies with inventory organizations reporting to the sub-department level have lower bus inventory dollars per vehicle than those reporting to the department level or executive level: 59

OCR for page 49
5.4.7. Percent of Items Stocked Out per Week There were two recession equations that can be used to predict the percent of items stocked out per week. En the first equation, two factors were selected as significant, however there is only a 43% confidence that the constant is not equal to zero. In the second equation, only one factor was selected as significant. Factor Coefficient Confidence Level Constant 0.36 43/0 Total employees involved with inventory management. 0.024 91% Organization type2 6.31 99~/0 Coefficient of determination bray: 28.~% Adjusted A: 25.0% Recession confidence level: 99+% Factor Coefficient Confidence Level Constant 0.94 91/0 Orgariization type 2 5.78 99+/O Coefficient of determination brig: Adjusted r2: Recession confidence level: 23.0% 20.9% 99+o/o 5.4.~. Average Days to Fill Bus Inventory Backorders Only one factor was selected to predict the average days to fill bus inventory backorders. This was the only factor with over 90% confidence that the coefficient was not equal to zero. The next closest factor was the use of an automated inventor system with only 67% confidence that the coefficient was not equal to zero. Factor l Coefficient Confidence Level Constant 24.8 99+% Use maintenance forecasts for replenishment -I I.2 96% Coefficient of determination (rib: Adjusted A: Recession confidence level: X.5% 6.6% 96% 74

OCR for page 49
5.4.9. Average Days to Fill Rail Inventory Backorders Only one factor was selected to predict the average days to fill rail inventory backorders. This was the only factor with over 90% confidence that the coefficient was not equal to zero. The next closest factor was whether the agency set target service levels with 83% confidence that the coefficient was not equal to zero. Factor Constant % coverage of storehouses by storekeeper Coefficient of determination arch: Adjusted r2: Regression confidence level: 5.4.10. Percent of Items Out of Balance 59.6/o 49.S% 93o/o Coefficient Confidence Level 82.8 -0.71 97o/o 93o/o The best regression equation for predicting the percent of items out of balance has a 68% confidence that the constant is not equal to zero. To derive an equation with a confidence level for the constant that is greater than 90%, all factors except "inventory orgaruzation reports to maintenance" must be eliminated Dom the equation. Factor Constant Inventory organization reports to maintenance Have multiple storehouses Have buses Total number of replenishment methods used Stock components in inventory Set target inventory levels Coefficient of determination bred: Adjusted r2: Regression confidence level: 5.4.~. Percent Obsolete Bus Inventory 23.7% 13.oo/o 94o/o 9.03 -6.51 10.93 14.97 -2.35 -9.60 -5.48 The factors selected to predict percent obsolete bus inventory were: 75 Coefficient Confidence Level 68% 96% 99o/o 94o/o 94o/o 94o/o 91%

OCR for page 49
Factor Coefficient Confidence Level Constant 5.91 gg+o/O Decentralized orgaruzation 54.5 p9+o/O Use"other'' technology 21.8 99+/0 Total bus inventory dollars 0.0000013 99/0 Total employees involved with inventory mat. -0.22 ~97/O Organization type 5 8.5 99/O Total bus material purchases -0.00000022 99/O Orgariization type 4 a. ~ggo/O Coefficient of determination arty: Adjusted r2: Regression confidence level: 5.4.12. Percent Obsolete Rail Inventory 86.7% 83.5o/o 99+o/o Only two factors were selected to predict the percent obsolete rail Inventor. These were the only factors with over 90% confidence that the coefficients were not equal to zero. The next closest factor was the amount of infonnation (number of reports/screens) available in the storehouse with only 58% confidence that the coefficient was not equad to zero. Factor Coefficient Confidence Level Constant 3.10 98% Total rail Inventory dollars 0.00000047 99+% Use "other'' technology -3.9S 94% Coefficient of determination arty: 91.9/0 Adjusted r2: 89.2% Regression confidence level: 99+/0 S.4.13. Total Inventory Dollars per Person The factors selected to predict total inventory dollars per person were: Factor Coefficient Confidence Level Constant 9~075 gg+o/O Have written policies and procedures 97274 99% Have rail vehicles 257481 99+% Use fixed period replen~shrnent method 210105 99+% 76

OCR for page 49
Set target inventory levels -74418 96% Coefficient of determination (r21: Adjusted r2: Regression confidence level: 57.5o/o 54.9o/o 99+o/o 5.4.14. Inventory Personnel Cost per Inventory Dollar The factors selected to predict inventory personnel cost per inventory dollar were: Factor Coefficient Confidence Level Constant 0.13 99% Average feet age (years) 0.00043 99~% Use manual inventory system 0.88 99+% % coverage of storehouses by storekeeper 0.0028 99+% Organization typed 0.13 95% Separate inventory planning -0.085 90% Use fixeUpenod method 0.18 98% Use blanket POs for inventory -.085 93% Coefficient of determination bras: Adjusted A: Regression confidence level: 78.4% 73.2% 99+o/o S.4.~. Total Inventory Transactions per Person No regression equation that could be denved to predict the total inventory transactions per week per person that contains a coefficient with a confidence greater than 90%. The highest confidence was for the factor "organization type 4" at 65%. As a result, 65% was also the highest confidence for the entire regression equation. 5~5 CONCLUSIONS The conclusions presented in this section are general conclusions implied by the findings outlines in the previous sections. Conclusions are Toupee by the organizational and inventory management decisions examined in this study. Conclusions regarding organizational structure: Organization type 5 provides some service level benefits over other organization types, such as higher fill rates and fewer days to fill back orders. However, these gains appear 77

OCR for page 49
. . . to be at the expense of higher inventory levels (e.g. dollars per vehicle) leading to highe percent of obsolete items. This is an illustration of the traditional trade-offs between the conDicting goals of increasing service level while lowering Inventory investment. The higher turnover experienced by organization type 5 is a by-product of the higher service levels, and appears to imply that service level increases more than compensate for the higher inventory levels. The more evolved inventory organization structures appear to process inventory transactions more efficiently. However, this transaction efficiency does not necessarily translate into efficiency In managing inventory dollars. In general, the organization types represent an evolution from no formal inventory organization (type 1) to a dedicated materials management group at the department level (type 5). The average number of inventory transactions per person increases from type 1 through type 5 (although not all differences have high confidence levels). However, the inventory dollars managed per person does not follow this pattern. Organization types 3 and 4 manage more dollars per person than the other types. Other differences between organizational type were isolated and inconclusive. For example, agencies with type 4 organization have a lower stockout rate than those with type 5. However, this was not a consistent pattern and no significant differences in stockout rate were found between any other combination of organization types. The existence of a full-time head and a separate inventory planning function appear to be key organizational attributes leading to the service level gains of inventory organizations. In addition, these attributes appear to contribute to more efficient management of inventory dollars. Inventory organizations with these attributes take less time to fill backorders and have a higher fill rate, respectively, and organizations with either or both attributes have higher annual tumover. These attributes appear to allow organizations to manage more inventory transactions and dollars per person. These attributes are also consistent with the higher inventory levels and percent of obsolete items observed in organizations with higher service performance. Including purchasing in the same organization as inventory management appears to lead to lower inventory levels and stockout rates. A centralized organization manages significantly more dollars per person than a decentralized organization. 78

OCR for page 49
. Inventory organizations that report to the maintenance department have better performance relating to inventory levels, but manage inventor th lower efficiency. These organizations have lower inventory dollars per vehicle and a lower percent obsolete items. However, inventory organizations reporting to maintenance manage less inventor dolBars per person and process fewer inventory transactions per person. Agencies with Inventory orgaruzations reporting to the finance department have a higher service level than those reporting to maintenance, but process fewer transactions per person than those reporting to any other department. In general, the higher in the agency that the Inventory orgariizations reports, the faster the orgaruzation responds to user needs. Organizations reporting to the executive level take significantly fewer days to fill backorders than those that report to the departmental level, and those reporting to the departmental level take significantly fewer days than those reporting to the sum department level. Sub-departmental orgaruzations have lower Inventory dollars per vehicle and a higher turnover rate than other orgaruzations. However, these organizations manage significantly fewer inventory dollars per person than orgar~ations reporting to higher levels. Conclusions regarding setline goals: Agencies that set target service levels achieve a higher level of service than organizations that do not set target seance levels. Inventory organ zations that set target service levels have a higher fig rate and higher annual inventory turnover. These agencies also manage more inventor dollars per person and have a higher percent obsolete inventory. Agencies that set target inventory levels have higher annual inventory turnover, but do not achieve lower inventory levels. There is no significant difference in inventory levels between those orgariizations that set target inventory levels and those that do not. Conclusions regarding replenishment methods: . No replenishment method or combination of replenishment methods resulted in higher inventory service levels, however the use of maintenance forecasts only appears to lead to lower inventory levels. 79

OCR for page 49
. The lower inventory levels for organizations using only maintenance forecasts are accompanied by lower inventory turnover, a lower percent obsolete items, and fewer Inventory dolDars managed per person. Several detrimental effects on inventory perfonnance are associated with having a safety stock percentage of 20% or greater, Including a higher percent of stockouts. Organizations with 20% or greater safety stock have higher inventory levels and lower turnover for rail inventory, take more tune to fill backorders, and have a higher percent obsolete items. In particular, since these orgaruzations have a higher percent stockouts, the high safety stock levels do not accomplish the primacy goal of safety stock (to reduce stockouts). Conclusions regarding E~urchasine practices: . The use of blanket orders and direct purchasing authority to replenish inventory material ho minimal effect on inventory performance. There was no general pattern on which to draw conclusions. Only isolated effects were observed relating to these variables. Using blanket orders resulted in higher inventory dollars per vehicle for both bus and raid inventones, and the more rail inventory material covered by blanket orders, the lower the service level (fill rate from inventory. Organizations using direct purchasing authority manage less inventory dollars per person, although the opposite is true for those using blanket orders. Conclusions regarding storehouse configuration and coverage: . . There was no storehouse configuration that consistently resulted In higher inventory performance, only a Sew isolated effects. Agencies with a hybrid configuration (a mixture of those tested in the surveys have a higher inventory fill rate, and independent storehouses have lower stockout rates. A central storehouse with satellites manages more inventor dollars per person, but also has a higher percent of items out of balance. Whether a storehouse was secured or not had no significant effect on inventory accuracy. Again, there were only a few isolated effects on inventory perfonnance relating to secured storehouses. Agencies with secured storehouses have higher inventory turnover, and those with less than 100/0 (but more than 0/0) manage more inventory doldars per person. Unmanned storehouses contribute to lower leads to higher inventory levels. 80 inventory service levels, but 100% coverage

OCR for page 49
The two ends of the spectrum, no coverage and 100% coverage, adversely effect inventory performance in one of the major conflicting goals of inventory management. Storehouses with no storekeepers contribute to lower inventory fib rates and storehouses that are 100% covered contribute to higher inventory dollars per vehicle, respectively. Storehouses with 75% or greater coverage (but less than 100%), have the highest annual inventory turnover. Storehouses run by maintenance are run less efficiently than those not run by maintenance. Maintenance run storehouses process fewer transactions per person and manage fewer inventory doDars per person. These storehouses also take longer to fib inventory backorders, but have a lower percent obsolete items. Conclusions regarding published procedures and catalog: . Written policies and procedures have minimal eject on inventory performance, but assist in the efficiency of managing inventory. Written policies and procedures and a published inventory material catalog both allow organizations to manage more inventory dollars per person, but contribute to a higher percent obsolete items. Conclusions regarding Physical and cycle counting: . . Conducting a complete physical inventory has no effect on inventory accuracy, regardless of the frequency. The percent of Steno out of balance was the statistically the same for agencies that conducted annual and semiannual physical inventories, as well as those that never conducted complete physical inventory. Quarterly cycle counting appears to result in significantly higher inventory accuracy than other frequencies. Those agencies that conducted quarterly cycle counts had a lower percent of items out of balance than those counting at other frequencies (daily, weeldy, monthly). Conclusions regarding repairing and stocking components: Agencies that repair and rebuild components are able to absorb the additional inventory activity and dollars with increased staff responsibility. Agencies that stock repaired and rebuilt components have significantly higher inventory levels than those that do not. The additional inventory transactions and inventory dollars 81

OCR for page 49
resulting from components lead to processing more inventory transactions and managing more inventory dollars per person rather than to higher staffing levels. Conclusions regarding use of systems and technology: . . Summary: Automated inventory systems and other technology allow orgaruzations to operate more efficiently and turnover inventory more open, but manual systems contribute to lower inventory levels. Agencies with automated inventory systems and/or "other'' technology process more inventory transactions per person, manage more inventory dollars per person, and have higher turnover for both bus and rail inventory. Agencies with manual system have lower inventory dollars per vehicle, and those with both have higher percent of items out of balance. Bar code technologr contributes to higher inventory seance levels. Agencies using bar code technology for inventory have higher inventory fig rates than those that do not use bar code. In general, no inventory orgaruzation factor, decision factor, agency or fleet characteristic has a comprehensive and consistent effect on inventor performance. Those factors that tend to favor higher service levels often do so at the expense of higher inventor levels and percent obsolete items. In addition, none of the factors affect more than one service level perfonnance indicator (e.g. inventory fill rates, days to fig back orders). Those that eject the efficiency of managing inventory seldom have more than minimal ejects on performance factors relating to inventor investment or service level. This mixture of ejects is further supported by the regression analysis and resulting equations. In 15 equations (one for each inventory performance indicators, 32 separate vanables were selected as significant. Four of these vanables appear in three equations, and one appears In five equations. The other 27 vanables appear In two equations at most. The variable that does appear In five equations, (percent of storehouses covered by a storekeeper) has a positive eject In three equations and a negative effect In two. The survey process yields valuable infonnation regarding the inter-relationships between Inventory management and organizational decision factors. However, the conclusions based on this information are preliminary at best. There are some inherent shortcomings in any survey process and the related statistical analysis. Most of the statistical tests applied during this study test a pair or group of factors and assume that "other things are equal". In addition, there was no way to verify the accuracy of survey responses beyond the application of common sense tests to the range of response values, comparing related responses in difference sections of the survey, and comparing an individual response to the nonnal range of responses. Finally, a survey cannot effectively capture information regarding the competence and attitude of the individuals actually involved with inventory management 82

OCR for page 49
at transit agencies. Therefore, the conclusions should be developed into hypotheses for more detailed testing in a more controlled environment where individual effects can be more effectively isolated and quantified. 83

OCR for page 49