Printer Friendly

Measuring and improving productivity: a new quantitative approach.

A significant number of books, articles, and papers have been written analyzing American productivity. Virtually every author admits that there are many disadvantages in American investment policy, technological improvement, quality control systems, organizational behavior, and structural organization of the economy. To solve this problem a great number of productivity improvement programs have been introduced and utilized in the U.S. Many of these are based on technological modernization, long-term investment policy, organizational improvements, etc. However, the results in manufacturing have not met management's expectations. This means that new, more innovated approaches are needed for substantial productivity increase. All productivity improvement projects should be based on a detailed analysis of the existing productivity movement. During this analysis several questions should be answered:

a. What are the factors that affect productivity?

b. How do we measure these factors?

c. What is the appropriate approach to analyze these factors?

d. What is the model that can be used for predicting productivity growth while varying the factors' values?

We attempted to analyze productivity in connection with losses which occurred during the production process.

Productivity and losses

Loss of labor and material resources (sometimes called waste) are a part of the production process. They can arise in any stage of the process and are due either to production necessities, or to disadvantages in the production process.

Variation of loss values affects productivity change. The standard definition of productivity is the ratio of output to input. Loss represents the input along with the productive expenses of material, labor, energy, and financial resources involved in the production process. The main difference between these two groups of input is that losses add only cost to input but not the value to product output. Decreasing the value of losses in input will contribute significantly to productivity growth.

Some manufacturing firms base their productivity improvement programs on the analysis of losses in the production process. Among the best examples of this kind of analysis is Toyota Motors Co. This company is one of the leaders in developing new approaches in production efficiency. Its program to improve productivity has been based on a critical review of company wastes (Bell, 1991). Among wastes analyzed are overproduction, idle time of workers and equipment, transportation delays, poor quality, etc. This analysis was used by the company to evaluate and improve the combination of equipment and processes to make the company more productive. Alan Lawlor (1985) presented a waste analysis as part of his productivity method study. He also classified the areas of waste and showed their influence on productivity.

Investigating these and other examples of loss-based analysis and productivity improvement, we may conclude that the following: labor and material loss play not only a great role as a productivity analysis base, but they also appear as an active element in productivity growth. However, companies do not use quantitative measurements or quantitative techniques to evaluate the real influence of loss on the productivity movement.

The main idea of this approach is to base productivity improvement on a new measurement system that fully describes the productivity behavior according to loss variation. The system should be able to produce scientifically based recommendations in productivity improvement. The assumptions of this approach are:

a. Achievement of significant productivity improvement without major expenses.

b. Elimination of the loss as a main basis for productivity growth.

c. Consideration of all possible losses.

d. Creation of an appropriate model describing loss influence on productivity.

According to Robert D. Pritchard, any productivity measurement system should produce an overall index of productivity. This system must be valid, which means the completion of the system and accuracy in the things measured. The next criterion is a system's flexibility in order to accommodate the changes in the organization's goals and policies over time. Another desirable feature of a productivity measurement system is the ability to aggregate the measurement systems of different units into a single system, and the ability to compare the productivity of different units. These assumptions generally specify the objectives of the newly-created measurement system that can be developed following these steps:

a. Identify all losses that can affect productivity.

b. Develop indicator(s) to measure each loss.

c. Design the model which fits the assumptions of the productivity measurement system.

d. Estimate productivity changes according to the created model.

e. Derive recommendations on how productivity can be improved by eliminating different losses.

Improvements in productivity

The first step in creating the new measurement system is to identify those sets of losses that may appear in a company, or in its specific unit, during the production process. Our analysis combines all possible losses into three classes (Figure 1). Measurement systems for different companies may use different elements due to specific operations conditions.

The next step requires that a set of indicators for the described measurement system should be developed. An indicator should quantitatively explain losses in the production process. It must cover each loss completely and be meaningful to personnel. Using these criteria, the set of indicators for the specified losses is created (Figure 1). The lower the value of an indicator, the lower the loss level, the greater the productivity index. The same loss can have different indicators depending on the particular production process.

One of the most important steps in the creation of a measurement system is the design of an appropriate model. This model should quantitatively describe productivity movements depending on different loss changes. Several statistical methods can be used to meet the measurement system's assumptions. If a limited number of variables (losses) is evaluated, the multiple regression analysis is the more relevant technique to describe productivity behavior. However, if a greater number of variables be included in the model, a close relationship and correlation between different variables should be expected. These features make factor analysis an appropriate technique for developing a productivity model.

Factor analysis is also used to identify a relatively small number of factors which can represent relationships among sets of many interrelated variables. The final mathematical model for factor analysis appears close to a multiple regression equation:

P = A1 [multiplied by] F1 + A2 [multiplied] F2 + ... Ak [multiplied by] Fk + B [1]

where F1, F2, ..., Fk = so called common factors,

A1, A2, ..., Ak = coefficients of factors,

k = number of common factors,

B = constant.

Each common factor can be estimated as a combination of several variables included in the set of all variables:

[F.sub.j] = Wj1 [multiplied by] X1 + Wj2 [multiplied by] X2 + ... + Wjm [multiplied by] Xm [2] where Wj1, Wj2, Wjm = factor score coefficients for jth common factor,

m = the number of variables for calculating common factors.

A good factor-model solution is both simple and interpretable. This means that the model should include as low a number of factors as possible, and each factor should be meaningful. Using factor analysis, all possible losses can be considered and their influence on productivity estimated. Equations [I] and [2] fully explain the behavior of productivity and, thus, can be used to prepare recommendations on how to improve productivity by eliminating different losses.

Using the new system

The described approach was used to evaluate changes in productivity for a manufacturing company producing electronic equipment. The company is located in Northern California, near San Francisco. The analysis of the production process showed fourteen losses to be included in the model. Among them are the following losses from Table 1: ineffective technological design (X1), poor automation policy (X2), poor tools supply (X3), ineffective facility layout (X4), unbalanced process line (X5), wrong work standards (X6), ineffective pay system (X7), overstocked level of inventory (X8), poor material requirement (X9), poor scheduling (X10), low level of product quality level(X11', poor quality (X12), poor job satisfaction (X13), ineffective incentives (X14).


Productivity was measured in a number of units of production per unit of time. The measurement of losses was based on their indicators. Among them were idle time of equipment, equipment downtime, number of defectives, average over-material usage, transportation costs between departments, workers' idle time, etc. In order to determine the effectiveness of the pay system and work incentives, as well as the level of job satisfaction, the special surveys were used. Information was gathered for the past three years at one month intervals. Thus, each variable was represented with 36 points. It took about 10 working days to collect and sort input information. All collected data were standardized to be used appropriately in the factor analysis.

Factor analysis usually proceeds in four steps [5]:

a. Calculate the correlation matrix and decide whether to use the factor analysis.

b. Determine common factors by factor extraction and validate how well they fit the data.

c. Transform the factors to make them more interpretable.

d. Compute scores for each factor that can be used to create final productivity model.

The whole factor analysis was done using SPSS statistical software. When the data were collected it took a relatively short time to make the decision.

First, the correlation matrix was computed (Table 2). It shows that almost half of the coefficients are greater than .3 in absolute value. That means they may share common factors and also that the factor model is appropriate. In order to obtain initial common factors we used principal component analysis. It showed that 91.8 percent of the total variance can be explained by the first five variables. The remaining nine factors together account for only 8.2 percent of the variance. Thus, the model with five common factors may be an adequate representation of the data. To identify factors that are substantively meaningful (in the sense that they summarize sets of closely related variables) a rotation of factors is used. We used Varimax rotation to create a rotated factor matrix (Table 3).


To identify the factors, it is necessary to group the variables that have large loadings for the same factors. We sorted the rotated factor matrix so that variables with high loadings on the same factor appear together. Small factor loadings (less than .5) can be omitted from such a table.

The last matrix (Table 3) shows how all variables are distributed among factors. These allow us to identify the factors as following: Fl ineffective technological design, F2 - overstocked inventory, F3 - poor job design, F4 - poor product quality, F5 - wrong work standards. While completing factor analysis we obtained five uncorrelated common factors that have 36 points each. These can be used to estimate productivity variation depending on the different variables changes. The final regression equation is the following:

P = -10.345 F1 -.342 F2 -3.434 F3 -8.151 F4 -.815 F5 +2915.303 [3]
 Sorted Matrix with Varimax Rotation
Varimax converged in five iterations

Rotated Factor Matrix:
X4 .97789
X5 .9259
X2 .90921
X1 .83390
X9 .96042
X3 .88065A
X8 .85912
X10 .82043
X7 .95895
X14 .95027
X13 .92318
X12 .93562
X11 .92912 .94675

According to the equation [3] all common factors have negative coefficient. That means productivity will grow if a common factor decreases. The maximum value of productivity can be reached if all factors (all losses) approach zero.


We developed the new measurement system which allows us to analyze productivity growth that depends on the elimination of different losses. This system is based upon a factor analysis of loss variables, which determines common loss factors, and also upon the multiple regression model of productivity with common factors as independent elements of the model.

A practical sample of an electronic equipment company showed that the main loss factors in terms of productivity improvement are: ineffective technological design, overstocked inventory, poor job design, poor product quality, wrong work standards. Equation [3] explains based upon the common factors the inadequate independence of productivity improvement. The biggest growth in productivity could be expected by decreasing the loss inherent in ineffective technological design (F1), poor product quality (F4), and poor job design (F3). These losses need to be lowered first, in order to achieve greater productivity growth.

It is interesting to point out that some of these common factors are close to the seven waste factors found in the Toyota study. That means the company empirically found the right way to analyze productivity. However, our approach based on scientific analysis of the issues gives proper quantitative results which can be used in productivity improvement.

For further reading

Belcher, J.G. Jr. 1987. Productivity Plus: How Today's Best Run Companies Are Gaining the Competitive Edge. Gulf Publishing Company Book Division, Houston. Bell, R.R. 1991. Managing Productivity and Change. South-Western Publishing Co., Cincinnati, Ohio. Lawlor, A. 1985. Productivity Improvement Manual. Grower Publishing Company Limited, England. McDonald, R.P. 1985. Factor Analysis and Related Methods. Lawrence Erlbaum Associates, Inc., Hillsdale, New Jersey. Norusis M.J. 1990. SPSS/PC + Statistics 4.0 for the IBM PC/XT/AT and PS/2. SPSS Inc., Chicago, Il. Pritchard, R.D. 1990. Measuring and Improving Organizational Productivity: A Practical Guide. Praeger Publishers, New York. Productivity Analysis at the Organizational Level 1981./Edited by Adam, N.R. and A. Dogramaci. Matinus Nijhoff Publishing, Kluwer Boston, Inc., Hingham, Massachusetts. Productivity in Organizations: New Perspectives from Industrial and Organizational Psychology 1988./By Campbell, J.P., J. Campbell, and Associates, Jossey-Bass Inc., Publishers. Scott Sink, D. 1985. Productivity Management: Planning, Measurement and Evaluation, Control and Improvement. John Wiley & Sons, Inc., New York.
COPYRIGHT 1992 Institute of Industrial Engineers, Inc. (IIE)
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Radovilsky, Zinovy D.; Gotcher, J. William
Publication:Industrial Management
Date:May 1, 1992
Previous Article:Getting the most for your public relations dollar.
Next Article:Is JIT really appropriate for American manufacturing?

Related Articles
The role of qualitative information in productivity measurement.
TQM measures: reliability and validity issues.
New directions in productivity analysis.
Operational Risk Gets Its Due.
Cost Structure and the Measurement of Economic Performance: Productivity Growth, Utilization, Cost Economies, and Related Performance Indicators.
Productivity Program Meeting.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters