Companies are faced with the challenge of promoting innovation for productivity improvement among employees. They create a work environment that promotes worker participation for productivity improvement. This sentiment underpins the concept of gainsharing.

This study evaluated the effectiveness of a gainsharing programme for productivity improvement in automotive parts manufacturing companies in South Africa (SA).

SA’s labour productivity, in the manufacturing sector, is low when compared with Korea, the United States of America, Taiwan, Japan, France and the United Kingdom. Hence, this study focused on gainsharing, given the low labour productivity levels in the South African manufacturing industries.

The two automotive parts manufacturing companies that have adopted a gainsharing strategy participated in the study. A third automotive parts manufacturing company that has adopted the 360-degree performance appraisal system was included for comparative purposes. These companies operated in the eThekwini District Municipality in KwaZulu-Natal. Study objectives were achieved by collecting pre- and post-quarterly data for spoilage, absenteeism, capital investment and labour productivity.

Results established that gainsharing improves productivity and reduces spoilage and absenteeism rates.

The South African companies are encouraged to revise their reward philosophies and develop strategies, policies and practices that help achieve productivity goals and support organisational change.

Gainsharing is a desirable alternative as it contributes to raising the competence levels and productivity improvement of an organisation. As a comparison, the 360-degree performance appraisal does not have an impact on labour productivity.

Despite the contraction of economic activity in 2009, the tepid recovery in 2010 and the overall soft labour market conditions, the real wage continued to increase rapidly in SA, outpacing the growth of labour productivity (Klein

This section discusses the overview of a gainsharing programme. It elaborates how gainsharing improves performance.

This section commences by defining a gainsharing programme. Its framework for employee participation is explained, and the experiences of US-based companies using a gainsharing programme conclude it.

According to Armstrong (

The participation structure of gainsharing varies by organisation and tends to grow and evolve over time (Masternak

Roman (

Gainsharing in various sectors of economic activity is increasing (Weiss

Employees under gainsharing reward structures are expected to engage in more cooperative behaviours, including sharing their ideas for saving costs and improving production, more so than employees under more competitive, individual-based compensation systems (Tjosvold

In the USA, Brazil and parts of Europe, gainsharing creates a working environment that encourages worker participation and provides opportunity for linking improved performance to better compensation (Wellbourne & Gomez-Mejia

The study is based on the following assumption:

H1: The implementation of a gainsharing programme leads to labour productivity improvement in the automotive parts manufacturing companies.

Ho: The implementation of a gainsharing programme does not lead to labour productivity improvement in the automotive parts manufacturing companies.

Having presented the main hypothesis for this study,

The sub-hypotheses for the study.

Sub-hypothesis | Code | Description |
---|---|---|

1 | H2 |
An increase in the spoilage rate increases labour productivity in the automotive parts manufacturing companies |

2 | H3 |
An increase in the absenteeism rate increases labour productivity in the automotive parts manufacturing companies |

3 | H4 |
An increase in the allocation of workers in production increases labour productivity in the automotive parts manufacturing companies |

4 | H5 |
The accumulation of past capital investments increases labour productivity in the automotive parts manufacturing companies |

The method for this research will be discussed under the following headings, namely research design and approach, companies that participated in the study, data collection, measurement of data and data analysis.

This study was quantitative in nature. Bryman and Bell (

The study is comparative in nature using quantitative research tools. A convenience sample utilising three large companies that are in the automotive parts industry situated within the eThekwini District Municipality in the Province of KwaZulu-Natal in SA was used. The two companies that adopted gainsharing as an incentive strategy agreed to participate in the study. They are identified as companies A and B. The third company, which adopted a 360-degree performance appraisal system, is identified as company C and is included in the study for comparative purposes. Company A had 1005 employees, whereas company B had 1300 employees, and both operate a three-shift system. Company C uses a 360-degree performance appraisal system and also operates a three-shift system. It is situated in the same municipal area and has 1400 employees.

The collection of data from two automotive parts manufacturing companies (A and B) was carried out in two phases. The first phase involved the collection of pre- and post-gainsharing results for spoilage, absenteeism, capital investment and labour productivity. The pre-gainsharing results were quarterly data reflecting each company’s performance over the 3 years prior to gainsharing implementation. This includes data from the first quarter of 2005 to the final quarter of 2007 (2005Q1–2007Q4). The post-gainsharing data reflect the company’s performance for the period of 3 years after the implementation of gainsharing. This involves data from 2008Q1 to 2010Q4. In the period between the 2011 and 2016, both companies A and B were involved in labour restructuring. This emanated from the strategic changes that took place in the motor assembly plant of the company they are supplying. Hence, the data from the restructuring period were excluded from the study. The next part of the study involved collecting pre- and post-quarterly data for spoilage, absenteeism, capital investment and labour productivity from the other automotive parts manufacturing company (C) that uses a 360-degree performance appraisal system. Company C’s data set ranges from 2004Q1 to 2009Q4. However, they adopted the 360-degree performance appraisal system in 2007Q1. The reason for including company C in the study was to compare gainsharing results with a different incentive system.

The company’s quarterly time series data on labour productivity, spoilage, absenteeism, number of workers involved in production and investment were used. Although companies A and B are relatively similar in nature, the researcher pooled the data from the two companies for an optimum sample size so that statistically valid results could be obtained. The measurements were based on 22 observations (after adjustments) per company. Therefore, the results are based on the total of 44 observations. Similarly, company C’s quarterly time series data on absenteeism, labour productivity and spoilage rates were used. The measurements were based on the total of 24 observations.

Additionally, a dummy variable that assumed the value of 0 and 1 to represent the pre- and post-gainsharing periods, respectively, was introduced into the ordinary least squares (OLS) model. The aim was to isolate the pre- and post-labour productivity effects. Consequently, if gainsharing proved to be a useful strategy in raising labour productivity levels, this would result in a statistically significant coefficient on the dummy variable.

Hence, the favourable findings regarding the co-integrating tests enabled the study to engage in quantitative analysis involving OLS in order to quantify the magnitude of the effectiveness that the implementation of gainsharing has had on labour productivity. Co-integration provides evidence of a long-run relationship between variables (Juselius

The OLS model used was as follows:

The above model assumes that labour productivity is a function of spoilage rate, absenteeism rate, the number of workers involved in production, investment and gainsharing strategy. The investment variable is the labour productivity lagged by 1 period (i.e. 1 quarter). This variable aims to capture previous machinery input (i.e. past capital investment).

For the study to achieve its objective, stationarity tests (as shown in

Augmented Dickey Fuller stationarity test results.

Variables | Company | Level | First difference | Critical values with percentage significant levels | Conclusion |
---|---|---|---|---|---|

Labour productivity | A | −0.929 | −3.952 | −3.831 (1%) | Stationary after 1st differencing |

B | 0.603 | −4.258 | −3.809 (1%) | Stationary after 1st differencing | |

C | −4.345 | - | −3.739 (1%) | Stationary in levels | |

AB | −1.559 | −5.780 | −3.597 (1%) | Stationary after 1st differencing | |

Spoilage rate | A | −3.628 | −6.685 | −3.809 (1%) | Stationary after 1st differencing |

B | −2.844 | −6.817 | −3.809 (1%) | Stationary after 1st differencing | |

C | −3.295 | −3.752 | −3.674 (5%) | Stationary after 1st differencing | |

AB | −5.470 | - | −4.186 (1%) | Stationary in levels | |

Absenteeism | A | −4.731 | - | −3.788 (1%) | Stationary in levels |

B | −4.853 | - | −3.809 (1%) | Stationary in levels | |

C | −5.392 | - | −3.738 (1%) | Stationary in levels | |

AB | −6.573 | - | −4.186 (1%) | Stationary in levels | |

Number of workers | A | −0.875 | −5.387 | −3.809 (1%) | Stationary after 1st differencing |

B | −0.982 | −5.194 | −3.809 (1%) | Stationary after 1st differencing | |

C | −1.719 | −4.837 | −3.752 (1%) | Stationary after 1st differencing | |

AB | −1.663 | −6.325 | −3.597 (1%) | Stationary after 1st differencing |

The data set spanned 2005Q1 to 2010Q4 for companies A and B. Data set for company C spanned from 2004Q1 to 2009Q4.

The stationarity tests for all (except spoilage rate for companies C and AB) were conducted on the assumption of intercept and no trend was used.

Company AB represents the pooled data for companies A and B.

All the critical values are based at the 1% significance level, except for spoilage rate of company C which was based at the 5% significance level.

A battery of other unit root tests (not reported) confirmed the above Augmented Dickey Fuller (ADF) test results.

The results for both companies’ (A and B) pooled data indicate that the variables exhibit mixed orders of integration. If one ran the OLS models that had non-co-integration level variables, this could have resulted in spurious regressions. As a result, the tests in

Johansen trace and maximum eigenvalue statistics for co-integrating vector.

Company | Trace test |
Maximum Eigenvalue test |
||||
---|---|---|---|---|---|---|

No. of hypothesised co-integrating equations | Trace statistic | 5% critical value | No. of hypothesised co-integrating equations | Maxi-Eigen statistic | 5% critical value | |

A | Ho: |
63.78 |
47.86 | Ho: |
34.78 |
27.58 |

Ho: |
29.00 | 29.80 | Ho: |
13.58 | 21.13 | |

B | Ho: |
53.28 |
47.86 | Ho: |
30.57 |
27.58 |

Ho: |
22.70 | 29.80 | Ho: |
14.47 | 21.13 | |

C | Ho: |
58.96 | 63.87 | Ho: |
26.57 | 32.12 |

Ho: |
32.39 | 42.92 | Ho: |
18.05 | 21.13 | |

AB | Ho: |
84.96 |
47.86 | Ho: |
40.76 |
27.58 |

Ho: |
44.20 |
29.80 | Ho: |
27.21 |
21.13 | |

Ho: |
16.98 |
15.50 | Ho: |
16.57 |
14.26 | |

Ho: |
0.412 | 3.841 | Ho: |
0.412 | 3.841 |

An unrestricted vector autoregression (VAR) of lag order two (that is,

Companies AB represent pooled data of companies A and B.

Models of companies A and B were based on the assumption that the level of data and co-integrating equations have linear trends.

Models for company C as well as company AB were based on the assumption that the level of data has co-integrating equations and linear trends.

, and **, denotes that the statistics under consideration are significant at the 1% and 5% significance levels, respectively.

The above tests show that the variables of companies A and B have a co-integrating relationship. This reflects that there is more than one co-integrating relationship in the pooled data of companies A and B (also shown as AB) in

Labour productivity data for spoilage and absenteeism rates, number of workers in production and dummy variables.

Regression | Coefficient | Probability | |
---|---|---|---|

Constant (Bo) | −8.682901 | −3.763349 | 0.001500 |

Spoilage rate | −0.055950 | −1.326403 | 0.202300 |

Absenteeism rate | 0.001439 | 0.044748 | 0.964800 |

Number of workers | 1.934386 | 5.700135 | 0.000000 |

Gainsharing dummy | 0.178659 | 4.006333 | 0.000900 |

0.940099 | 66.70077 | ||

Adjusted |
0.926005 | Prob ( |
0.000000 |

S.E. of regression | 0.061079 | Mean dependent var. | 4.528331 |

S.D. dependent var. | 0.224537 | Durbin–Watson stat. | 1.196897 |

Constant (Bo) | 13.63182 | 3.444978 | 0.003100 |

Spoilage rate | 0.092150 | 0.850904 | 0.406600 |

Absenteeism rate | 0.107928 | 1.681369 | 0.111000 |

Number of workers | −1.369704 | −2.415123 | 0.027300 |

Gainsharing dummy | 0.616282 | 6.579194 | 0.000000 |

0.8256680 | 20.13048 | ||

Adjusted |
0.784664 | Prob ( |
0.000003 |

S.E. of regression | 0.100087 | Mean dependent var. | 4.597399 |

S.D. dependent var. | 0.215684 | Durbin–Watson stat. | 1.945990 |

Constant (Bo) | −43.89323 | −1.566675 | 0.132100 |

Spoilage rate | 0.561978 | 3.61252 | 0.001600 |

Absenteeism rate | 0.206197 | 1.25233 | 0.224200 |

Number of workers | 6.749395 | 1.733268 | 0.097700 |

360˚ dummy | −0.00692 | −0.029883 | 0.976400 |

0.453045 | 4.348594 | ||

Adjusted |
0.348863 | Prob ( |
0.010202 |

S.E. of regression | 0.301106 | Mean dependent var. | 5.748966 |

S.D. dependent var. | 0.373150 | Durbin–Watson stat. | 1.320818 |

Model 1, Company A regression notes: The OLS estimation is based on the equation: Productivity = Bo + B1 Past capital investment + B2 Spoilage + B3 Absenteeism + B4 Number of workers + B5 Gainsharing dummy. Regression data: 2005Q1–2010Q4. 22 observations after adjustment.

Model 2, Company B regression notes: The OLS estimation is based on the equation: Productivity = Bo + B1 Past capital investment + B2 Spoilage + B3 Absenteeism + B4 Number of workers + B5 Gainsharing dummy. Regression data: 2005Q1–2010Q4. 22 observations after adjustment.

Model 3, Company C regression: Productivity = Bo + B1 Past capital investment + B2 Spoilage + B3 Absenteeism + B4 Number of workers + B5 Post-360° dummy. Regression data: 2004Q1–2009Q4. 24 observations.

OLS, ordinary least squares, S.E., standard error, S.D., standard deviation, Prob, probability, stat., statistics, var., variable.

Illustrates labour productivity data as a dependent variable to past capital investment (lagged by 1 quarter).

Regression | Coefficient | Probability | |
---|---|---|---|

Constant (Bo) | −1.387463 | −0.784840 | 0.444800 |

Past capital investment (lagged by 1 quarter) | 0.810535 | 6.547857 | 0.000000 |

Spoilage rate | 0.041050 | 1.499065 | 0.154600 |

Absenteeism rate | −0.002225 | −0.127798 | 0.900000 |

Number of workers | 0.323907 | 1.010184 | 0.328400 |

Gainsharing dummy | 0.079908 | 2.827210 | 0.012700 |

0.983569 | 179.5779 | ||

Adjusted |
0.978092 | Prob ( |
0.000000 |

S.E. of regression | 0.032805 | Mean dependent var. | 4.541183 |

S.D. dependent var. | 0.221635 | Durbin–Watson stat. | 2.354170 |

Constant (Bo) | 9.308245 | 1.886486 | 0.078700 |

Past capital investment | 0.215905 | 1.185821 | 0.254100 |

Spoilage rate | 0.075752 | 0.697804 | 0.496000 |

Absenteeism rate | 0.086224 | 1.310002 | 0.209900 |

Number of workers | −0.878498 | −1.332745 | 0.202500 |

Gainsharing dummy | 0.474492 | 3.426895 | 0.003700 |

0.846551 | 16.550470 | ||

Adjusted |
0.795401 | Prob ( |
0.000012 |

S.E. of regression | 0.099707 | Mean dependent var. | 4.594075 |

S.D. dependent var. | 0.220432 | Durbin–Watson stat. | 2.273307 |

Constant (Bo) | 12.98467 | 0.491943 | 0.628400 |

Past capital investment | 0.340728 | 2.328527 | 0.031100 |

Spoilage rate | 0.490514 | 3.809934 | 0.001200 |

Absenteeism rate | 0.214852 | 1.669034 | 0.111500 |

Number of workers | −1.405156 | −0.377915 | 0.709700 |

360° dummy | 0.297861 | 1.434140 | 0.167800 |

0.522472 | 4.157645 | ||

Adjusted |
0.396806 | Prob ( |
0.010139 |

S.E. of regression | 0.234600 | Mean dependent var. | 5.793536 |

S.D. dependent var. | 0.302065 | Durbin–Watson stat. | 2.046693 |

Model 1, Company A regression: The OLS estimation is based on the equation: Productivity = Bo + B1 Absenteeism + B2 Spoilage + B3 Workers + B4 Post-gainsharing dummy. Regression data: 2005Q1–2010Q4. 22 observations after adjustment.

Model 2, Company B regression: The OLS estimation is based on the equation: Productivity = Bo + B1 Absenteeism + B2 Spoilage + B3 Workers + B4 Post-gainsharing dummy. Regression data: 2005Q1–2010Q4. 22 observations after adjustment.

Model 3, Company C regression: The OLS estimation is based on the equation: Productivity = Bo + B1 Absenteeism + B2 Spoilage + B3 Workers + B4 Post-360° dummy. Regression data: 2004Q1–2009Q4. 24 observations.

OLS, ordinary least squares, S.E., standard error, S.D., standard deviation, Prob, probability, stat., statistics, var., variable

Results for companies A and B in

Results for the three companies as illustrated in

Results for both companies (A and B) in

Results for both companies (A and B) in

Results of

Results for both companies (A and C) in

Results for both companies (A and B) in

Results for companies A, B and C in

Results for companies A, B and C in

Results for both companies (A and B) in

In addition,

This section analyses data using factorial designs. It incorporates box plots to determine whether the factorial ANOVA assumptions of normality and homogeneity of variance have been met. Porkess (

Box plots determining the normality and homogeneity of variance.

_{mean} for company A (labelled as company 1) and company B (labelled as company 2). Although the D_{means} are similar, the variances of company C (labelled as company 3) are much larger when compared to company A and company B. In all cases, the D_{means} for the three companies are normally distributed.

_{means} for companies A, B and C (labelled as 1, 2 and 3, respectively).

Estimated marginal means of D_{means} for companies A, B and C (labelled as 1, 2 and 3, respectively).

_{means}. This means that policy interactions that took place between companies A and B are significant in relation to company C. The effect of marginal means of the policy variable at companies A and B is relatively different to the marginal means of company C. The mean difference is partly based on the fact that the method of operations, the product classification and policy types (i.e. gainsharing system) for companies A and B are almost similar to each other, whereas company C uses the 360-degree performance appraisal system. Thus, there are no mean changes for company C when comparing the mean changes for both companies (A and B) in relation to pre- and post-policy periods.

Therefore, the results indicate that the implementation of gainsharing improved labour productivity for companies A and B. In contrast, they indicate that the implementation of 360-degree performance appraisal does not improve labour productivity, as measured in company C. The results are confirmed by study analysis from sections ‘Labour productivity as a dependent variable to dummy variables’.

The study results indicate that absenteeism and the number of workers involved in production have no relation to labour productivity for companies A, B and C. However, spoilage rate has a relationship with labour productivity for company C. In addition, there is no relationship between labour productivity and the 360-degree performance appraisal system. The study revealed the relationship between gainsharing programme and labour productivity. The positive relationship indicates that the implementation of the gainsharing programme does increase labour productivity. This is supported by Fourie (

The study also provides the comparative results of a gainsharing programme and the 360-degree performance appraisal system to labour productivity (1 quarter after the three companies have invested in capital). The results show that absenteeism and the number of employees involved in production have a relation with both the gainsharing programmes and the 360-degree performance appraisal system (1 quarter after the companies have invested in capital). In addition, there is no relation between spoilage rate and labour productivity (for both companies that have adopted gainsharing). Similarly, the results showed no relations between the 360-degree dummy variable and labour productivity. In contrast the gainsharing variable as well as capital investment (for the three companies) has a positive relation with labour productivity (1 quarter after the companies have adopted the systems). The relationship indicates that the implementation of a gainsharing programme and capital investment increase labour productivity. Hence, Rondeau (

The South African companies are encouraged to revise their reward philosophies and develop strategies, policies and practices that help to achieve new business goals and support organisational and cultural change. This must be based on an understanding of the economic factors affecting pay, the significance of the psychological contract and the practical implications of motivation theory as it affects the provision of both financial and non-financial rewards. Besides the achievement of study objectives, the following conclusions can be made:

The implementation of the gainsharing programme increases labour productivity in the automotive parts manufacturing companies in SA. As a comparison, the 360-degree performance appraisal does not have an impact on labour productivity.

Capital investment plays a role in labour productivity improvement. Companies that have invested capital (after the implementation of a gainsharing programme) experience an improvement in labour productivity.

The study was limited to the automotive parts manufacturing industry within the eThekwini District Municipality. The investigation was conducted in two companies that have adopted gainsharing and a single one that has adopted the 360-degree performance appraisal system. As there are 378 registered automotive parts manufacturers in SA (SAinfo

In order to maximise performance, a comprehensive performance policy must be developed, which aligns pay (and other incentives) to performance. Gainsharing creates a working environment that encourages worker participation and provides opportunity for linking improved labour productivity to compensation. However, it is not a quick fix for inherent problems. Hence, the results indicate that there is no relationship between both absenteeism and the number of workers in production and labour productivity under a gainsharing programme.

The nature of this study did not allow the investigation to determine the long-term labour productivity survival to a wider sector of economic activity. It is recommended that future studies should examine the following issues in greater depth:

when to use and when not to use a gainsharing programme or 360-degree performance appraisal system

the applicability of gainsharing to other industrial sectors

a more comprehensive investigation should be carried out using a randomised sample of the registered automotive component manufacturers that use gainsharing to see if the results can be generalised.

This article is based on the thesis (A comparative investigation into the applicability of gainsharing programmes for the improvement of productivity in the automotive sector of South Africa) presented to the Faculty of Commerce, Law and Public Administration at the University of Zululand by R.W.D.Z. It is submitted in fulfilment for the degree of Doctor of Commerce (D.Com). The author acknowledges Mr A. Moorley, Dr M. de Beer, Prof. K. Nel, Prof. E. Contogiannis and Dr I. Kaseeram for their guidance and support. He also thanks the companies for granting permission to conduct the study.

The author declares that he has no financial or personal relationships that may have inappropriately influenced him in writing this article.