Inclusive Manufacturing: An Analysis of Use of Employees with Disability and their Impact on Productivity
Sriram Narayanan (), Tharo Soun (), Kalyanmoy Deb()
Michigan State University, East Lansing, 48824
There is significant amount of interest in workplace to create a more inclusive and diverse environment. In particular, employers have focused on providing more opportunities to people with disabilities. Public policy initiatives such as the Occupational Safety and Health Act (OSHA); and the Americans with Disabilities Act (ADA) of 1990 push employers in creating a more inclusive, and safe, work environment. The ADA prevents employers from discriminating employees based on their disabilities. Several companies – particularly in services sector – such as AMC Theaters, Home Depot, Microsoft are employing American workers with Physical or Cognitive Disabilities (Cann 2012). Despite this, only an estimated 20% of workers with these conditions participate in the workforce and 14% of this group is unemployed (Cann 2012). To provide more opportunities for people with disabilities, the department of Labor mandates that firms will have to employ at least 7% of their workers with disabilities should they bid for federal contracts (Associated Press 2013). While these are excellent measures for improving the employability of individuals, very few studies have so far taken a systematic view of how the employment of people with disabilities influences productivity in shop-floor operations. In this study, we examine the impact of use of workers with disabilities and their influence on manufacturing productivity. The context is an apparel manufacturing cell. Using a simulation approach, we perform multi-criteria optimization that not only maximized productivity, but also disability diversity. We examine the pareto-frontier that trades-off the degree to which disabled people are employed in the team and productivity. Our research answers the following questions:
- What is the effect of diversity (of disability and languages) of a team on its productivity?
- Is there any differential effect of different diversity factors (diversity of employees with varying types of disabilities, and different language backgrounds) in influencing productivity?
- How can the knowledge gleaned from answering questions 1 and 2 be systematized to develop heuristic rules for team assignment problems in maximizing productivity of teams that have individuals with disabilities?
Our research is in partnership with a local (not-for-profit) apparel manufacturing firm, and was funded by them.The firm provided access to its manufacturing cells and data. The organization (700 employees at least 70% of which have some form of significant disabilities) prides itself in providing opportunities for individuals with significant disabilities in their quest for independence and self-sufficiency. The results presented in this extended abstract are for one garment. We are in the process of checking replication to multiple garments.
Approach
Data for this study is related to running shorts. The production of the apparel needed 23 steps. The team assignment was performed using the time study data obtained from a historical (time study)records of the firm. The data is collected from employees with different disability and diversity classes. The nature of disabilities among employees range from Physical (deaf, hard of hearing, or visual impairment), Mental (anxiety, Schizophrenia), Cognitive disabilities, Learning disabilities and short term disabilities (Substance abuse, emotional impairment). Our data also contain substantial fraction of individuals with no disabilities.In addition, employees come from 22 different language backgrounds (diversity) grouped intofive world regions. The time study data stored the actual time (in min) spent by an employee coming from a particular diversity group in completing a categorized complexity task from on-the-job assembly lines for various tasks was used to estimate mean and standard deviation of times. This time study data was scaled to the particular process used for the garment. Our methodology was to solve the following multi-objective optimization problem to examine a set of trade-off team assignments:
,
,
subject to
To handle conflicting nature of the two objectives and complexities involved in evaluating an assignment (team), we used a customized evolutionary multi-objective optimization (EMO) methodology to solve the problem. Ateamwas formed by choosing nappropriate employees (n is the number of tasks) from the time study database so that both skill and task constraints are satisfied. The times were assumed to be stochastic based on mean and standard deviation of time to complete a task for a particular category of employee. This stochastic simulation allowed us to estimate the number of pieces of garment sewn per hour, the productivity metric. In addition, we examined the composition of the team to decipher different levels of diversities (related to both disability and language) involved in the team. To assess diversity, we used the sum of Blau Index for both disability and languages. We now summarize our result across 21 replications.
Results
The surprising finding from the simulation is that the mean piece rate flow of the population that has no workers with disabilities is statistically insignificant (p=0.717) from a mix of able and disabled workers. In all other cases, there are significant differences (p<0.001) among the groups. The groups that are exclusively mentally disabled appear to perform the least with 27.11% lower productivity as compared to the benchmark case, while the group with short-term disabilities have a productivity 16.82% lower than the benchmark case. We are currently in the process of replicating the finding across multiple garments.
References
Associated Press (2013) “Labor department rules to Increase hiring of veterans, disabled workers” Reported in Washington Post, August 28 2013, Accessed online 2/20/2017
Cann (2012) “The debate behind disability hiring” Fast Company, Accessed online 2/20/2017 from