International Journal of Business and Applied Social Science (IJBASS)
©Center for Promoting Education and Research (CPER) USA www.cpernet.org
VOL: 4, ISSUE: 12
December/2018
https://ijbassnet.com/
E-ISSN: 2469-6501
BIG DATA ANALYTICS OF LABOR COST OVER THIRTY YEARS
MOUSUMI BHATTACHARYA
Charles F. Dolan School of Business
Fairfield University
Fairfield, CT 06824
Tel: (203) 254-4000 ext.2893
Fax: (203) 254-4105
E-mail: mbhattac@fairfield.edu
USA
ABSTRACT
Businesses need to analyze labor cost so that they understand what components of businesses affect the
amount of resources spent on employees. Interpreting this will help make managerial decisions in adjusting
costs of production and develop new strategies regarding investing in workers. Labor cost can vary due to
many factors such as employee numbers, assets, liabilities, sales, and debt. In this study, I analyzed labor cost
in relation to other firm characteristics, using a large panel data set over thirty years with over 24,000 firm-
year observations. I used two different statistical software, R and SPSS and the results are compared. Results
show that they both give the same output when running a multivariate regression. Both software are powerful
enough to analyze large datasets; however the form of input data and treatment of missing values, matter in
determining which one is more efficient. Managers and practitioners can use business analytics with big data
to draw conclusions and make important managerial decisions. Future study should include big data analysis
using more complex analytical techniques.
Keywords: HR Analytics, Labor cost
Acknowledgement: I want to thank Siddharth Jain, Research Assistant for his help in data analysis.
INTRODUCTION
The application of big data analysis in human
resource (HR) management is currently at its early stage.
Big data analytics is growing fast as organizations are
beginning to leverage these to gain competitive advantage
(Grover and Kar,2017). This paper examines the relevance
of analyses of big data on labor cost and whether
different types of software make a difference in the
analytics. I applied R software and SPSS statistical
package to a panel and cross-sectional data set on
companies from the COMPUSTAT North America
database from Wharton Research Data Services
(WRDS) over the past thirty years. I ran multivariate
regressions on labor cost and labor cost variability, as
related to several other variables, to compare the results
of the two statistical software. Although the statistical
results where similar, we found differences in the
process of computing, which might affect the choice of
analytics software and technique.
Big data analytics offers various benefits to the
organizations, by providing visual tools and multiple-
loop analysis to give fine-grained results that enhance
the quality of decisions taken (Li, Tao, Cheng, and
Zhao, 2015). So far, the issue is that statistical packages
like SPSS may not be able to find fine-grained
relationships in the data because it analyses data using
single-loop modeling. This means that it develops
relationships between variables in the form of
mathematical equations. In contrast, R software uses
machine learning techniques to perform the statistical
analyses. This is a significant research area because
machine learning is an algorithm that can learn from
data without relying on rules-based programming. It can
detect smaller relationships between variables, which
63