Bayesian and Classical Approaches Applied on Saving Habits of Employees Data: A Comparative Study
Genanew Timerga
Department of Statistics, Debre Birhan University, Debre Birhan, Ethiopia.
Nigatu Degu *
School of Mathematical and Statistical Sciences, Hawassa University, Hawassa, Ethiopia.
*Author to whom correspondence should be addressed.
Abstract
480 employees questionnaire were collected on saving habit at Debre Birhan town during February, 2010 to October, 2011. Descriptive statistics, Binary logistic regression and Bayesian statistical methods were used. Average private worker, low government worker were associated with a significantly lower likelihood of saving regularly versus not saving. Binary logistic regression indicates that age, education, dependent family members, transport, job satisfaction, expenditures and inflation significantly affect saving habits of employees. (Coeff-0.569, OR 0.566, P=0.000, CI 0.468, 0.685) the odds of saving decreases by 43.4% for one unit increase in dependent family members. The regression coefficient for the consumption growth rate on the one-period lagged consumption growth rate is expected to be positive. Capacity of employees’ utilized, formal method of saving institutes is higher than informal saving institutes. Our measures that are expected to capture various precautionary saving habits, that is, number of earners in a family and job security of the head of household, are not perfect in capturing uncertainty about future income.
Keywords: Logistic regression, Bayesian analysis, saving habit, MCMC methods