Technical efficiency of pension schemes and provident funds in Kenya: an application of data envelopment analysis
Abstract
Pension schemes and provident funds in Kenya have expanded in terms of members and
contributions. However, little is known about their level of technical efficiency. The purpose
of this study was to estimate technical efficiency of pension schemes and provident funds in
Kenya and to identify the factors that are likely to influence their technical efficiency.
In this study, data from 161 pension funds and provident funds are sampled. Of the 161
pension schemes, 118 schemes are pension funds while 43 schemes are provident funds.
Input oriented data envelopment analysis is used to calculate constant returns to scale and
variable returns to scale technical efficiency scores. The study also estimated both Tobit
model and liner model to identify the determinants of technical and scale efficiency of
pension funds and provident funds in the second stage. In the second stage, regression is
estimated having the technical and scale efficiency scores as dependent variables. The
independent variables for the regression analysis are age, size, market share of the pension
scheme or provident fund, employer contribution rate and the employee contribution rate to
the pension scheme.
The results revealed that of the 118 pension funds only 4 were fully technically efficient with
the majority (73 pension funds) having technical efficiency score of less than 50 per cent. Of
the 43 provident funds ,15 had technical efficiency score of 70 per cent and above. The mean
scale efficiency scores for pension funds was 83.9 per cent while for provident funds was
74.9 per cent. A scale efficiency score of less than 100 per cent implies that pension funds
and provident funds in Kenya are not operating at optimal scale or size.
Market share and size was positively related to technical efficiency of pension funds while
age, employer contribution rate and employee contribution rate were negatively related to technical efficiency for pension funds. On the other hand, the newly formed pension funds
were found to be technically efficient when compared to older pension funds. The higher the
rate of contribution to a pension fund by the employer the less technically efficient will the
pension fund operate. On scale efficiency, age and size of a pension fund were found to be
negatively related to scale technical efficiency while market share, employer contribution
rate, and employee contribution rate had a positive relationship.
For provident funds, market share and employer contribution rate had a positive relationship
with technical efficiency while age, size, and employee contribution rate had a negative
relationship. The implication of these results is that the higher the provident fund’s market
share is in the industry, the greater will be its technical efficiency. The higher is the employer
contribution rate; the provident fund will tend to operate technically efficiently. On the other
hand, older provident funds and tend to be less technically efficient. Further, the bigger the
size of a provident fund, the less technical efficient it will be. On scale technical efficiency,
age, size, and employer contribution rate had negative relationship with scale technical
efficiency while market share had a positive relationship.
The low technical efficiency score reflects that a greater amount of inputs to the pension
schemes is wasted. Pension schemes need to improve their technical efficiency since the
levels observed are below the frontier. Further, the results of this study can heighten the
awareness of policymakers in Kenya regarding the technical efficiency of the pension
schemes in light of its primary objective of providing income at retirement.
Publisher
University of Nairobi