Performance and Efficiency of Public and Private Hospitals in Papua New Guinea: Comparison of Stochastic Frontier and other Approaches of Measuring Efficiency Scores

Tuesday, June 25, 2019
Exhibit Hall C (Marriott Wardman Park Hotel)

Presenter: Mahmud Khan

Co-Authors: Xiaohui Hou; Ibrahim Demir; Rifat Haider; Olga PM Saweri

Economic growth of Papua New Guinea (PNG) declined from an average rate of 9% during 2012-2015 to less than 2.5% since 2015. Slowing of the economy has been associated with lowering of total health expenditure per capita. With the decline in economic support, it has become even more important for PNG to ensure efficient use of available health care resources. This study estimates efficiency and performance scores of secondary and tertiary health facilities in PNG and identifies policy-sensitive measures for improving hospital performance in an environment of declining resource availability.
Methodology: All upper level facilities (defined as level 5-7 facilities in PNG’s health system) operational at the time of the survey were selected for survey. Sixty mid-level (level 3-4) publicly-run and church-run facilities were selected from 11 provinces. Only the functional facilities were considered in the selection of facilities for the survey. Since Church-run facilities provide about 50% of services at this middle level, attempts were made to select equal number of publicly-run and church-run facilities. To measure the performance and efficiency of health facilities surveyed, a number of performance parameters were compared. Efficiency scores were estimated using both stochastic and non-stochastic approaches. Possible factors affecting efficiency and performance of facilities were also examined.
Results: Bed occupancy rates were around 40% for Church-run and Publicly run secondary health facilities implying that 60% of beds at this level were vacant. Bed occupancy rates increase with increasing complexity of upper level facilities, reaching 80% for the national referral hospital. A number of other parameters such as cost per unit of output, output per clinical staff, input-output ratio, Lasso efficiency, etc. were compared. Stochastic frontier (SF) and DEA were also used to obtain efficiency scores. Both SF and DEA imply that the variability across facilities was due to random variations rather than any systematic variations. Constrained regression model and order-alpha approach generated efficiency scores for all facilities in the sample. Ownership status, facility size, geographic location, personnel mix, degree of autonomy of facility managers in terms of personnel and financial decision-making, etc.
Conclusions: In general, larger hospitals turned out to be relatively more efficient compared to smaller hospitals. Larger hospitals show lower vacancy rates for allotted positions. Urban location of facilities allows them to provide significantly more services than others. Among the mid-level facilities, Church-run facilities performed better than public sector managed facilities. It appears that greater autonomy in terms of personnel decisions and financial management helps to improve efficiency and performance measures.