ESTIMATES OF THE DEMAND FOR MEDICAL CARE UNDER DIFFERENT FUNCTIONAL FORMS

Abstract

The trend of rising expenditure on medical care could pose substantial cost on the Singapore economy but yet, statistical estimate on the demand for medical care in Singapore is negligible. The purpose of this empirical study is to provide insights on the demand for medical care in Singapore and this is done through the following three objectives. The first objective of this study is to show that different functional forms of estimating equation produce different results independent of the data.1 The second objective is to present a quantitative representation of the effect of both price and non-price determinants of demand for medical care and hence, a framework for empirical study on the demand for medical care in Singapore. The third objective is to find an appropriate model for estimating the price and income elasticities for the demand for medical care. The selection is based on a set of criteria which include theoretical consistency, goodness-of-fit and the overall significance of the estimated regression. To show the effect from using different functional forms, three specifications have been employed on the same set of data input. The three specifications are the linear model, log-linear model and the almost ideal demand system by Deaton and Muellbauer. The demand for medical care is represented by physician office services and hospital inpatient care. The data input for this study is extracted from a cross-sectional survey conducted on a sample of the Singapore population. The method of multi-stage sampling have been employed and a total of 108 respondents have been selected from the Singapore population. Comparing the regression results, we found that there exist wide variations in the significant level of the explanatory variables for the three models. In some cases, the direction of influence is reversed. These results are consistent with the conclusion by Hunt-McCool, Kiker and Ng (1994). The price elasticities estimated for the three models range from 0.02469 to -0.3057. The income elasticities range from -0.2498 to 1.1509. Given that the same data has been employed for the three models, the implication is that researchers should not over rely on a single estimate. Turning to the second objective, a comparison of the regression results yield three explanatory variables which have performed consistently well across the three models. They are the existence of major physical limitations, health status and race. The strength and consistency of their influence provided the supporting evidence on the relevance of these variables in the market for medical services. For the third objective, we put restriction on the explanatory variables and compare the results for the models under three selected criteria. The log-linear model satisfied all these criteria and is judged to be the appropriate model. Under this model, the estimated price elasticity for physician services is -0.1350 while the estimated income elasticity is -0.0054. The result suggested that physician services have low price elasticity and is regarded as an inferior good. 1 This approach is similar to the study by Hunt-McCool, Kiker and Ng ( 1994) on the demand for medical care in the US.