کشش قیمت هزینه در خدمات بهداشتی و درمانی

Policymakers in countries around the world are faced with rising health care costs and are debating ways to reform health care to reduce expenditures. Estimates of price elasticity of expenditure are a key component for predicting expenditures under alternative policies. Using unique individual-level data compiled from administrative records from the Chilean private health insurance market, I estimate the price elasticity of expenditures across a variety of health care services. I find elasticities that range between zero for the most acute service (appendectomy) and −2.08 for the most elective (psychologist visit). Moreover, the results show that at least one third of the elasticity is explained by the number of visits; the rest is explained by the intensity of each visit. Finally, I find that high-income individuals are five times more price sensitive than low-income individuals and that older individuals are less price-sensitive than young individuals.

U.S. policymakers are debating health care reform options to reduce the large and growing cost of medical care. The Medicare Modernization Act of 2003 and the Affordable Care Act of 2010 both attempt to address this issue, relying crucially on assumptions regarding consumer responsiveness to out-of-pocket costs. To evaluate the potential benefit of different health care reform options in reducing costs, policymakers need a better understanding of expenditure elasticity, i.e., the ways in which consumer demand for health services changes in response to differences in out-of-pocket costs (referred to as “price” in this paper). Consumers hoping to limit their own out-of-pocket costs respond to price in two ways: by changing the frequency of service or by changing the quality of care to reduce per-visit costs. Understanding how individuals’ trade off frequency and quality is thus crucial for policymakers seeking to control costs. Unfortunately, there is little empirical evidence on how consumers respond to differences in price and the ways in which this varies by type of health service (e.g., emergency room care, routine visits) and by individual characteristics (e.g., income, age, education, socioeconomic status). This lack of research is primarily due to the difficulty of identifying exogenous, or externally caused, variation in prices. In health insurance markets, individuals select their plans using information that they – and not the provider or insurer – possess about their health status. This “information asymmetry” can lead to selection bias in those individuals who expect to use more services than the average person or the ones that are risk-averse. Thus, any type of health event or shock that is related to the individual's health status will therefore be correlated with the coinsurance rate of the chosen plan, creating an endogeneity problem that biases estimates for price elasticity. Only a few studies have been able to identify sources of exogenous price variation in health care usage, and they have been able to do so only in limited settings. For example, using the RAND Health Insurance Experiment of the 1970s, Manning et al. (1987) randomized consumers into health insurance plans with varying levels of generosity. Kowalski (2009) and Eichner (1998) used health shocks to individuals with large families who shifted their coinsurance rates (i.e., the percentage of medical expenses, beyond the deductible, that must be covered by the patient) from a fixed amount to zero to estimate the impact of such a change on individual's total expenditures. In these papers, researchers focused on total expenditures but were not able to examine how sensitivity to price varied across different types of health services or different individual characteristics. In the literature of price elasticity, this is the first paper, to my knowledge, that estimates elasticities in a context of a middle-income country, like Chile. Therefore this could be a good starting point for future research. Moreover, the estimates are comparable to the ones found in the literature for high-income countries, like the U.S. In this study I use a unique and detailed data set to describe new evidence on how health care consumers respond to changes in the price of care. I describe the ways in which price elasticities vary both by type of health service and consumer demographics. This study uses individual-level census data from the Chilean private health insurance market. Several features of the Chilean data make it useful for understanding the price elasticity of health expenditures in the United States. First, the health care system in Chile incorporates several policy mechanisms currently under debate in the U.S., such as health insurance exchanges, individual mandates, and regulations on the private health insurance system (e.g., premiums based on community ratings and minimum levels of benefits). Second, as in the United States, the private insurance market is a significant part of health care system in Chile. Third, detailed information from five datasets is available concerning patient characteristics, family member characteristics, plan characteristics, and prices (e.g., claims and out-of-pocket expenditure by individual and health care service). I combine these data to construct a panel of plan choices, fees for services at service providers, coinsurance rates and insurer payment caps for all participants in private sector plans. The Chilean health care system is thus an interesting case to study on several levels, and access to the complete private-sector administrative records allows me to analyze the entire population insured under the private system. Finally, the coverage and richness of the data allow me to examine the heterogeneity of price responses, allowing for a richer understanding of consumer behavior in response to differences in price. The paper is organized as follows. Section 2 provides a brief literature review. Section 3 presents background information on the health insurance system in Chile. Section 4 describes the econometric approach and instrumental variable strategy used in this study. Section 5 describes the data and sample selection. Section 6 outlines concerns with the approach. Section 7 shows the results of the main regressions across health care services for two main types of health services: urgent (acute) health care services and elective health care services, while Section 8 presents a deeper analysis of the price elasticity across age and income. Section 9 outlines the checks for robustness. Section 10 is the conclusion.

By using detailed and previously unavailable administrative data from Chile, this paper presents new empirical evidence on how health care consumers respond to prices. The analysis takes advantage of unique institutional features of the Chilean system to resolve identification biases that limited previous research on the price elasticity for health care. I found that consumer response varies by type of health service. Specifically, individuals are more sensitive in their demand for elective care (home visits, psychologist and physical therapy evaluations) than for acute care (appendectomy, cholecystectomy [gallbladder removal] and arm casts). The estimated demand elasticities for elective health care are −1.88, −2.08 and −0.32, respectively, whereas the demand elasticities for acute health care are close to zero. The results on elective care are comparable to the ones found by Kowalski (2009) for a U.S. population. I also estimated the price elasticity for the number of visits. I found that almost one-third of the elasticity for expenditure on home care is explained by the number of visits (extensity) and two-thirds by the intensity (the price of the health service) of each visit. For psychologist visits, the number of visits explained 40% of the elasticity; for physical therapy evaluations, the number of visits explained 60% of the elasticity. Given the richness of the data, I extended my analysis to examine how demographic factors affect price elasticity. I found that price sensitivity increases with income and decreases with age. The results for high-income people may be due to the fact that high-income individuals are better at calculating complicated out-of-pocket expenditures because they are more educated. On the other hand, it is possible that this income-based heterogeneity is related to the variety of available pricing structures, with high-income individuals able to more easily move from one physician to another than low-income individuals. These results reveal heterogeneity in price responsiveness based on types of services and demographics. Several robustness checks, using two time periods and alternative distributions for prices, support these main findings. One of the main concerns in the U.S. health care reform debate is the challenge of predicting and controlling health care costs. This paper addresses these points by quantifying expenditure elasticity and revealing how it varies. Policymakers can use this paper's results, for example, to understand how coinsurance rates or other features of the insurance plan might vary based on the type of service or demographic characteristics of the consumer. These design measures would allow policymakers to mitigate consumers’ tendency to overuse unneeded care and would reduce average health care expenditures, regardless of whether instituted by private or (if established) a public insurance provider. Although this paper provides a deeper understanding of consumer behavior in health markets, there remain unresolved questions for future research. I do not address preventive care, which may have important policymaking implications. For example, we do not know how consumers respond to changes in the price of preventive care and to what extent they will use free (subsidized) preventive care services. Understanding this topic may prove useful to further controlling future health care expenditures. In future research, I plan to study the impact of preventive care on future medical expenditure. Some of the paper's results need to be explored further. Specifically, understanding what drives the low levels of price responsiveness among older individuals is of the utmost importance in both academic and policymaking arenas. Is it that this population is not well informed of the pricing structure, or is it that they cannot understand the calculations involved in pricing? A formal experiment may help illuminate this topic by examining how new information alters consumer behavior for older consumers. On a final note, the results of this study are applicable to this particular population, which is younger, healthier and richer than the Chilean population as a whole. Therefore, we need to exercise care in applying the results to other samples. Nonetheless, the findings provide a starting point for moving in that direction.