Ana Aizcorbe

Changing mix of medical care services: Stylized facts and implications for price indexes

The utilization of health care services has undergone several important shifts in recent years that have implications for the cost of medical care. We empirically document the presence of these shifts for a broad list of medical conditions and assess the implications for price indexes. Following the earlier literature, we compare the growth of two price measures: one that tracks expenditures for the services actually provided to treat conditions and another that holds the mix of those services fixed over time. Using retrospective claims data for a sample of commercially insured patients, we find that, on average, expenditures to treat diseases rose 11% from 2003Q1 to 2005Q4 and would have risen even faster, 18%, had the mix of services remained fixed at the 2003Q1 levels. This suggests that fixed-basket price indexes, as are used in the official statistics, could overstate true price growth significantly.

Moore's Law and the Semiconductor Industry: A Vintage Model

In this paper we develop a vintage model to gain a better understanding of the semiconductor industry and its role in recent U.S. productivity gains. Unlike previous work, in our model the observed price declines of individual chips are driven by the introduction of better vintages rather than by learning economies. Dominated chips, nonetheless, continue to be produced, for a time, due to sunk investments in chip-specific production equipment. The model lends partial support to Jorgensons hypothesis that an exogenous increase in Moores Law could have generated the more rapid price declines, and faster productivity growth, seen after 1995.

Moore's Law, Competition, and Intel's Productivity in the Mid-1990s

In the mid-1990s, a pickup in measured productivity growth for the semiconductor industry coincided with an economy-wide acceleration in labor productivity growth. The pickup in semiconductor markets reflected an increase in the growth of real output that was generated by what Dale Jorgenson (2001) called an “inflection point” in the price indexes for the semiconductor industry. Jorgenson hypothesized that the inflection point reflected increases in the rate of product innovation made possible by an increase in Moore’s Law, a stylized description of technology that currently states that the number of electrical components on a chip will double every eighteen months. Within semiconductors, microprocessors (MPUs) produced by Intel—the world’s largest producer of the chips that serve as a computer’s central processing unit—were the primary contributor to the inflection point in the semiconductor index. The inflection point in the price index coincided with two changes in the price contours for Intel’s chips. First, price contours for Intel’s chips became steeper around 1995. Because most price index formulae boil down to functions of weighted averages of price change, steeper price contours translate directly into more rapidly declining price indexes. At the same time, the product lifecycle for MPUs—the length of time chips are sold in the market—shortened and Intel began to introduce chips more frequently. What caused these changes in pricing and product cycles? This paper provides a simple framework to help gain some intuition on these issues. The model provides a set of conditions under which an increase in Moore’s Law is consistent with both of these stylized facts. In the model, an increase in Moore’s Law raises the quality of future chips relative to today’s chips. If consumers view these chips as substitutes, then increases in the quality of tomorrow’s chips push down the prices for today’s chips and can, under certain conditions, generate an inflection point in the price index. However, the framework also suggests that changes in the attributes of contemporaneous substitutes can have the same effects. Thus, the model suggests that increases in the quality of competitor’s chips can generate an inflection point through the same channel. This is an important possibility to consider because Intel faced increasing competition from AMD beginning in the mid-1990s, about when the inflection point occurred.

Measuring Health Care Costs of Individuals with Employer-Sponsored Health Insurance in the U.S.: A Comparison of Survey and Claims Data

As the core nationally representative health expenditure survey in the United States, the Medical Expenditure Panel Survey (MEPS) is increasingly being used by statistical agencies to track expenditures by disease. However, while MEPS provides a wealth of data, its small sample size precludes examination of spending on all but the most prevalent health conditions. To overcome this issue, statistical agencies have turned to other public data sources, such as Medicare and Medicaid claims data, when available. No comparable publicly available data exist for those with employer-sponsored insurance. While large proprietary claims databases may be an option, the relative accuracy of their spending estimates is not known. This study compared MEPS and MarketScan estimates of annual per person health care spending on individuals with employer-sponsored insurance coverage. Both total spending and the distribution of annual per person spending differed across the two data sources, with MEPS estimates 10 percent lower on average than estimates from MarketScan. These differences appeared to be a function of both underrepresentation of high expenditure cases and underestimation across the remaining distribution of spending.

When do Matched-model and Hedonic Techniques Yield Similar Measures?

Hedonic techniques were developed to control for quality differences across goods and over time in order to construct constant-quality aggregate price measures. When the available data are a panel of high-frequency data on models whose characteristics are constant over time, matched-model price indexes can also be used to obtain constant quality price measures. We show this by demonstrating that, given data of this type, certain matched-model indexes yield price measures that are numerically close to those obtained using hedonic techniques. ; * This paper is a condensed version of a paper that was presented at the CRIW workshop on Price Measurement at the NBER Summer Institute, July 31-August 1, 2000 and is available at http://www.nber.org.

Price Measures for Semiconductor Devices

This note provides quality-adjusted price indexes and nominal shipments data for highly disaggregate classes of semiconductor devices. These data may be used to construct indexes under different assumptions from those used in indexes that are currently available. Because the construction of these building blocks require some assumptions, the indexes are compared with similar price measures constructed by Bruce Grimm (1998) and by the Bureau of Labor Statistics.

The role of semiconductor inputs in IT hardware price decline: computers vs. communications

Sharp declines in semiconductor prices are largely responsible for observed declines in computer prices. Although communications equipment also has a large semiconductor content, communications equipment prices do not fall nearly as fast as computer prices. This paper partly resolves the puzzle - first noted by Flamm(1989) - by demonstrating that prices for chips used in communications equipment do not fall nearly as fast as prices for those chips used in computers, and those differences are large enough to potentially explain all of the output price differences.

Policy makers will need a way to update bundled payments that reflects highly skewed spending growth of various care episodes.

Bundled payment entails paying a single price for all services delivered as part of an episode of care for a specific condition. It is seen as a promising way to slow the growth of health care spending while maintaining or improving the quality of care. To implement bundled payment, policy makers must set base payment rates for episodes of care and update the rates over time to reflect changes in the costs of delivering care and the components of care. Adopting the fee-for-service paradigm of adjusting payments with uniform update rates would be fair and accurate if costs increased at a uniform rate across episodes. But our analysis of 2003 and 2007 US commercial claims data showed spending growth to be highly skewed across episodes: 10 percent of episodes accounted for 82.5 percent of spending growth, and within-episode spending growth ranged from a decline of 75 percent to an increase of 323 percent. Given that spending growth was much faster for some episodes than for others, a situation known as skewness, policy makers should not update episode payments using uniform update rates. Rather, they should explore ways to address variations in spending growth, such as updating episode payments one by one, at least at the outset.

The value of microprocessor designs

Applying a centuries-old technique to modern cost estimation.