Does Price Transparency Improve Market Efficiency? Implications of Empirical Evidence in Other Markets for the Health Sector
Congressional research reportApr 29, 2008
Ask Donna
What actually matters in this document.
Text
ȱ
ȱȱ¢ȱ ȱȱ
¢ǵȱ ȱȱȱ
ȱȱȱȱȱȱ ȱ
ȱ
ǯȱ ȱȱ
¢ȱȱȱ¢ȱ
ȱ ǯȱ ȱ
ȱȱȱȱ¢ȱ
ȱŘşǰȱŘŖŖŞȱ
ȱȱȱ
ŝȬśŝŖŖȱ
ǯǯȱ
řŚŗŖŗȱ
ȱȱȱ
Prepared for Members and Committees of Congress
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
¢ȱ
Consumer advocates, proponents of wider use of market incentives in the health care sector, and
some policy makers have called for greater price transparency. These measures might include
posting prices in an accessible form or regulations constraining price discrimination (different
prices charged to different customers). Price transparency implies that consumers can obtain price
information easily, so they can usefully compare costs of different choices. Price transparency
may also mean consumers understand how prices are set and are aware of price discrimination. In
health care markets consumers often have difficulty finding useful price data. In particular, few
consumers have a clear idea of what hospital stays or hospital-based procedures will cost, or
understand how hospital charges are determined.
Many empirical studies have investigated how changes in price transparency have affected
various markets. Most of this evidence, largely relating to advertising restrictions and lower
search costs on the Internet, suggests that price transparency leads to lower and more uniform
prices, a view consistent with predictions of standard economic theory. If this evidence could be
applied to the health market, it would suggest that reforms that increase transparency would
reduce prices. In cases involving NASDAQ and Amazon.com, public reaction created pressure to
change pricing strategies. A few studies, involving intermediate goods in one case and less clearly
identified advertising effects in others, found that transparency raised prices.
However, the special characteristics of the health market make it difficult to directly apply
empirical evidence gathered from other markets. These characteristics include limits on
competition among hospitals, complicated products that vary in quality, intermediate agents
(physicians) who make choices, and third-party payment of costs through insurance. The
dispersion of prices for similar health care procedures is high, which suggests that these markets
are not working well with respect to price outcomes, as would be expected in ordinary
competitive markets. In addition, prices paid by different types of payers vary dramatically. On
average, patients without insurance or who pay their own bills pay much more relative to what
private insurers, Medicare, and Medicaid pay.
Despite these complications, greater price transparency, such as accessibly posted prices, might
lead to more efficient outcomes and lower prices. Some markets where lifting advertising
restrictions led to lower prices also involved complicated products such as eye care, suggesting
that the complex nature of health care may not be a barrier to benefits from price transparency.
Internet comparison shopping sites also appear to have lowered prices for many products. Better
price information might allow patients, either directly or through their physicians, to obtain better
value for health care services. Several states and health insurers now provide online data on
hospital costs. These price transparency initiatives, at least so far, have had little visible effect on
pricing. Public pressure, which in some cases has caused hospitals to curtail aggressive bill
collection tactics, might change hospitals’ and health care providers’ pricing behavior. This report
will be updated as events warrant.
ȱȱȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
Introduction ..................................................................................................................................... 1
Empirical Evidence on the Effects of Price Transparency .............................................................. 2
What Are the Implications for Health Care Markets? ..................................................................... 4
Why Do Different Prices Persist? Differentiated Products and Price Discrimination .............. 4
Cost Structures and Pricing ....................................................................................................... 5
Special Characteristics of the Health Care Markets.................................................................. 7
Health Care Is Complicated ................................................................................................ 7
Physicians as Agents ........................................................................................................... 7
Patients Pick Physicians and Hospitals Pick Physicians..................................................... 8
Other People’s Money Pays for Most Hospital Care .......................................................... 8
Patients Have Poor Information About Hospital Quality and Costs ................................... 9
Summary: Special Characteristics of Health Care Markets .................................................... 10
Hospital Pricing..............................................................................................................................11
Nuts and Bolts ..........................................................................................................................11
Price Variation Among Hospitals ............................................................................................ 13
How Does Hospital Price Dispersion Compare To Other Markets? ....................................... 17
Implications of Hospital Price Dispersion .............................................................................. 19
Price Transparency Initiatives of Governments, Insurers, and Interest Groups ............................ 21
Does Price Transparency Reduce Price Variability? Some Preliminary Results .................... 24
How Would Greater Price Transparency Affect the Health Care Sector? ..................................... 27
Internet Price Comparison Sites.............................................................................................. 29
Will the Health Sector Change Like Other Industries? ........................................................... 29
ȱ
Figure 1. Hospital Charges for Selected Diagnostic Tests............................................................. 13
Figure 2. Hospital Charges for Two Common Analgesics ............................................................ 14
Figure 3. Distribution of Average Charges Per Stay for Normal Vaginal Birth............................. 18
Figure 4. Distribution of Average Charges Per Stay for Heart Failure.......................................... 19
Figure 5. Price Distribution for a Samsung HP-R6372 HDTV..................................................... 20
Figure 6. Distribution of Average Charges Per Stay For Normal Birth, 2003-2006...................... 25
Figure 7. Scatter Plot for Changes in Avg. Daily Charges and Discharges for Normal
Birth (DRG 373)......................................................................................................................... 26
ȱ
Table 1. Average Costs and Charges for Selected Hospitals, By Type of Payer............................ 15
Table 2. Hospital Payment-To-Cost Ratios By Type of Payer, 1991-2000 ................................... 16
Table 3. Variability of Average Hospital Charges and Samsung HDTV Prices ............................ 21
ȱȱȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
¡ȱ
Appendix. Review of Empirical Studies on Price Transparency................................................... 31
ȱ
Author Contact Information .......................................................................................................... 43
ȱȱȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
ȱ
Price transparency helps consumers obtain price information easily, which allows them to make
useful comparisons of costs of alternative choices. Price transparency may also mean that
consumers understand how prices are set and are aware of any price discrimination (different
prices charged to different customers). In health care markets consumers often have difficulty
finding useful price information. In particular, few consumers have a clear idea of what hospital
stays or hospital-based procedures will cost, or understand how hospital charges are determined.
Prices charged by hospitals vary significantly across hospitals and vary within hospitals across
categories of patients.
Transparent prices play a key role in the efficient allocation of goods and services. Under certain
conditions, the decentralized and self-interested decisions of firms and households in a price
system yield resource allocations that avoid waste and that match what suppliers make and what
consumers want, which is how economists define efficiency. Financial economics researchers
typically define markets as efficient when prices reflect all available information and when prices
adjust swiftly as new information arrives. If buyers and sellers do not know what prices are, then
some mutually agreeable trades will fail to occur, thus creating inefficiencies. If buyers can see
and compare prices for the same good offered by different sellers, the buyers then save money by
choosing the cheapest vendor. If goods are similar but not identical, buyers then can compare
prices and qualities offered by different sellers and pick whichever offer suits them best. The
buyers’ ability to choose an offer that suits them best puts tremendous pressure on all sellers to
lower prices, improve quality, or both. Without such competitive pressure firms that are less
efficient or that are earning excess profits can remain in the market, and prices will be higher than
they would otherwise be.
Lack of transparent prices may also contribute to price discrimination, which can cause different
customers to pay higher prices, an outcome that may be acceptable in some markets but may lead
to undesirable consequences in others. For example, if the customers with the least bargaining
power also tend to be those with the least ability to pay, such discrimination may be deemed
particularly undesirable.
Barriers to price transparency include both explicit restrictions on information (such as
government restrictions on price advertising or concealment by firms of prices or price-setting
approaches, including negotiated prices) and costs of search by consumers. The simplest theories
suggest that more information about prices should decrease prices and also bring prices closer
together, but certain theories predict that more price information could raise average prices, and
advertising might raise prices by increasing demand or brand identification.
The first section of this paper briefly reviews the empirical studies of the effect of changes in
price transparency on prices and quality of goods in a variety of industries. Most of this evidence
relates to markets where buyers are the final end users of the good, and the bulk of evidence
suggests that more transparent prices lead to lower prices and transactions costs. This section
includes examples of direct effects of price transparency acting through normal market
mechanisms (as in the case of lifting advertising restrictions or reducing search costs) as well as
instances in which publicity about pricing strategies altered firms’ behavior. (An appendix
contains a more detailed discussion.)
ȱȱȱ
ŗȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
The second section addresses the extent to which this evidence might be applicable to the health
care market. It addresses certain special characteristics of the health care market which may
reduce the importance of prices as signals, for example, the complicated nature of health care, the
intermediation of physicians in making health care choices including choosing hospitals, and the
presence of third party payment (e.g., insurance companies).
The third section then turns to a closer examination of how prices are actually set by hospitals and
the evidence that exists on price dispersion both across hospitals and across patient categories.
The fourth section discusses some initiatives undertaken by governments, insurers, and interest
groups to improve information about prices and to regulate price discrimination.
The final section draws the pieces together, suggesting that while it is difficult to determine the
consequences of greater consumer price transparency, it is reasonable to believe that greater
transparency would improve outcomes.
ȱȱȱȱȱȱȱ
¢ȱ
Isolating the effects of price transparency from other determinants of price is empirically difficult,
and the literature contains a variety of approaches used to identify these effects. A more detailed
discussion of this extensive literature is presented in the appendix.
Some examples of the effects of price transparency relate to the effect of publicity about pricing
practices that may be viewed as inappropriate and that may lead to fears of regulatory
involvement or consumer backlash. One such example relates to NASDAQ. In 1994, William
Christie and Paul Schultz, two Vanderbilt University financial economists, noticed that NASDAQ
dealers almost never quoted prices using odd eighths (i.e., 1/8, 3/8, 5/8, and 7/8) for many highvolume stocks of companies such as Microsoft, Intel, and Apple. This practice effectively created
a quarter dollar minimum spread between sellers’ asks and buyers’ bids, which increased the
trading profits of dealers. The day after these economists issued a press release about their
findings the practice was abandoned, and spreads for several major stocks fell by about half.1
Some other examples of transparency in financial markets suggested transparency lowered prices.
When Island, an electronic communications network, ceased displaying limit order data in 2002,
trading costs rose; when Island resumed a year later, trading costs fell. Another study found prices
more volatile after hours than during regular market hours when trades are immediately reported.
A second example of the effect of publicity involves the case of Amazon.com, the internet seller.
Amazon, according to reports, used characteristics gathered about individual customers from the
Internet itself (such as whether a customer was new to the site, what browser the customer was
using and what the customer purchased in the past, etc.) to charge different prices to different
1
William H. Christie and Paul H. Schultz, “Did NASDAQ Market Makers Implicitly Collude?,” Journal of Economic
Perspectives, vol. 9, summer 1995, pp. 199-208.
ȱȱȱ
Řȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
individuals. Once this strategy was publicized, the protests led Amazon to cease the pricing
variations and apologize.2
Another case study focused on the intermediate market. In 1993, the Danish Competition
Authority required that all ready-mixed concrete contracts be made public, which it hoped would
stimulate greater competition. Instead, average prices rose by 15%-20% and other factors such as
changing demand conditions played no discernable effect.3 There are two possible explanations
for this unexpected increase in prices with publicity. First, public prices may make collusion
among sellers easier. Rivals can observe sellers who undercut their competitors, and may be able
to mete out punishments in various ways. Second, price transparency may alter the strategic
incentives of sellers, inducing them to become tougher bargainers.
A larger body of studies estimates the effects of restrictions on advertising and posting of prices.
Most of these studies involved comparing jurisdictions that banned certain types of advertising,
primarily for vision exams and eyeglasses. Some studies focused on the effects of restrictions on
the advertising of prescription drugs and alcoholic beverages and restrictions on posting gasoline
prices. (It is important with advertising, which can increase demand for branded products, to
examine cases where some outside authority, in this case the government, restricts advertising.)
Two studies examined the effects of local advertising of food prices, one examining the effects of
the 1978 newspaper strike in New York City and another where researchers provided advertising
via direct mail. Although studies of quality are more difficult to undertake, two studies examined
these effects: one study examined the effect of mandatory fat content labeling and another the
effect of requiring restaurants to post hygiene quality grade cards. Almost all of these studies
found that more information on prices and quality lowered prices, improved quality, or both.
The final part of the appendix discusses the relatively new and growing body of studies on the
effect of better price information and lower search costs through computers and the Internet.
Studies have examined a wide range of items: automobiles, books and CDs, airline tickets, and
life insurance. The evidence was mixed for cars and for books and CDs, but showed reductions in
prices for airline tickets and insurance. These studies suggested that consumers using comparison
sites did pay lower prices and later studies, as the Internet became more common, more
frequently pointed to lower prices. Part of the difficulty of studying the effect of the Internet is
that Internet sellers may offer benefits to customers compared to conventional sellers, so that the
evidence on price comparison sites, which appeared to reduce prices and price variation, may be
more relevant than comparing prices of Internet and conventional sellers.
Considering all of the evidence of price transparency, the majority of the empirical studies tend to
find that greater price transparency, including advertising and reduction in costs of finding
information through the Internet, leads to lower and more uniform prices.
2
Robert M. Weiss and Ajay K. Mehrotra, “Online Dynamic Pricing: Efficiency, Equity and the Future of Ecommerce,” Virginia Journal of Law & Technology, vol. 6, no. 11 (2001), available at http://www.vjolt.net.
3
Svend Albaek, Peter Møllgaard, and Per. B. Overgaard, “Government Assisted Oligopoly Coordination? A Concrete
Case,” Journal of Law and Economics, vol. 45, December 1997, pp. 429-443.
ȱȱȱ
řȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱȱȱ ȱȱ ȱȱǵȱ
Can the evidence from other markets be used to analyze the effects of greater price transparency
in health care markets, or provide guidance about what measures might best be considered?
While the special features of the health care market that distinguish it from other markets are well
known among health economists, researchers and policy makers have sought ways to capture the
potential gains from increasing efficiency in the health care sector by the introduction of marketlike reforms. Whereas published prices in other markets provide important signals of the true
economic value of goods and services in other parts of the economy, the impenetrability of many
health care billing practices creates a barrier to rational decision making and analysis.
Prices in the health care markets reflect physician charges, hospital pricing, prescription drugs,
costs for medical devices and diagnostics, as well as other types of health care goods and
services. Certain market characteristics of industries that provide many of these products are
important in analyzing the effects of price transparency: they are subject to quality differences
(and are thus not entirely homogeneous products); the product may be one whose nature and
benefits are not easily understood by the customer; sellers charge different prices to different
customers and customers pay different (and often small) shares of the costs because of insurance;
and within specific geographic areas there may be few providers, at least in the case of hospitals.
These aspects of the health care market not only mean that prices will vary but they also (in many
cases) complicate the consumers’ understanding of expected prices or their response to price
differences; they also may mean that it is difficult for prices to bring about economic efficiency
(for example, because of lack of competitive markets). All of these aspects of the health care
market therefore may mute the effects of transparency on prices.
Prices clearly vary in the health industry, and why they vary is relevant to the implications of
price transparency. The discussion below reviews basic aspects of pricing that lead to different
prices in a market and are relevant to discussing barriers to the effect of transparency on prices in
the health market. The first section discusses two reasons that different prices persist for the same
product: product differentiation and price discrimination. As we shall see, both characteristics
exist in the health care market. Secondly, the cost structure of an industry may lead to market
power that allows different prices to be charged. Following that discussion, some specifics of the
health care market and how they relate to pricing characteristics are discussed. Many of these
characteristics are directly related to the role of price in consumers’ decisions. Finally, the
empirical evidence on price transparency presented in the first part of this report is examined in
light of these issues.
¢ȱȱȱȱǵȱȱȱȱȱ
ȱ
The “Law of One-Price,” which states the same good will sell for the same price, is a simple
consequence of buyers’ ability to pick the most advantageous offer. In many situations, however,
prices will vary. This may happen because two goods are not identical. For example, a store in a
more convenient location can charge more than a store in an out-of-the-way location. Spending
time in a resort during peak season is different than spending time in the same resort during low
season. Conversely, as the real estate maxim states, if the price of an apartment with a view is the
same as an otherwise similar apartment without a view, then there really isn’t a view. Moreover,
products that otherwise seem quite similar may be differentiated, if no more than in consumers’
ȱȱȱ
Śȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
minds, by brand, and certainly a great deal of advertising appears directed at differentiating
similar products, which permits suppliers to increase prices and profits. Because health care
depends on location, quality, and patient characteristics it is not a homogeneous product, and so
some price differential is expected.
Some sellers may gain larger profits by charging different prices to different groups of
consumers. For this to happen, firms must have some market power, meaning that they can raise
their average selling price by cutting back on the amount they sell. If the seller can identify
different groups that differ in their sensitivity to price changes, and if buyers cannot resell or use
arbitrage, then firms will earn higher profits by charging groups with lower price sensitivity a
higher price.4 For instance, airlines know that business travelers are usually less sensitive to
prices than leisure travelers. By imposing “Saturday night stayover” requirements for cheaper
fares, airlines are able to charge higher prices to business travelers who want to sleep in their own
beds on weekends.
Firms can price discriminate in a number of ways.5 Consumer electronics manufacturers offer
mail-in rebates in order to charge higher prices to customers who either value their time highly or
who are poorly organized, and who therefore fail to obtain those rebates. Car dealers charge
different prices for identical cars, an outcome of the bargaining process.
Price discrimination often benefits some classes of consumers: those who would probably pay
higher prices under uniform pricing. If airlines could not charge business passengers higher fares,
leisure travelers would certainly have to pay higher fares. Some price discrimination schemes,
such as college financial aid, are often justified on the grounds of fairness, although they can also
be explained by the desire to maximize profits. Hospitals before the Medicare Act often sought to
justify charging different rates to different customers on the grounds of fairness, although some
economists who examined the issue at the time were skeptical.6 Hospitals in the current health
finance environment—dominated by large insurers and managed care firms on the private side
and Medicare and Medicaid on the public side—typically attempt to charge more to uninsured
patients who have less ability to negotiate, even though uninsured patients are more likely to have
lower incomes than insured patients.7
ȱȱȱȱ
The structure of costs within an industry has important effects on the nature of pricing. Firms
with market power, which often arises from cost structures, will have some ability to set prices
differently from cost, and may be more resistant to competitive pressures that result from price
transparency.
4
In economic theory charging different groups different prices is called “third-degree” price discrimination. First
degree price discrimination occurs when sellers have information on the price sensitivity of individuals, and second
degree price discrimination occurs when sellers use quantity discounts.
5
For one list of marketing techniques designed to charge different prices to different customers, see F. M. Scherer and
David Ross, Industrial Market Structure and Economic Performance (Boston: Houghton-Mifflin, 1990), pp. 491-494.
6
R. A. Kessel, “Price Discrimination in Medicine,” Journal of Law and Economics, vol. 1, 1958, pp. 20-53. Kessel
examines physicians’ fees and argues the setting of higher fees for those with more means to be a standard case of price
discrimination to maximize profits.
7
This point is analyzed in more detail in the section entitled “Price Variation by Payer.”
ȱȱȱ
śȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
Firms will have limited market shares and will face strong competitive pressures to keep profit
margins low when firms have
•
fixed costs that are small relative to operating costs that can be added or cut in
the short run (variable costs), and
•
unit costs that increase as output increases.
On the other hand, if fixed costs are large relative to variable costs or if firms use an increasing
returns technology, then uniform pricing may be difficult to maintain, especially if the firm
cannot store its output.8 Economic theory suggests that industries that have high fixed costs and
which sell perishable goods or services face strong pressures to charge different customers
different prices and compete in markets subject to unstable prices.9 Increasing returns can often
be found in industries with network characteristics. For example, a phone connection is more
valuable within a large network than within a small one because more connections are possible.
Learning-by-doing effects are another example of increasing returns.
In addition, many hospitals provide indigent care for which they are not wholly compensated.
Such hospitals must find other ways to finance this care, which often involve cross-subsidies. In
these conditions, a simple flat-rate price system may not be a viable strategy for hospitals.
Therefore, imposing greater transparency of health care prices may require closer attention to
cross-subsidies and uncompensated training and care.
The hospital industry has some natural monopoly or natural oligopoly characteristics. A natural
monopoly exists where incremental costs fall as output rises through the relevant range of output
for a market. A natural monopoly would suffer losses if it set prices equal to incremental cost,
which is a standard condition for socially efficient pricing. Therefore, a natural monopoly must be
supported by some subsidy or must charge prices above incremental cost, which from an
economic perspective causes inefficiencies and market distortions. Industries with natural
monopoly characteristics are often regulated, and prices are often set administratively through
rate-of-return type regulations. The outputs of industries regulated under rate-of-return
procedures, however, are much simpler than the set of outputs which hospitals provide. For
example, electric power distribution, which generally has been subject to rate-of-return
regulation, deals with a single commodity which is uniform in its physical characteristics.10
Entry of new firms in an industry with natural monopoly characteristics is inefficient because at
least some firms will be forced to operate at inefficiently low levels. For example, entry of a new
hospital might cause the average number of empty beds in a market area to increase, which
increases average prices. Because of the hospital industry’s natural monopoly characteristics,
state and federal regulators have often imposed restrictions, such as Certificates of Need, on entry
of new hospitals. Theoretical models have been developed to better understand the tradeoffs
between the gains in competitive pressure and the loss of scale economies.11 In U.S. v. Carilion
8
Technically, fixed costs and increasing returns create non-convexities in a firm’s production function. This will cause
gaps in the firm’s supply curve, so that supply and demand curves might not intersect. In this case, there is no market
equilibrium. Experimental evidence suggests that pricing can be extremely erratic in such cases.
9
Lester G. Telser, “Competition and the Core,” Journal of Political Economy, vol. 104, no. 1, 1996, pp. 85-107.
10
Even if electricity is physically homogeneous, costs of generation and demand for power vary by time of day.
Nonetheless, electric power, even if differentiated by time of day, is much more homogeneous than outputs provided by
the health sector.
11
Rabah Amir, “Market Structure, Scale Economies, and Industry Performance,” working paper University of Southern
(continued...)
ȱȱȱ
Ŝȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
Health System a federal district court accepted the argument of two hospitals that wished to merge
that higher market concentration would lead to lower prices, and rejected the Department of
Justice’s claim that the merger would raise prices, providing an illustration of a case where the
scale economy argument prevailed.12
ȱȱȱȱ ȱȱȱ
Health care markets differ from markets for standardized commodities described in economics
textbooks in several important ways. The special features of health care have had a strong effect
on the evolution of health care markets. Five key features of health care markets are discussed
below; in general, they point to price being a less important signal than it typically is in other
markets. Prices could, however, become more important with a shift to insurance types such as
Health Saving Accounts where consumers confront higher prices at the margin.
ȱȱ ȱȱ
By its nature, health care cannot be easily standardized, making price dispersion difficult to
monitor. Different diseases affect different people in different ways, and treatments that work for
one patient may fail to help another. Patients may not know what disease or condition is affecting
them, and may have difficulty in articulating what is wrong with them and what they would like
treatment to accomplish. Hospitals are sometimes described as “job shops” to emphasize their
dissimilarity to assembly lines. Thousands of different types of procedures may be performed in
an average general hospital, and even specialized hospitals must be equipped to face a wide range
of conditions and complications.
Because hospitals produce many different outputs with many of the same inputs, allocating costs
to particular outputs or to specific patients can be somewhat arbitrary. There is no unambiguous
way to allocate the costs of employing nurses, pathologists, accountants, and billing clerks to
specific procedures or patients. Hospital management strategies that seek to assign such costs to
specific “profit centers” appear to rely more on rules of thumb than on precise economic
calculations.
¢ȱȱȱ
Because patients cannot always know what they want, physicians must serve as their agents. In
most cases, physicians will make a preliminary diagnosis, recommend which specialists will be
seen, and determine whether a patient is admitted to a hospital or not. It is true that ethical and
professional guidelines stress that physicians must act in the best interests of the patient. Still,
physicians may be swayed directly or indirectly by insurers, pharmaceutical companies, hospitals,
and peers in ways that might not benefit patients. While the vast majority of physicians feel a
strong professional compunction to provide the best care possible, they also face pressure to
(...continued)
Denmark at Odense, August 2000, available at http://www.econ.ku.dk/wpa/pink/2000/0008.pdf
12
U.S. v. Carilion Health System, 707 F. Supp. 840 (Western District of Virginia, 1989). For an economic and legal
analysis of this case see David Eisenstadt, “Hospital Competition and Costs: The Carilion Case (1989),” in John E.
Kwoka and Lawrence J. White, eds., The Antitrust Revolution: The Role of Economics (New York: Harper Collins,
1994).
ȱȱȱ
ŝȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
reduce costs to patients or insurance companies. The problem of agents considering their own
interests, along with those on whose behalf they act, exists in this market as well as many other
markets.13
ȱȱ¢ȱȱ ȱȱ¢ȱ
Because patients rely upon physicians as their agents, patients often do not choose which hospital
they enter. Rather, patients choose a physician, and the physician’s admitting privileges determine
where the patient goes. Hospital credentials committees decide which physicians get admissions
privileges based on a physician’s training, residency program, malpractice record, and other
relevant information. Although some physicians have admitting privileges at more that one
hospital, the available evidence suggests that most physicians admit the bulk of their patients to
one hospital.14 A patient needing an operation may have some choice of hospital if her physician
provides referral to more than one surgeon. While this provides the patient with some choice, the
patient rarely has detailed information about cost and quality, and is rarely in a position to make
an informed choice.
If a patient wishes to go to a certain hospital, then the patient must select a physician with
privileges there. Insurance companies offer physician directories which list hospital affiliations,
and hospitals often sponsor “physician-finder” services that feature “their” M.D.s. Therefore,
patients may have sufficient information to figure out which physicians they would need to
choose in order to go to a particular hospital in the event of some medical condition. (Emergency
admissions are generally sent to the nearest hospital with an emergency room or to a hospital
which specializes in trauma cases.) In some cases they may have information on quality through
studies that rank hospitals. The fact remains, however, that patients are usually in a poor position
to choose a hospital which best suits their needs because they lack the right information and
because they are shielded from information about cost differences among hospitals.
ȱȂȱ¢ȱ¢ȱȱȱ ȱȱ
Hospitals get slightly less than a third of their revenue from Medicare, another third from private
insurers and slightly more than a sixth of their revenue from Medicaid.15 While public or private
insurance protects patients from the financial consequences of a hospital stay, insurance also
makes patients insensitive to prices. By the time a patient reaches a hospital deductible, out-ofpocket payment limits for most insurance policies may have been reached for many patients. In
particular, for the most complicated episodes (which account for a disproportionate share of
hospital costs), most patients may be fully covered or fully bankrupt. In either case, price plays
little or no role in either choice of treatment or location of treatment.
Patients may indirectly choose their hospital and nature of their care through their choice of
insurance plan, and as noted above, through their choice of physician. Many plans using Preferred
Provider Organization (PPO) approaches restrict policy holders’ choice of hospital, or impose
13
The principal-agent problem, as it is referred to in economics, occurs in many contexts, including corporate managers
acting on the behalf of stockholders and tenants making decisions that affect owners of property.
14
Douglas R. Wholey and Lawton R. Burns, “Convenience and Independence: Do Physicians Strike a Balance in
Admitting Decisions?,” Journal of Health and Social Behavior, vol. 32, no. 3 (September 1991), pp. 254-272.
15
Uwe E. Reinhardt, “The Pricing of U.S. Hospital Services: Chaos Behind A Veil of Secrecy,” Health Affairs, vol. 25,
January/February 2006, pp. 57-69.
ȱȱȱ
Şȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
financial penalties for using hospitals outside the PPO network. One plan may be cheaper than
another because it is able to drive a harder bargain with hospitals or because it can restrict the cost
or amount of care which policy holders receive. While consumers can obtain information about
features of different insurance plans, that information arguably is often incomplete and confusing.
ȱ ȱȱ ȱȱ ȱ¢ȱȱȱ
Patients may also be in a poor position to choose their own hospital because they have little
access to information about hospital prices and quality or are not familiar with the information
that is available (such as hospital ratings). As with any other good or service, a good decision
about hospital selection must be supported with adequate information on costs and quality.
Hospitals in most states are not required to make public individual prices for items, and other
resources for comparative pricing information are limited. Aetna, for example, has provided price
information for physician services in selected areas, but this information is available only to its
subscribers.16
The impenetrability of hospital bills is legendary. Hospital bills for privately insured patients
routinely run for pages and contain hundreds of individual items. Hospital billing and coding
have become arcane arts, practiced by highly specialized clerks and consultants. Insurers and
government analysts have access to files that can be used to generate meaningful average costs,
but this information is not available to patients.
Compounding the problems patients face, they generally have access to little useful information
about health care quality. In part this is due to the inherent complexity of medical care and the
difficulty of defining and measuring quality. Consumers can quickly judge the quality of most
goods they buy on a daily or weekly basis, and make changes in shopping routines accordingly. In
some cases, such as obstetrics, word-of-mouth and reputation may lead patients to reasonably
well-informed choices among hospitals. In general, however, hospital stays are for most an
infrequent event and not many patients have enough experience or connections to compare
experiences in a range of different hospitals.
Large corporations, insurance companies, and government agencies have developed extensive
databases containing information reflecting the quality of health care. The development of large
electronic databases has opened the possibility of creating quality indices based on sophisticated
statistical methods. Presently, however, these data are largely unavailable to consumers.
Traditional approaches to quality monitoring in health care focus on “zero/one” indicators.
Physicians are licensed, and others are barred from providing medical care. Hospitals are
accredited and providers are certified for Medicare reimbursement. Such measures, however,
serve only to set lower bounds.
16
Testimony of Robin Downey, Vice President and Head of Product Development, Aetna, in U.S. Congress, House
Committee on Ways and Means, Subcommittee on Health, Hearing on Price Transparency, 109th Cong., 2nd sess., July
18, 2006. The expansion to the Washington area was reported in January W. Payne, “The Secret’s Out: Aetna Members
Gain Access to Care Price, Quality Data,” Washington Post, August 22, 2006, pp. F1, F4.
ȱȱȱ
şȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
Providing consumers with more useful data on outcomes may improve health care quality.17 Of
course, outcome data must include risk adjustments, so that statistics reflect the fact that healthier
patients will on average have better outcomes. For example, the United Network for Organ
Sharing, established by Congress in 1984, collects data on all transplant operations in the United
States. Risk-adjusted outcome data for each transplant center are available at
http://www.unos.org. Public availability of risk-adjusted outcome data puts pressure on surgeons
and transplant centers to improve performance. New York State has published risk-adjusted
average mortality rates for cardiac surgery since 1991. Once New York State started publishing
average mortality rates, patients at a top-performing hospital or surgeon reportedly had about half
the chance of dying as did those who picked a hospital or surgeon from the bottom-performing
25%.18 Massachusetts maintains a website with death rates for coronary artery bypass graft
(CABG) operations for specific hospitals and surgeons. This site lists the number of procedures
performed by specific surgeons for several other types of operations.19 Pennsylvania published a
report on cardiac surgery that listed hospital-specific data on average charges, average payment
by commercial insurers and Medicare, and risk-adjusted mortality and readmission rates. This
report also listed surgeon-specific data on risk-adjusted mortality and readmission rates for
CABG procedures.20 Data presented in the report showed little connection between average
charges and adjusted mortality rates.21
¢DZȱȱȱȱ ȱȱȱ
If the market satisfies conditions of the model of perfect competition, which imply that
consumers are fully informed and can choose the lowest price, prices will converge to the cost of
producing the last unit of output and goods will be distributed efficiently.22 More generally, the
“Law-of-One-Price” asserts that consumers’ ability to choose the most advantageous offer will
ensure that the same good will sell for the same price. To the extent that transparent pricing helps
markets rapidly converge by bringing prices in line with incremental costs, it promotes economic
efficiency.
Many markets do not satisfy conditions of the model of perfect competition. If consumers are
poorly informed, or hindered from taking their most advantageous option, prices might not
converge to efficient levels, if they converge at all. While such problems can arise in markets for
simple goods, the problems are exacerbated for more complex goods and services, such as health
care. Several aspects of health markets, including natural differentials in the product due to
differences in quality and patient characteristics and the widespread practice of price
discrimination, limit the effects of price transparency. In addition, other important characteristics
interfere with price signals and competitive pricing outcomes: the product is complicated,
physicians rather than consumers tend to determine the product purchased, patients generally do
17
For a more extensive analysis of the potential of giving consumers with useful outcome data see Michael E. Porter
and Elizabeth O. Teisberg, Redefining Health Care: Creating Value-Based Competition on Results (Allston, Mass:
Harvard Business School, May 2006).
18
Ashish K. Jha and Arnold M. Epstein, “The Predictive Accuracy of The New York State Coronary Artery Bypass
Surgery Report-Card System,” Health Affairs, vol. 25, no. 3, 2006, pp. 844-855.
19
These data reports are available at http://www.mass.gov/healthcareqc.
20
Pennsylvania Health Care Cost Containment Council, “Cardiac Surgery in Pennsylvania 2005,” June 2007.
21
Reed Abelson, “In Health Care, Cost Isn’t Proof of High Quality,” New York Times, June 14, 2007.
22
Efficiency here means no waste in production and that all gains from trade are exhausted.
ȱȱȱ
ŗŖȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
not directly pick hospitals, many costs are covered by third parties, and patients have poor
information about costs.
In sum, health care patients often have only a limited and indirect ability to choose which hospital
they will be treated in the event of some medical episode. Choosing a different hospital may
require a change of physician or of insurance plan. Even if patients could switch among hospitals
more easily, their incentives to search for cheaper hospital care are dulled by third-party payment,
and patients typically lack price and quality data that would be necessary for them to make a fully
informed choice. Much of the difficulty in instituting market-like reforms in the health care sector
stems from the nature of health care itself, and from the ways health care institutions have
evolved to deal with special features of health care. Improvements, while possible, would
probably be neither quick nor easy.
These characteristics, however, all point to some important conclusions. Prices as signals are
diluted and muted in the health care market as compared to many other markets. That muting of
price signals tends to suggest that improvements in price transparency may be less effective in the
health care market than in other markets and that this problem is particularly serious with hospital
pricing. At the same time, the lack of understandable price information in the health care market
may suggest significant room for improvement. To understand this last issue, it is important to be
clear about just how complex and dispersed hospital pricing is, an issue considered in the
following section.
ȱȱ
As the previous section suggests, the barriers to direct consumer choice are high for hospitals, and
it is for hospitals that many initiatives, discussed below, have been made to improve information
and transparency. Hospital costs are also a major portion of health care costs, accounting for 31%
of the $2 trillion of costs in 2005.23 To interpret and apply the evidence on price transparency
requires a more specific understanding of how hospitals set prices. This section provides an
overview of how hospitals set and administer prices. This section also investigates the variability
of hospital prices.
ȱȱȱ
Every hospital maintains a “chargemaster,” a document which lists prices for each item and
procedure offered by the hospital.24 A chargemaster may contain about 10,000 to 20,000 separate
items. By comparison, the U.S. tariff schedule has about 10,000 separate rate lines, and a regular
supermarket sells about 15,000 items.25 A Lewin Group study of hospital pricing practices found
that few hospitals in its sample conducted systematic reviews of their chargemasters.26 Many
23
Centers for Medicare and Medicaid Service, Office of the Actuary, National Health Statistics Group.
Uwe E. Reinhardt, “The Pricing of U.S. Hospital Services: Chaos Behind A Veil of Secrecy,” op. cit.
25
U.S. International Trade Commission, Simplification of the Harmonized Tariff Schedule of the United States,
Investigation No. 332-388, Publication 3318, June 2000; Food Marketing Institute Facts and Figures website, available
at http://www.fmi.org/facts_figs/superfact.htm.
26
Allen Dobson, Joan DaVanzo, Julia Doherty, and Myra Tanamor, “A Study of Hospital Charge Setting Practices,”
Lewin Group report no. 05-4, December 2005, available at http://www.medpac.gov/publications/contractor_reports/
Dec05_Charge_setting.pdf.
24
ȱȱȱ
ŗŗȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
hospitals stated that their charges had little relation to costs, although hospitals that were larger,
urban, or which conducted substantial amounts of research were more likely to report some link
between costs and chargemaster prices. Supplies and pharmaceutical charges appeared to be
reviewed more regularly and were more likely to be related to costs. Most hospitals in the Lewin
sample charged higher markups on less-expensive items.
Prices listed on the chargemaster bear little resemblance to what is actually paid. On average,
insurers and patients paid hospitals about 38% of their “charges” in 2004.27 Medicaid pays about
17% of total hospital revenues.28 Medicaid payment arrangements differ by state. All states use a
prospective payment system for Medicaid hospital reimbursement, with most either paying a flat
fee according to diagnosis related groups (DRGs) or paying a flat per diem rate. All states also
make special payments to hospitals for unusually high-cost cases, and most make payments to
hospitals that serve low-income or medically needy populations.29
Medicare pays a flat fee for inpatient care based on the average relative cost of a case within one
of about 600 DRGs. A DRG weight, reflecting the relative cost and complexity of a given
diagnosis code, is multiplied by a monetary conversion factor. Medicare payments are adjusted to
reflect differences in regional labor costs and some other local factors. Other adjustments are
made for outliers (extraordinarily complex cases with exceptionally high costs) and
“disproportionate share” adjustments made for hospitals that serve a larger than usual portion of
indigent patients. DRG weights are recalculated to account for changes in technology, practice
patterns, and other trends. Congress typically adjusts the monetary conversion factor each year.
From time to time, the Medicare Payment Advisory Commission (MEDPAC) proposes technical
changes in the definition of DRGs and in payment and adjustment details.
Private insurers are responsible for about a third of the hospitals’ revenues (hospital revenues
were $612 billion in 2005).30 Private insurers’ payment arrangements vary: some pay a fixed
portion of charges, some pay negotiated per diems or pay flat fees according to DRGs. Private
insurers typically use Medicare’s list of DRGs, but may assign their own weights. Medicare’s
calculations of DRG weights use claim experiences of Medicare beneficiaries, who are older than
the average private health plan policy holder, and so may not reflect relative costs for younger
patient populations. Private insurers vary in their ability to extract discounts from hospitals, and
arrangements between insurers and hospitals are tightly guarded trade secrets.
According to many analysts familiar with health care finance, Medicare and Medicaid payments
on average fall short of the fully allocated costs associated with patients in those programs.31
Thus hospitals must shift costs to private insurers, increase efficiency, or reduce services to
balance their books. As a result, payments for a particular patient’s case will reflect not just the
complexity of the case and the resources used, but also depend on the negotiating prowess of the
patient’s insurer.
27
Uwe E. Reinhardt, “The Pricing of U.S. Hospital Services: Chaos Behind A Veil of Secrecy,” op. cit., p. 57.
Ibid., p. 61.
29
Some specialized hospitals, such as psychiatric hospitals, are often reimbursed differently than general hospitals. For
a detailed description of Medicaid hospital reimbursement policy, see CRS Report RL32644, Medicaid Reimbursement
Policy, by Mark Merlis.
30
Ibid., for the share. Total hospital costs are from the Centers for Medicare and Medicaid Services, Office of the
Actuary, National Health Statistics Unit.
31
Allen Dobson, Joan DaVanzo, and Namrata Sen, “The Cost-Shift Payment ‘Hydraulic’: Foundation, History, And
Implications,” Health Affairs, vol. 25, no. 1 (2006), pp. 22-33.
28
ȱȱȱ
ŗŘȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
ȱȱȱ ȱ
Prices for specific items may vary wildly from one hospital to the next, as Figure 1 and Figure 2
show. For instance, a comprehensive metabolic panel, which costs $97 at San Francisco General,
costs $1733 at Doctors Hospital in Modesto, about 18 times more expensive. To some extent
disparate prices reflect different markup formulae, which act to allocate hospital overhead costs
among items.
Figure 1. Hospital Charges for Selected Diagnostic Tests
Doctors, M odesto
Sutter General,
Sacramento
West Hills Hospital,
West Hills
Cedars-Sinai, Los Angeles
Complete blood count
UC Davis, Sacramento
Comprehensive metabolic panel
Scripps M emorial, La
Jolla, San Diego
Chest X-Ray (Two views, basic)
San Francisco General,
San Francisco
$0
$200
$400
$600
$800
$1,000 $1,200 $1,400 $1,600 $1,800
Reproduced from Lucette Lagnado, “California Hospitals Open Books, Showing Huge Price
Differences,” Wall Street Journal, Dec. 27, 2004, p. A1. Data obtained from individual hospitals.
Source:
ȱȱȱ
ŗřȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
. Hospital Charges for Two Common Analgesics
Figure 2
Doctors, Modesto
Sutter General, Sacramento
$0.00
$35.50
$0.00
$26.79
$3.28
West Hills Hospital, West
Hills
Cedars-Sinai, Los Angeles
$27.86
$0.12
$6.50
$1.00
UC Davis, Sacramento
$15.00
Scripps Memorial, La
Jolla, San Diego
$7.06
San Francisco General, San
Francisco
$5.50
$6.68
$11.44
$0
$5
$10
Tylenol (or acetaminophen) one
tablet, 325 mg
Percocet (one tablet, 5-325 mg)
$15
$20
$25
$30
$35
$40
Source: See source for Figure 1.
Table 1 presents data on average costs and charges by type of payer for three hospitals, all located
in urban areas. In each case, the average charges for managed care patients were about 20%-30%
above average operating costs as reported to the Centers for Medicare and Medicaid Services
(CMS). By contrast, average charges for uninsured patients were substantially higher.
ȱȱȱ
ŗŚȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
. Average Costs and Charges for Selected Hospitals, By Type of Payer
Table 1
Palm Beach Gardens
Community Hospital
(Tenet Healthcare)
Palm Beach Gardens, CA
San Jose, CA
St. Louise Regional
(Catholic)
West Gilroy, CA
$1,631
$1,376
$1,501
$1,940
$1,773
$1,774
$5,951
.258
$5,508
.289
$7,414
.205
97%
96%
32%
O’Connor
Hospital
Avg. Operating Cost per
Patient per Day
Collected from Managed
Care
Billed the Uninsured
Cost-to-Charge Ratio
Collection Rate from the
Uninsured
Heartland Institute analysis of Centers for Medicare and Medicaid’s Medicare Cost Reports data for
2002. Reproduced from Randy Suttles and Merrill Matthews, Jr., “Hospital Pricing: Separate and Unequal,” Health
Care News, September 1, 2003.
Source:
Table 2 presents payment-to-cost ratios by type of payer for community hospitals from 1991 to
2000. The mix of services that each type of payer funds differs, which precludes direct
comparisons of payment rates across payers. Nonetheless, these data underline the point that the
relationship between costs and payments differs among payers. Average Medicare payments since
the mid 1990s nearly match hospital costs, while Medicaid payments, on average, fall short of
covering costs. The ratio of payments to costs is highest for private payers (i.e., private insurers
and managed care firms), although that ratio fell significantly during the 1990s. The ratio of
payments to costs is lowest for uncompensated care, although many hospitals receive subsidies
from state and local governments, not reflected in Table 2, that serve to defray expenses
associated with uncompensated care.
Data in Table 2 lump all uninsured payers together, although these payers include indigent
patients, from whom much smaller payments may be received, and non-indigent patients. We
were unable to locate aggregate data that would separate these two groups. However, an
illustration based on aggregate California data, provided in testimony by Glenn Melnick, shows
the importance of this distinction. The average chargemaster price for an appendectomy in 2002
was $18,229; the indigent uninsured paid $1,783, the Medicare payment was $4,805, the
managed care payment $6,174, and payments by the non-indigent uninsured was $8,143.32 The
payment from the non-indigent, indeed, did fall below the list price, and may reflect both ad hoc
discounts and failure to collect the payment. Even so, the uninsured non-indigent paid a third
more than the managed care patients and 70% more than Medicare patients for this procedure.
Melnick also points out that the list price remains important not only because some uninsured
patients are charged the list price, but also because of stop-loss provisions in contracts (where list
is paid above a threshold), lack of contracts with all third party providers, and out-of-network use.
He also points out that increasing revenues is an incentive to charge a high list price.
32
Testimony of Glenn Melnick, in U.S. Congress, House Committee on Ways and Means, Subcommittee on Health,
March 9, 2004.
ȱȱȱ
ŗśȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
. Hospital Payment-To-Cost Ratios By Type of Payer, 1991-2000
Table 2
Year
Uncompensated
Care
Medicare
Medicaid
Private Payers
1991
88.4%
81.6%
19.6%
129.7%
1992
88.8
90.9
18.9
131.3
1993
89.4
93.1
19.5
129.3
1994
96.9
93.7
19.3
124.4
1995
99.3
93.8
18.0
123.9
1996
102.4
94.8
17.3
121.5
1997
103.6
95.9
14.1
117.6
1998
102.6
97.9
13.2
113.6
1999
101.1
96.7
13.2
112.3
2000
100.2
96.1
12.1
112.5
MEDPAC analysis of American Hospital Association data. Reproduced from Table B-11 in Medicare
Payment Advisory Commission, Report to the Congress: Medicare Payment Policy, March 2002. Data for years
2001 and after are unavailable.
Source:
Some hospitals have been strongly criticized for charging uninsured patients, who typically have
less ability to pay for care than insured patients, far higher prices. In some apparently isolated
circumstances, news stories detailing some hospitals’ attempts to use aggressive collection
methods against uninsured patients purportedly caused those hospitals to cancel those debts.33
Although some hospitals and hospital associations have argued that some federal regulations
prohibit hospitals from offering discounts and fee waivers on a case-by-case basis,34 the Centers
for Medicare and Medicaid Services (CMS) contends that no federal law prevents hospitals from
reducing or waiving charges for an indigent uninsured patient so long as such reductions or
waivers conform to the hospital’s indigency policy. Moreover, the CMS Inspector General has
stated that it is “highly unlikely” that hospitals that waived charges to indigent uninsured patients
would run afoul of the federal anti-kickback statute.35
More detailed analysis of hospital charge and cost data shows that uninsured and self-pay patients
are charged, when confronted with the full list price, on average, about 2½ times more than what
insurers pay hospitals, and about three times Medicare-allowable costs.36 The gap between what
uninsured and self-pay patients pay and what insurers pay hospitals appears to have widened
since the mid 1980s.
33
Randy Suttles and Merrill Matthews, Jr., “Hospital Pricing: Separate and Unequal,” Health Care News, September 1,
2003, available at http://www.heartland.org/Article.cfm?artId=12775.
34
Metropolitan Chicago Healthcare Council, Report of the Task Force on Charity Care and Collection Practices of the
Illinois Hospitals Association and the Metropolitan Chicago Healthcare Council, September 11, 2003, available at
http://www.pfho.org/HousingInitiatives03/charity%20care-hospitals.pdf.
35
Centers for Medicare and Medicaid Services, “Questions On Charges For The Uninsured,” available at
http://www.cms.hhs.gov/AcuteInpatientPPS/downloads/FAQ_Uninsured.pdf.
36
Gerard F. Anderson, “From ‘Soak The Rich’ To ‘Soak The Poor’: Recent Trends In Hospital Pricing,” Health
Affairs, vol. 26, no. 3 (May/June 2007), pp. 780-789.
ȱȱȱ
ŗŜȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
ȱȱ ȱȱȱȱȱȱǵȱ
Chargemaster prices charged by different hospitals for the same procedure can vary wildly, as
noted in Figure 1 and Figure 2. Actual charges for specific procedures, which are generally
lower than chargemaster prices, also vary widely, although information on them is unavailable.
Chargemaster prices are nevertheless important, because they are prices billed to uninsured
patients who do not have discounts, and are the starting point for discounted prices.
Figure 3 shows the distribution of average charges per stay for normal vaginal birth for
California hospitals in 2004, which aside from newborn care is the most common DRG.37 (The
discrete data are converted into a smoothed curve using a technique called kernel density
estimation.) Average charges at the mode of the smoothed distribution lie between $5,000 and
$10,000. The distribution has a fat right-hand tail, indicating more variation on the high side of
charges than on the low side.
Figure 4 shows average charges per stay by hospital for heart failure and shock in 2004. Unlike
the conditions which occur at birth, heart attack victims have substantially less time to plan their
hospital stay. Indeed, the typical heart attack victim has no time to select a hospital or to consult
with his physician about treatment options. Further, because of variation in the time spent in
hospitals, the variation of total charges per episode is much greater in the case of heart attacks,
with a handful of hospitals having average charges above $500,000.
37
Details of the California price transparency initiative are discussed below. Because of the way Kaiser-Permanente
health plans are structured, Kaiser hospitals did not report charges and are therefore excluded from the analysis.
ȱȱȱ
ŗŝȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
Figure 3. Distribution of Average Charges Per Stay for Normal Vaginal Birth
.00012
.0001
Density
.00008
.00006
.00004
.00002
.00001
4000
6000
8000
10000
12000
16000
20000
24000
28000
Average Charges Per Stay (Current Dollars)
State of California, Office of Statewide Health Planning and Development, Healthcare Information
Division. Data are smoothed using two related methods, one yielding the jagged line and the other the smooth
line. Both lines represent kernel density estimates. For the jagged line the kernal halfwidth is set at $300. For the
smooth line the kernal halfwidth is set to minimize mean square error for Gaussian distributions. Unit of
observation is the hospital.
Source:
ȱȱȱ
ŗŞȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
. Distribution of Average Charges Per Stay for
Heart Failure
Figure 4
.000035
.00003
Density
.000025
.00002
.000015
.00001
15000
25000
35000
45000
55000
65000
75000
100000
125000
140000
Average Charges Per Stay (Current Dollars)
Source:
See source for Figure 3. Dotted line represents kernal density estimator with halfwidth of $1,500.
ȱȱ ȱȱȱ
Does evidence on the effects of price transparency in other markets, which by and large supports
the view that better information on pricing reduces prices, imply that greater price transparency
would affect health markets in the same way, despite the specific structures and characteristics of
the health care market? Of course, how the examples are applied depends on how pricing
information is provided. For example, allowing the public to examine charge books and data on
actual average charges at a hospital’s finance office provides more limited access than posting
that information on the Internet. Additional information could be conveyed by providing
information on the pricing of complete, but typical, procedures as well. (Selective reporting could
provide opportunities for hospitals to game the system by lowering costs on the reported
procedures and raising costs on others.) Even more information could be conveyed by also
reporting prices for the different categories of patients (Medicare, Medicaid, uninsured, or insured
by specific health plans).
One caution is that prices of goods commonly sold on the Internet also show substantial price
variation, although the degree of price dispersion may be less than for average daily hospital
charges. For example, Figure 5 shows prices for a Samsung HP-R6372 high definition television
found using two common Internet price search engines, Froogle.com and Pricescan.com. The
mode of the smoothed distribution is slightly less than $6,000, which is slightly less than the
mode for daily average charges for normal vaginal birth. However, the right-hand tail of the
distribution of TV prices is not nearly as fat as the distribution of average daily charges for
normal vaginal birth. (Note: horizontal scales differ for each figure.) Only one seller listed a price
for the Samsung TV above $8,000, and 37 of 47 sellers posted prices between $5,500 and $7,500.
ȱȱȱ
ŗşȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
By contrast, 10 of 251 California hospitals charged more than $9,000 per day and 33 charged less
than $3,000 per day.
. Price Distribution for a Samsung HP-R6372 HDTV
0
.0002
Density
.0004
.0006
.0008
.001
Figure 5
4000
6000
8000
Price
10000
12000
Searches conducted at http://www.froogle.com and http://www.pricescan.com on August 9, 2006. Price
distributions smoothed using same methods as described in Figure 3.
Source:
A formal way of comparing variability of distributions with different averages is to compute
coefficients of variation. Table 3 presents statistical estimates of price variability for two of the
most common types of hospital episode as well as for a similarly expensive consumer good,
namely, a particular HDTV. The coefficient of variation is a dimension-free measure, and thus is
an appropriate tool for comparing different distributions.38 As expected, the coefficients of
variation for average hospital charges for normal birth and for heart attacks are substantially
greater than for the Samsung television.
The comparison of HDTV prices advertised by retailers and average charges per day for a given
DRG is not an “apples-to-apples” comparison. Prices advertised by retailers do not necessarily
represent actual sales prices. Posted prices on the Internet may vary considerably, even if prices at
those websites that make most of the sales vary less.39 To the extent that most units are sold by
sellers with prices near the minimum posted price, sales-weighted measures of price variability
will be less than unweighted measures.
38
The coefficient of variation is the standard deviation divided by the mean (average). Among hospitals that reported
charges for both procedures the coefficient of variation for charges per stay for Heart Failure and Shock (DRG 127)
was 0.678 and 0.420 for Vaginal Delivery w/o Complicating Diagnoses (DRG 373).
39
Erik Brynjolfsson and Michael D. Smith, “Frictionless Commerce? A Comparison of Internet and Conventional
Retailers,” Management Science, vol. 46, April 2000, pp. 563-585.
ȱȱȱ
ŘŖȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
.Variability of Average Hospital Charges and Samsung HDTV Prices
Table 3
Average
Vaginal Delivery w/o Complicating
Diagnoses
Heart Failure and Shock
per day
per stay
per day
per stay
Samsung HP-R6372 HDTV
Source:
$5,280
$10,350
$5,696
$36,840
$6,254
Standard
Coefficient of
Deviation
Variation
$1,933
$4,286
$3,451
$31,273
$1,253
0.366
0.414
0.606
0.849
0.200
N
242
393
47
See preceding figures. Data are for 2005.
We checked whether combined average daily charges for mother and baby varied less than
average charges for normal vaginal birth alone, which could occur if different hospitals allocated
charges to mother and baby differently. The coefficient of variation for the sum of average daily
charges for normal newborn and normal vaginal birth, however, was about the same (.356) as for
normal birth alone (.366). The coefficient of variation for average stay charges for the sum of
normal birth and normal newborn care (.416) was nearly the same as for normal birth alone
(.414).
These illustrations are just examples of pricing variability and do not constitute a statistically
valid universe. Nevertheless, they do indicate considerably more price variability for medical
procedures than for an expensive consumer durable that might be expected to show much more
variation than more frequently purchased commodities. They also show much more variation for
an unanticipated procedure (heart failure) than for an anticipated one (birth). They are suggestive,
therefore, of a considerable amount of price variability in hospital costs.
ȱ¢ȱ ȱȱ ǰȱ
ǰȱȱ ȱ ȱ
Several states have enacted regulations intended to enhance price transparency in the health
sector in general, and hospital pricing in particular. Several private insurers also allow
policyholders access to online tools that allow some price comparisons for medical procedures.40
California has required hospitals to provide a variety of pricing data to the public, discussed in
more detail below. Average hospital charges per day and per stay for selected DRGs are available
on state government websites sponsored by Arizona, California, Florida, Maryland, and
Massachusetts.41 Other states, such as Iowa, New Hampshire, and Wisconsin, in cooperation with
state hospital associations, provide some pricing information.
Aetna has published price information for physicians and hospitals in the Cincinnati area, and
recently extended this program to other parts of the country. Other insurers, including Cigna,
40
National Conference of State Legislatures, “State Legislation Relating to Disclosure of Hospital and Health
Charges,” April 2007, available at http://www.ncsl.org/programs/health/Transparency.htm#Table1.
41
The National Conference of State Legislatures has links to the state websites, along with many private insurers’
websites: http://www.ncsl.org/programs/health/Transparency.htm.
ȱȱȱ
Řŗȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
Humana, United HealthCare, and Wellpoint have created websites that provide price comparison
data for certain procedures.42
Hospitals also submit data to the Centers for Medicare and Medicaid Services (CMS), which
compiles annual Medicare Cost Reports (MCR). The MCRs contain extensive information about
hospitals’ cost structures and finances. These reports, which are quite large and complex, are
available for download on the CMS website. The website HospitalVictims.com provides cost data
for individual hospitals derived from Medicare Cost Reports, suggesting that hospitals with high
charge-to-cost ratios be avoided, or that the patient negotiate for a discount.43 It suggests that a
high charge-to-cost ratio is evidence of a significant amount of price discrimination and a
likelihood that the uninsured patient will be charged a high price.
Initiatives to impose price transparency requirements on hospitals, such as allowing the public to
inspect chargemaster data or have access to average daily charges data, were motivated in part by
a desire to allow consumers to make informed choices about selecting hospitals. Better
information on prices, according to this view, would increase competitive pressure on hospitals,
slowing the growth of hospital prices and reducing price variability. These initiatives are
relatively new and do not yet appear to have had significant effects on the level and dispersion of
medical costs.
In August 2006, Executive Order 13410 called for greater transparency of quality and price
information and for more widespread use of information technology in federal health care
programs using compatible data standards.44 The executive order also directed federal agencies to
develop health care quality measurement programs. The National Coordinator for Health
Information Technology within the U.S. Department of Health and Human Services oversees
these initiatives, although other offices also have major responsibilities.
The executive order directed federal agencies to “make available ... to the beneficiaries or
enrollees of a Federal health care program (and at the option of the agency, to the public) the
prices that it, its health insurance issuers, or its health insurance plans pay for procedures to
providers.” It was reported that the Bush administration earlier in 2006 declined to release
Medicare claims data to the Business Roundtable, which had requested them.45 The Business
Roundtable is an association of chief executives of very large corporations. Whether the order
signifies a change in policy or there was another reason for not releasing these data remains
unclear.
The effects of the executive order on pricing information are unknown, including how widely
available the information is, since the implementation of the order is in its early stages. But it is
an example of another government initiative to provide more information about pricing.
In some areas, the initiatives outlined by the Executive Order parallel ongoing efforts. Several
federal agencies have already taken some measures to provide federal health care users with
42
Ibid.
See http://www.hospitalvictims.com/.
44
Executive Order 13410, “Promoting Quality and Efficient Health Care in Federal Government Administered or
Sponsored Health Care Programs, August 22, 2006. Additional information is available at the Department of Health
and Human Services’s “Value-Driven Health Care” website: http://www.hhs.gov/transparency/fourcornerstones/
index.html.
45
Robert Pear, “Employers Push White House to Disclose Medicare Data,” New York Times, April 11, 2006.
43
ȱȱȱ
ŘŘȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
better price and quality information. For example, the Federal Employees Health Benefit Program
(FEHBP) provides a website that compares premiums, plan details, and customer satisfaction
measures for all plans.46 Changes made in response to the order appear minor.47 CMS (Centers for
Medicare and Medicaid Services) now publishes summarized inpatient price data for the 30 most
common elective DRGs for individual hospitals. These data are taken from Medicare Provider
Analysis and Review (MEDPAR) data, which has been collected since 1991. However, locating
this information on the CMS website may be difficult for consumers since the website covers a
range of material.48
Medicare’s “Hospital Compare” website, accessed via the Medicare.gov site, allows beneficiaries
to see data that compares how closely different health care providers follow accepted treatment
protocols.49 Whether such initiatives give consumers enough relevant information in an easily
accessible way, whether patients and their families would be able to locate such information, and
whether such information would motivate patients to make major changes in their treatment plans
is unclear. One consulting firm concluded that while a previous version of the Hospital Compare
website “does an average job of presenting the quality information, it lacks the robust data found
in commercially available products and leaves consumers fumbling with insufficient help.”50 In
June 2007 a redesigned HospitalCompare website was launched that allows limited comparisons
of hospital mortality rates for heart failure and heart attacks with national mortality rates.51
However, nearly all hospitals were judged to have mortality rates “no different than the U.S.
national rate.” Out of almost 4,500 hospitals, only 17 were recognized as “above average” in
treating heart attacks and only seven were rated “below average.”52
On the other hand, more information about health care provider quality and pricing is becoming
available on the Internet, and these sources will continue to evolve. A major review of
information available on websites that provide hospital price and quality information expressed
concern that some consumers might be confused rather than enlightened by the reported data, but
also noted that momentum continues to build for making health care data more easily available to
consumers.53
The following section of this report presents an analysis of the California price transparency
initiative’s effect on the dispersion of hospital prices.
46
http://www.opm.gov/insure/health/
The Office of Personnel Management’s efforts to implement these initiatives are described on its website, available
at http://www.opm.gov/insure/health/executiveorder.asp.
48
http://www.cms.hhs.gov/HealthCareConInit/02_Hospital.asp#TopOfPage
49
http://www.cms.hhs.gov/HospitalQualityInits/25_HospitalCompare.asp
50
Katy Henrickson, Hospital Comparison Tools Scorecard Summary: Centers For Medicare And Medicaid Services:
Key Findings From “The Forrester Wave: Hospital Comparison Tools, Q4 2005,” October 13, 2005, available at
http://www.forrester.com/Research/Document/Excerpt/0,7211,37929,00.html.
51
Hospital Quality Alliance, “HQA Adds Enhanced Hospital Quality Information to ‘Hospital Compare’ Web Site,”
June 22, 2007, available at http://www.fah.org/issues/quality_initiative/
HQA%20Adds%20Enhanced%20Hospital%20Quality%20Information%20to%20Hospital%20Compare%20Web%20S
ite.pdf.
52
Michael S. Gerber, “Hospitals Are Beyond Compare: Data on Cardiac Care Show Almost No Differences
Nationwide,” Washington Post, July 3, 2007.
53
Delmarva Foundation, The State-of-the-Art of Online Hospital Public Reporting: A Review of Fifty-One Websites,
Report Prepared for the CMS Hospital Three State Pilot Project, July 2005. Available at
http://www.delmarvafoundation.org/newsAndPublications/pressReleases/2005/08_18_05.pdf.
47
ȱȱȱ
Řřȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱȱ¢ȱȱȱ¢ǵȱȱ
¢ȱȱ
The California hospital price transparency initiative, according to analysis of available data, has
had negligible or no observable effect on hospital prices.
In September 2003, California legislators passed Assembly Bill 1627,54 which required hospitals
(except for certain small and rural hospitals) to make chargemaster data public by July 1, 2004,
either in electronic form or by allowing onsite inspection.55 Sponsors of this bill contended that
these reporting requirements would prevent hospitals from “gouging” customers and would make
patients into better-informed consumers.56 In July 2005, hospitals had to begin submitting
chargemaster data to the Office of Statewide Health Planning and Development Healthcare
Quality & Analysis Division (OSHPD). In 2004 and 2005, hospitals also had to list charges for 25
common services or procedures. In 2006, hospitals were required to submit data on average
charges for 25 common diagnosis-related groups (DRGs). The state of California makes these
data available online.
If patients became better-informed customers, most economists would expect that hospitals that
raised their prices more would lose patients, unless there were offsetting increases in the quality
of medical care or level of amenity.57 That is, if customers are sensitive to and aware of prices,
increases in hospital prices would be negatively related to changes in hospital admissions, other
things equal.
If consumers were becoming more sensitive to price as a result of greater price transparency, then
one might expect to see stronger effects for procedures for which patients can plan ahead.
Expectant mothers planning for a normal vaginal birth can compare what various hospitals have
to offer and their prices, unlike victims of sudden medical emergencies. Figure 6 shows the
distribution of average daily charges (adjusted for general inflation) for normal vaginal birth at
California hospitals in 2003, 2004, 2005, and 2006. Over this time period, the modal (i.e., most
frequent) nominal price drifts upwards because average daily charges have been rising faster than
the general price level. The distribution of prices shows no signs of convergence.58
54
Helen Sanderson, “Cost of Care: New Law Lets Patients Examine Hospital Price Lists,” North Coast Weekly, August
11, 2005, available at http://www.northcoastjournal.com/081105/cover0811.html.
55
The Wall Street Journal article, referenced in Figure 1 above, was based on these chargemaster data. For further
details on these requirements, see the State of California’s Office of Statewide Health Planning and Development
Healthcare Information Division’s website at http://www.oshpd.ca.gov/HID/Products/Hospitals/Chrgmstr/
FAQ.html#Q2.
56
California State Assembly, “Bill Analysis for Assembly Bill 1627,” available at http://info.sen.ca.gov/pub/03-04/bill/
asm/ab_1601-1650/ab_1627_cfa_20030527_173625_asm_floor.html
57
This also presumes that demand factors, such as demographics and income, were held constant. Changes in these
factors over the period analyzed here are likely small relative to changes in hospital prices.
58
The same analysis was performed for average charges per stay, rather than average daily charges. The results were
nearly identical.
ȱȱȱ
ŘŚȱ
ȱ
. Distribution of Average Charges Per Stay For Normal Birth, 2003-2006
Figure 6
State of California’s Office of Statewide Health Planning and Development, Healthcare Information Division
Note: Kaiser hospitals submitted no average charge data, and so are excluded. GDP Price Index used to convert charges into 2006 dollars. Also see notes for Figure 3.
Source:
ȬŘśȱ
ȱ
. Scatter Plot for Changes in Avg. Daily Charges and Discharges for Normal Birth (DRG 373)
Figure 7
State of California’s Office of Statewide Health Planning and Development, Healthcare Information Division.
Notes: GDP Price Index used to convert charges into 2006 dollars. Kaiser hospitals, which submitted no average charge data, are excluded.
Source:
ȬŘŜȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
Hospitals that had increased average daily charges for normal vaginal birth, on average, did not
lose patients. Figure 7 (above) presents a scatter plot with percentage change in hospital
discharges on the vertical axis and percentage change in average daily charges on the horizontal
axis. Different plotting symbols divide hospitals into four categories defined by the number of
(normal) births in 2003. If expectant mothers avoided hospitals that raised their prices, then a
downward-sloping relationship would be evident between the two variables. Regression analysis
shows a statistically significant, albeit small, positive relationship between changes in average
charges and changes in hospital volume for normal births over the 2003-2006 period.59
Several explanations are possible for the lack of a discernable relationship between changes in
average charges and changes in hospital volume. Differences in perceived quality or care or
amenity levels may matter more than price for many patients, especially if insurance coverage
insulates them from prices. Patients’ relationships with their physicians and those physicians’
relationships with hospitals might reduce patients’ sensitivity to hospital prices. Alternatively,
patients may care about prices, but might be unable, unwilling, or disinclined to examine online
price data. Finally, changes in prices might correlate to offsetting changes in quality or amenity
levels. Distinguishing among these explanations would require more sophisticated data. However,
the available evidence, while preliminary, suggests that the California price transparency
initiative so far has had little observable effect where it might have been expected to have the
greatest effect.
ȱȱ ȱȱ¢ȱȱȱ
ȱȱǵȱ
The experience of the Danish Competition Authority, noted above (in the section titled “Empirical
Evidence on the Effects of Price Transparency”), suggests that imposing price transparency in
negotiations between sellers and buyers of intermediate goods does not necessarily lead to
sharpened competition or lower prices. At the same time, some evidence suggests that
information about the process of setting prices, including practices of price discrimination, may
produce a change in pricing, as in the NASDAQ and Amazon cases. Much of the remaining
evidence also suggests that transparency lowers prices and makes them more uniform.
The evidence cited on price transparency involves two types of effects: a response through
publicity effects and a response through normal market mechanisms. The price discrimination
that occurs in hospitals—brought about partly by government policies with respect to Medicare
and Medicaid, partly due to bargaining power of insurance companies (and the desire to set high
list prices to leave room for discounting),60 and partly through providing free care for the
indigent—leads to potentially high prices for a small segment of uninsured individuals. As more
of these pricing differences are revealed and spotlighted, public opinion might force a reduction
in cross subsidies (charging some patients higher prices to cover the costs not fully applied to
59
One outlier was excluded. An indicator variable for Tenet Healthcare hospitals, which Bill 1627’s sponsors claimed
had aggressively sought to increase hospital prices, was statistically insignificant. Regression results available upon
request.
60
See the discussion in Lucette Lagnado, “California Hospitals Open Books, Showing Huge Price Differences,” Wall
Street Journal, December 27, 2004, p. A1.
ȱȱȱ
Řŝȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
other patients).61 Indeed, such a response has already occurred. An example is the case of the state
of Minnesota, which entered into a voluntary agreement with most of its hospitals to limit the
charges for uninsured patients. Under the agreement, uninsured patients with $125,000 or less in
annual income would pay no more than the amount paid for the procedure by the private
insurance company that provided the greatest amount of the hospital’s revenue.62
The Minnesota example suggests that publicity can affect pricing. What about the effects through
normal market mechanisms? The survey of evidence included cases where price transparency did
not affect prices, or in some instances, led to higher prices. In addition, many of the studies
analyzed goods that lack the special characteristics of health care. These shortcomings do not,
however, necessarily mean that price transparency in the health sector would not be beneficial.
First, the ready-mix concrete example, in which price information resulted in higher prices also
involved for-profit businesses, which presumably were attempting to maximize profits. This
example may not apply to many hospitals that are non-profit and may have different behavioral
responses.63
Second, the evidence from the advertising studies includes not only simple uniform goods such as
alcoholic beverages and gasoline, but complex differentiated products such as vision exams,
where quality matters. And while insurance pays much of the cost of medical care, the studies
summarized above also included examples of price reductions for prescription drugs after direct
advertising to consumers, whose prices are also subject to third party payment.64 In these cases as
well, the evidence suggests that prices fell after advertising was permitted, without deterioration
in quality.
Third, evidence from the Internet suggested that price comparison sites may help reduce
commodity prices, including differentiated commodities that are subject to bargaining
(automobiles). Over time, price comparison websites have become more sophisticated and are
playing an increasingly important role in consumer behavior in many markets.
61
Cross subsidies are intra-firm transfers that support some activities or lines of business using net revenues earned in
other activities or lines of business.
62
Minnesota Hospital Association and Office of Minnesota Attorney General Mike Hatch, “Hospitals Step Forward to
Sign Voluntary Agreement with Attorney General’s Office on Billing and Collection Practices,” June 2, 2005.
63
Objectives of not-for-profit hospitals are an important area of research in health economics, and although there has
been a presumption that any surplus might be used for charitable care, evidence on that issue is mixed. While providing
a complete review is beyond the scope of this report, see for example, Richard G. Frank and David S. Salkever, “The
Supply of Charitable Services by Nonprofit Hospitals: Motives and Market Structure,” RAND Journal of Economics,
vol. 22, autumn 1991, pp. 430-440; and Edward C. Norton and Douglas O. Staiger, “How Hospital Ownership Affects
Access to Care for the Uninsured,” RAND Journal of Economics, vol. 25, spring 1994, pp. 171-185. Mark Pauly once
described hospitals as “doctors’ workshops,” whose decisions were made with an eye to maximizing the welfare of
physicians. Since that time hospital administrators reportedly have become more professional and more powerful. See
Mark Pauly, Doctors and Their Workshops: Economic Models of Physician Behavior, (Chicago: Univ. of Chicago
Press, 1980). In any case, non-profits’ objectives may differ from those of for-profit firms. Some evidence suggests that
non-profits maximize output or revenues; e.g., Richard Steinberg, “The Revealed Objective Functions of Non Profit
Firms,” RAND Journal of Economics, vol. 17, winter 1986, pp. 508-526.
64
John F. Cady, “An Estimate of the Price Effects of Restrictions on Drug Advertising,” Economic Inquiry, vol. 44,
December 1976, pp. 493-510; Steven W. Kopp, “Direct-To-Consumer Advertising and Consumer Prescription Prices,”
Drug Information Journal, vol. 30, 1996, pp. 59-65.
ȱȱȱ
ŘŞȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
ȱȱȱȱ
The Internet has begun to affect the availability of price information in the health care sector,
although this does not appear to have influenced a large proportion of consumers. According to
the New York Times, 32 states require hospitals to publish price information.65 Some new websites
provide consumers with data on health care costs. For example, Vimo http://www.vimo.com
provides information on average list prices and average negotiated prices charged by hospitals for
specific procedures. One company, My Medical Control, provides a negotiation service for
consumers through its website http://www.mymedicalcontrol.com. A consumer forwards a bill,
via the website, to a claims adjuster who negotiates a reduced rate with the provider. This
company then deducts a 35% fee and returns the remainder to the consumer. At present, such
websites have little observable effect on health care markets. In the future, however, such sites
could have large effects.
The Carol.com website allows consumers to compare prices and offerings of health providers in
the Twin Cities region of Minnesota, and in addition it allows them to book services. The
intention of the website’s creators is to follow the example of web-based booking services such as
Expedia.com, which have transformed the travel industry in the past decade. Many state
governments have opened their own sites that allow consumers to compare prices or provider
characteristics.66
The effect of information on quality is much more difficult to obtain, and it is hard to make a
judgment based on the available evidence. As noted above, the United Network for Organ
Sharing publishes risk-adjusted outcome data on its website http://www.unos.org. Some other
organizations, also noted above, also publish some data reflecting quality of health care.
ȱȱ ȱȱȱȱȱ ǵȱ
One of the most important differences between hospital care and other commodities is that
typically patients pick physicians and physicians pick hospitals. Although this characteristic
means that direct consumer pressure to hold down prices (or at least have a sensible pricing
system) is more difficult, it does not mean that physicians would not become more sensitive to
differences in costs among various hospitals on behalf of their patients, particularly if their
patients raise questions about these costs. Not everyone in a market is required to be attentive to
price for pressure to be exerted at the margin. Moreover, publicity about price differentials may
result in voluntary compliance by hospitals. Nevertheless, this aspect of the delivery of hospital
services makes it more difficult to apply evidence from other markets to the expected outcome of
introducing more price transparency in health care markets.
Changes in the airline industry might provide some insight into how increased price transparency
and competition could affect the hospital industry. While the air travel and hospital industries
have important structural differences, airlines, like hospitals, have high fixed costs and offer a
non-storable product. Before the Airline Deregulation Act of 1978 (P.L. 95-504), airlines
competed largely on the basis of amenity levels rather than on price. The Airline Deregulation Act
65
Michael Mason, “Bargaining Down That CT Scan Is Suddenly Possible,” New York Times, February 27, 2007.
Many of the state government health comparison and information sites are listed at
http://www.healthtransformation.net/cs/leading_examples.
66
ȱȱȱ
Řşȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
restricted the Civil Aviation Board’s price administration powers, and led to the abolition of the
board in 1984. After deregulation, several new airlines entered the market, while several major
airlines went bankrupt and exited the market. Increasing competitive pressure led airline
companies to cut back amenities to passengers and led to contentious negotiations with labor
unions that resulted in sharply reduced wages in many cases.67 Employees with highly specialized
skills, such as pilots and mechanics, appeared to fare better in resisting wage and salary
reductions compared to other employees. Air service to some small cities, supported by implicit
cross subsidies, ceased, while service to some other small cities expanded, in part because some
airlines found ways to serve such markets at lower cost. Lower fares (in real terms) led to an
enormous expansion in air travel and increases in air travel employment. Some relatively new
airlines, such as Southwest Airlines, prospered and expanded, while other airlines struggled,
including several major carriers that declared bankruptcy.
Were price transparency to improve, and if consumer choice in health care were to become more
sensitive to price differentials, then economic analysis would suggest that these effects would
increase pressure on hospitals to become more productively efficient, that is, to use fewer inputs
to produce the same or greater output. Cost-cutting measures would put pressure on health sector
salaries and wages, which some occupational groups would resist more successfully than others.
Services, such as indigent care, now in part supported by implicit cross subsidies, could face
cutbacks unless direct subsidies to support such services were increased. Innovative providers,
however, may find ways to expand access to health care by the indigent using more efficient and
cheaper methods. Some prices might fall, along with amenity levels. Lower prices, in turn, could
expand access to health care, and to the extent that demand for medical procedures is sensitive to
price, could expand the volume of medical services provided. Some existing health care
providers, especially those unable to change their cost structures and operating procedures
quickly, would be at a comparative disadvantage to more nimble providers. Such changes would
produce both winners and losers, just as airline deregulation produced winners and losers.
Increased price transparency, however, to the extent that it allowed health care markets to
function more efficiently, would be expected to generate more gains than losses.
67
U.S. Government Accountability Office, Airline Deregulation: Reregulating the Airline Industry Would Likely
Reverse Consumer Benefits and Not Save Airline Pensions, GAO-06-630, June 2006.
ȱȱȱ
řŖȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
¡ǯ ȱȱȱȱȱȱ
¢ȱ
ȱȱȱȱȱ
The effects of price transparency on how financial markets function depend on how those markets
are set up. Financial exchanges are structured as auction markets or dealer markets.68 In an
auction market, such as the New York Stock Exchange (NYSE), investors send orders to a
specialist, who coordinates trading for a particular stock. Investors can send market orders, which
are to be executed immediately for the best possible price, or limit orders, which instruct a broker
to buy a stock at a set price or to sell a stock at a set price. The specialist executes market orders
by matching them with orders from the other side of the market, or by buying or selling on his
own account. Limit orders that are not executed are entered into the specialist’s order book. In a
dealer market, such as NASDAQ, orders for a particular stock flow to market makers who then
post bids (i.e., prices at which buyers are willing to trade) with asks (i.e., prices at which sellers
are willing to trade) via an electronic market. In NASDAQ each stock must have at least two
market makers, and for major stocks there may be 30 or more market makers.
Price transparency can mean several things in financial markets. The most basic form of price
transparency is the timely reporting of executed trades. A second form of transparency is
information about outstanding limit orders listed in a specialist’s order book. Order book
information can signal impending price movements, and a trader with special knowledge about
outstanding orders can make profits. For example, an order book with many buy orders just above
the market price and very few sell orders may signal that the market price is about to rise, and a
specialist who buys before that rise occurs will reap profits. A third form of transparency concerns
information about how dealers or specialists handle orders. A dealer often has some discretion in
how and when orders are executed and may sometimes exploit that discretion to earn profits at
the expense of the investor who placed the order.
Past NASDAQ pricing practices illustrate the importance of the third form of price transparency.
In 1994, William Christie and Paul Schultz, two Vanderbilt University financial economists,
noticed that NASDAQ dealers almost never quoted prices using odd eighths (i.e., 1/8, 3/8, 5/8,
and 7/8) for many high-volume stocks of companies such as Microsoft, Intel, and Apple. This
practice effectively created a quarter dollar minimum spread between sellers’ asks and buyers’
bids, which increased the trading profits of dealers. The day after these economists issued a press
release about their findings the practice was abandoned, and spreads for several major stocks fell
by about half.69
68
For a more detailed description of the structure of modern financial markets, see Hans R. Stoll, “Electronic Trading
in Stock Markets,” Journal of Economic Perspectives, vol. 20, no. 1 (winter 2006), pp. 153-174.
69
William H. Christie and Paul H. Schultz, “Did NASDAQ Market Makers Implicitly Collude?,” Journal of Economic
Perspectives, vol. 9, summer 1995, pp. 199-208. The U.S. Department of Justice (DOJ) concluded that the even-eighths
convention was “vigorously enforced through industry-wide peer pressure, and intimidating telephone calls to, and
refusals to deal with, market makers who did not quote bid and ask prices in conformance with the convention.” See the
DOJ “Competitive Impact Statement,” U.S. v. Alex Brown & Sons Inc., et al., U.S. District Court for the Southern
District of New York, pp. 16-17. Also see U.S. General Accounting Office, Security Market Operations: The Effects of
SOES on the Nasdaq Market, GAO/GGD-98-194, August 1998.
ȱȱȱ
řŗȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
Economists often argued that collusion is difficult or impossible with large numbers of traders. A
more careful argument is that collusion depends on the ability to make explicit or implicit
agreements, and maintaining agreements may be more difficult for larger groups. Should a large
group of sellers collude, each seller has strong incentive to increase his or her market share by
making small reductions in price. If a seller can reduce its price and increase sales without other
sellers noticing, then it will reap extra profits. For example, many members of the Organization of
Petroleum Exporting Countries (OPEC) have been suspected of making hidden side deals which
allow them to sell more oil than their OPEC quotas specify, which may have led to softened oil
prices.
In the case of NASDAQ, it appears that a very simple rule—no trading on odd eighths—created
artificially high trading spreads, which allowed dealers to reap higher profits. Young traders
reportedly were cautioned not to narrow inside spreads by using odd-eighths, and traders who
violated the no-odd-eighths convention may have been subject to intimidation or isolation. In
addition, securities dealers rely on trades with other dealers to rebalance their inventories of
stocks in order to minimize financial risks associated with sudden price movements. If other
dealers refused to trade with a dealer who violated the pricing convention, then that dealer would
be exposed to higher levels of financial risk. Furthermore, the practice of “preferencing” among
NASDAQ dealers, which involves guaranteeing flows of orders to preferred dealers or dealer
subsidiaries, meant that order flows were less sensitive to spreads. A dealer who violated the
pricing convention in order to attract order flow would therefore gain little additional market
share because of existing “preferencing” arrangements, which in turn reduced incentives for
dealers to compete by narrowing spreads.
While no evidence was found that this pricing convention was the result of an explicit collusive
agreement, that convention enhanced traders’ profits for many years.70 While investors and
regulators were not aware this pricing convention existed, transparency of dealers’ prices, which
were visible on trading monitors, made enforcement of the pricing convention possible. All other
dealers could immediately observe if any dealer violated the no-odd-eighths convention.
While prices of stocks and other equities are publicly published, bond prices are less transparent,
which has put small investors at a distinct disadvantage relative to large trading institutions.
Trading of stocks is highly centralized, but except for U.S. Treasury securities, most bond trading
occurs “over the counter.”71 In the past decade transparency of bond prices has improved. In
particular, the Trade Reporting and Compliance Engine (TRACE), which was launched by the
National Association of Security Dealers (NASD) in July 2002, reports bond trades within 15
minutes, and covers a large portion of the fixed income and bond market.72 Before the
introduction of TRACE, some argued that improved transparency of prices would come at the
cost of reduced market liquidity, meaning that some large bondholders or dealers would trade less
frequently in order to protect proprietary pricing information. However, the expansion of trading
volume has improved or maintained market liquidity.73
70
Ibid.
Testimony of Vanguard Group Principal Christopher M. Ryon, in U.S. Congress, Senate Committee on Banking,
Housing, and Urban Affairs, “An Overview of the Regulation of the Bond Markets,” hearings, 108th Cong., 2nd sess.,
June 17, 2004.
72
Annette L. Nazareth, SEC Commissioner, “Remarks Before the TBMA Legal and Compliance Conference,” U.S.
Securities and Exchange Commission, New York, New York, February 7, 2006.
73
M. Goldstein, E. Hotchkiss, and E. Sirri, “Transparency and Liquidity: A Controlled Experiment on Corporate
(continued...)
71
ȱȱȱ
řŘȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
Because large traders may gain some proprietary advantage from keeping the traded prices of
bonds hidden, the advance of bond price transparency has been slow. The European Union’s
“Markets in Financial Instruments Directive,” which comes into force November 1, 2007,
requires traders to provide real-time trading data for a wide range of financial instruments and
markets.
Finance researchers contributing to the new “market microstructure” literature have taken several
approaches to analyzing the effects of price transparency. The market microstructure approach
looks at how individual traders act in financial markets.74 In addition to the well-known Christie
and Schultz work on NASDAQ, other research has found other ways to examine the effects of
price transparency. An electronic communications network named Island discontinued displaying
limit order data in three exchange-traded funds (ETFs) in which it played a dominant role from
September 2002 to the end of October 2003 rather than comply with a regulatory mandate.
During this time period ETF prices adjusted less quickly and trading costs rose. When Island
resumed displaying limit order data, trading costs fell.75 Another study compared trades before,
during, and after the regular trading day to examine the effects of price transparency. Trades made
during the regular trading day are immediately reported and available to all investors. Much less
information is available about trades that occur before or after regular hours, and because trade
volumes are much smaller, there are fewer prices to report. Barclay and Hendershott found that
prices are more volatile after hours, suggesting that pre-open and after-close markets are less
efficient than markets during regular trading hours.76
In some cases existing firms have beat back efforts to improve transparency in order to keep a
strategic advantage over would-be entrants. For instance, following Mexico’s 1994 financial
crisis the World Bank sought to create a credit registry, which would list all assets pledged as
collateral by borrowers.77 A credit registry allows any lender to see what a loan applicant has
already pledged in collateral, thus giving potential lenders a better opportunity to price risk.
Incumbent banks strongly opposed creation of the registry because they could already obtain
information about lenders, while less established lenders could not. Although a credit registry
would have expanded and strengthened Mexico’s financial system, it was thwarted by banks who
feared a more competitive environment.
ȃ¢ȱȄȱȱ£ǯȱ
The Internet seller Amazon.com’s “dynamic pricing” experiment illustrates how marketing
strategies can affect prices and create a consumer backlash. Dynamic pricing is a term that has
(...continued)
Bonds,” Babson College working paper, 2005, available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=686324.
74
For surveys of the market microstructure literature, see Bruno Biais, Larry Glosten, and Chester Spatt, “The
Microstructure of Stock Markets,” Institut d’Économie Industrielle (IDEI) working paper #253, Toulouse, France, May
28, 2004; and Ananth Madhavan, “Market Microstructure: A Survey,” Journal of Financial Markets, vol. 3, no. 3,
August 2000, pp. 205-258.
75
Terrence Hendershott and Charles M. Jones, “Island Goes Dark: Transparency, Fragmentation, and Regulation,”
Review of Financial Studies, vol. 18, no. 3, 2005, pp. 743-793.
76
Michael J. Barclay and Terrence Hendershott, “Price Discovery and Trading After Hours,” Review of Financial
Studies, vol. 16, winter 2003, pp. 1041-1073.
77
Raghuram G. Rajan and Luigi Zingales, “The Road to Prosperity: Saving Capitalism from Capitalists,” World Bank
Transition Newsletter vol. 14, July/August/September 2003, pp. 1-3.
ȱȱȱ
řřȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
come to be used to refer to a particular type of price discrimination. Price discrimination usually
takes the form of sorting customers into groups with different price sensitivity based on their
purchasing behavior.78 For instance, for airline fares a Saturday night stay-over requirement
separates price-insensitive business travelers from price-sensitive tourists. Price discrimination
typically raises prices for some groups and lowers them for others. In 2000, two episodes of
differential pricing by Amazon were publicized; the second episode involved the sale of DVDs.
Amazon, according to reports, used characteristics gathered about individual customers from the
Internet itself (such as whether a customer was new to the site, what browser the customer was
using and past purchases, etc.) to charge different prices to different individuals.
Amazon stated that it was simply conducting tests, but nevertheless apologized and promised not
to do it again.79 The same availability of information that permits individualized pricing also
makes it much easier to expose such price differentials, as occurred with the Amazon case. Once
this strategy was publicized, the protests led Amazon to cease the pricing variations and
apologize. This example illustrates that a consumer backlash against differential pricing affected
pricing behavior and provides evidence that people generally disapprove of price discrimination
based on individual characteristics.80
¢Ȭ¡ȱDZȱ ȱȱ¢ȱȱ¢ȱ
Antitrust authorities and consumer groups have often advocated price transparency on the
grounds that consumers could more easily make comparisons among sellers, which would
sharpen competition among sellers. Competition generally leads to greater efficiency and lower
prices. Price transparency, however, can change the workings of markets in unexpected ways,
which can lead to higher prices. For example, in 1993, the Danish Competition Authority required
that all ready-mixed concrete contracts be made public, which it hoped would stimulate greater
competition. Instead, average prices rose by 15%-20% and other factors such as changing
demand conditions played no discernable effect.81
There are two possible explanations for this unexpected increase in prices. First, public prices
make collusion among sellers easier. Rivals can observe sellers who undercut their competitors,
who may be able to mete out punishments in various ways. Second, price transparency may alter
the strategic incentives of sellers, inducing them to become tougher bargainers. Hviid and
Møllgaard, motivated by the Danish concrete case, developed a model in which different buyers
negotiate with a seller of an intermediate good.82 Some buyers are better informed than others,
which makes them tougher bargainers. If less-informed buyers can observe prices negotiated by
78
The Robinson-Patman Act (15 U.S.C. §13) prohibits some forms of price discrimination for the purpose of
destroying competition. Many lawyers, however, consider it difficult to win cases based on this act. The U.S. Supreme
Court (FTC v. Ruberoid Co., 343 U.S. 470, 72 S. Ct. 800, 96 L. Ed. 1081 [1952]) contended that the language of the act
was “complicated and vague in itself and even more so in its context.”
79
This event has been discussed in many articles; see, for example, David Streitfeld, “On the Web, Price Tags Blur,”
Washington Post, September 27, 2000, p. A01.
80
This may depend on how price differences are described. For instance, although senior-citizen discounts are common
and uncontroversial, a surcharge for children and working age adults, which would have the same effect, might be a
controversial pricing policy.
81
Svend Albaek, Peter Møllgaard, and Per. B. Overgaard, “Government Assisted Oligopoly Coordination? A Concrete
Case,” Journal of Law and Economics, vol. 45, December 1997, pp. 429-443.
82
Morten Hviid and H. Peter Møllgaard, “Countervailing Power and Price Transparency,” CIE discussion papers 200001, University of Copenhagen, Department of Economics, 2000.
ȱȱȱ
řŚȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
more-informed buyers, then the seller will become less willing to offer lower prices to the
informed buyers. This happens because the seller understands that by offering an informed buyer
a better price creates an obligation to offer less-informed buyers a better price. Thus, in this
model greater price transparency, in the sense that less-informed buyers are allowed to see prices
negotiated by informed buyers, can actually increase average prices. The underlying logic
resembles that of price discrimination, where different prices are charged to different groups of
consumers, with lower prices for those who are more sensitive to price. In this model buyers
differ in their bargaining power, whereas in standard price discrimination models consumers
differ in price sensitivity. Bargaining power and high price sensitivity are related because both
depend on the availability of good alternatives.
More generally, some competition experts argue that some exchanges of information about
production costs or prices among sellers often have anti-competitive effects. In particular, flows
of information among firms can make coordination or collusion easier, as the cases of the Danish
ready-mix concrete and the NASDAQ odd-eighths pricing convention suggest. On the other hand,
flows of information to buyers make price comparisons among sellers easier and thus make
consumers more sensitive to prices. Therefore, in general giving firms better price or cost
information about other firms may harm competition, but giving consumers better price or cost
information may enhance competition.83
The Hviid and Møllgaard model in some ways resembles the negotiations between hospitals and
insurers. Hospitals engage in negotiations with private insurers, which make about one-third of
hospital payments. Some insurers are in a stronger bargaining position than others due to better
data analysis, larger size, or managerial talent. The Hviid and Møllgaard model and the Danish
experience with price transparency in the concrete market suggest that it is not inevitable that
greater price transparency in hospital markets would lead to lower average prices. Most of the
evidence discussed below, however, suggests that for a variety of markets more information on
prices leads to lower overall prices.
ȱȱȱ
In determining the effects of greater price transparency, restrictions on advertising which vary
across jurisdictions or across time can provide evidence on the effects, as advertising can make
price comparisons easier. Advertising has sometimes been banned for some goods and services.
For example, many states prohibited lawyers and other professionals from advertising prices. In
general, most studies that examined prices across jurisdictions that restricted or permitted
advertising found lower prices in those jurisdictions that permitted advertising. Some studies also
examined changes over time, involving a control group, which allows a simple method of
controlling for other variables. The findings of this body of evidence may provide important
insights about how improving price information for consumers in the health sector, where priceoriented advertising is uncommon and basic information about prices is often unavailable or
difficult to obtain, might affect the level and dispersion of prices.
83
Per Baltzer Overgaard, “Market Transparency, Information Exchange and Competition,” presented at the workshop
on Competition Strategies and Competition Law hosted by the Center for International Economic Law and Department
of Economics, Swedish School of Economics and Business Administration, Helsinki, October 14, 2003, available at
http://www.econ.au.dk/vip_htm/povergaard/pbohome/webpapers/transpcomphelsinki.pdf
ȱȱȱ
řśȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
An important consideration is distinguishing between voluntary advertising and restrictions by a
third party. Also, for some types of commodities, it is possible that low prices signal inferior
quality. The clearest test of the effect of advertising occurs when a third party (usually the
government) prevents advertising. Note that many of the studies discussed below are older
because most advertising bans no longer exist, although their findings remain relevant.
ȱ¡ȱȱ¢ȱ
Several of the studies that compared effects across jurisdictions focused on optometry and the
pricing of vision exams and eyeglasses, as past rules restricting advertising varied across states.
Benham84 examined the effects of advertising on eyeglasses by comparing prices paid in states
with and without advertising restrictions in 1963. He first pointed out that the effect of advertising
is theoretically ambiguous, as it may increase demand as well as competition. Subsequently, he
separated the sample into states that permitted no advertising, that permitted advertising but not
price advertising, or that permitted any type of advertising. He found the lowest prices in states
with no restrictions, but also some benefit from advertising without price advertising. Overall,
complete advertising restrictions caused prices to be higher by 25% or more. In two subsequent
studies Feldman and Begun compared prices for vision examinations, controlling for quality
(using length of exam and office equipment).85 In the first study they found that state bans on
price advertising by either opticians or optometrists had an insignificant effect on prices, but
prices were higher by 16% when advertising was banned for both. In the second study they found
that prices were higher by 11% when state governments and state optometry boards imposed
bans. This study also indicated that the variance of prices increased with advertising restrictions.
Maurizi and Moore found that eyeglasses and contact lenses are less expensive “if the optician or
optometrist provides price information by telephone and advertises outside the telephone book.”86
Bond, Kwoka, Phelan, and Whitten87 report the results of an experiment where survey
interviewers were sent to report on both the prices and characteristics of vision exams and
eyeglasses and outcomes measured by an examination of the quality of the eyeglasses and
evaluation of prescriptions. The study found that prices were lower in cities where advertising
was restricted and chain firms did not operate; quality was about the same. Kwoka88 studied
exams by optometrists, dividing the observations into cities where advertising was not allowed,
and cities where it was, which included non-advertisers who practice in professional-looking
offices, those who do not advertise but have prominent signs in storefronts, small firms who are
affiliated with firms that do advertise, and those who advertise heavily. This study found that
advertisers offered lower prices than non-advertisers and also that non-advertisers in non84
Lee Benham, “The Effect of Advertising on the Price of Eyeglasses,” The Journal of Law and Economics, vol. 25,
October 1972, pp. 337-352.
85
Roger D. Feldman and James W. Begun, “The Effects of Advertising Restrictions: Lessons from Optometry,”
Journal of Human Resources, vol. 13, 1978, pp. 247-262; “Does Advertising of Prices Reduce the Mean and Variance
of Prices?” Economic Inquiry, vol. 18, July 1980, pp. 487-492.
86
Alex R. Maurizi and Ruth L. Moore, “The Impact of Price Advertising: The California Eyewear Market After One
Year,” Journal of Consumer Affairs, vol. 15, no. 2, 1981, pp. 290-300.
87
Ronald S. Bond, John E. Kwoka, Jr., John J. Phelan, and Ira Taylor Whitten, Effects of Restrictions on Advertising
and Commercial Practice in the Professions: The Case of Optometry, Federal Trade Commission Staff Report,
Washington, DC, U.S. Government Printing Office, September 1980.
88
John E. Kwoka, “Advertising and the Price and Quality of Optometric Services,” American Economic Review, vol.
74, March 1984, pp. 211-216.
ȱȱȱ
řŜȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ
restricted markets offered lower prices than firms in restricted markets, but the differences were
not nearly as large.
Time spent in the exam provides a proxy measure for quality. Optometrists that advertised spent
less time in exams, but non-advertisers in markets in which advertising was allowed spent more
time in exams than those in markets with advertising restrictions. These findings suggest that high
quality service is not endangered by advertising. Overall, the analysis found quality was higher
and price lower when advertising was permitted. Haas-Wilson89 explored other restrictions on
optometrists, but found media advertising reduced average prices and no effect on quality. HaasWilson and Savoca,90 who analyzed survey data collected by the Federal Trade Commission,
found that advertising restrictions on optometrists had no effect on the quality of contact lens
outcomes.
ȱȱ
Restrictions on advertising prescription drugs, according to some research, also lead to higher
prices. In 1976, Cady91 found that prescription drug prices were 4.3% higher on average in states
restricting advertising of prices than in states allowing such advertising. The restrictions
examined included limitations on outdoor signs with information identifying the products and
prices offered by the pharmacy, prohibitions on implying the pharmacy has discount drugs,
prohibitions on price advertising, and prohibitions of promotional schemes such as senior
citizens’ discount. He also found that the quantity of prescription drugs bought was unaffected.
Cady found no evidence that advertising or lower prices would increase the consumption of
drugs, as supporters of advertising restrictions had contended. (This result would not necessarily
apply to drug manufacturers’ current advertising to consumers that promotes potential benefits of
drugs, but does not advertise prices.) Kopp analyzed how the initiation of direct-to-consumer
advertising affected retail drug prices from 1986 through 1992. He found that average retail
margins of 13 drugs that were advertised fell on average by 40% after the introduction of directto-consumer advertising, while the change in average price for 120 drugs that were not advertised
did not fall.92
ȱ
The final set of cross section studies related to restrictions on posting gasoline prices. In 1972,
Maurizi93 compared prices in cities with ordinances against posting large signs advertising price
at gasoline stations and found that ordinances against the signs increased the variation in prices,
but reduced the average price. He considered the price differences unimportant because he was
unable to completely control for discounts in wholesale prices in areas subject to price wars, but
89
Deborah Haas-Wilson, “The Effect of Commercial Practice Restrictions: The Case of Optometry,” Journal of Law
and Economics, vol. 29, April 1986, pp. 165-186.
90
Deborah Haas-Wilson and Elizabeth Savoca, “Quality and Provider Choice: A Multinomial Logit-Least-Squares
Model with Selectivity,” Health Services Research, vol. 25, February 1990, pp. 791-809.
91
John F. Cady, “An Estimate of the Price Effects of Restrictions on Drug Advertising,” Economic Inquiry, vol. 44,
December 1976, pp. 493-510.
92
Steven W. Kopp, “Direct-To-Consumer Advertising and Consumer Prescription Prices,” Drug Information Journal,
vol. 30, 1996, pp. 59-65.
93
Alex R. Maurizi, “The Effect of Laws Against Price Advertising: The Price of Retail Gasoline,” Western Economic
Journal, vol. 10, October 1972, pp. 321-329.
ȱȱȱ
řŝȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
did consider the variation evidence that restrictions on signs reduce competition. A subsequent
critique by Marvel94 argued that Maurizi’s results were not valid because of a statistical issue,
except for premium gasoline. A subsequent study by Maurizi and Kelly,95 with access to a more
extensive database, indicated that posting prices reduced prices.
ȱȱ
Two studies analyzed changes in restrictions on advertising and alcoholic beverages. Luksetich
and Lofgreen96 examined the effect of an accidental repeal of liquor advertising restrictions in
Minnesota, which led to an ability to post prices and distribute price lists. The result was a decline
in price and slightly more variability in price. This latter effect was not predicted by the simple
theory; however, the authors suspect it arose from abandoning wide usage of the manufacturer’s
suggested retail price. Milyo and Waldfogel97 found that when restrictions on advertising in
Rhode Island were eliminated, advertising stores cut prices on products they advertised and on
products advertised by rivals. Non-advertising stores did not change prices, and advertising stores
did not change prices of items not advertised. Also stores with the initial lower prices were more
likely to advertise, and advertising stores drew more consumers.
¢ȱȱȱȱ ȱ
Some studies were also done on changes in information. Two studies related to food prices.
Glazer98 used the 1978 newspaper strike in New York City to examine the pattern of price
movements compared to neighboring jurisdictions on several commodities whose prices could
easily be altered, such as produce and meat. He found that prices went up in stores that normally
advertised in the newspapers, relative to stores in other jurisdictions and to stores that did not
advertise. He found the effects relatively small and that they declined over time, speculating that
individuals may have found other sources of information on prices, such as radio. Grant and
Devine99 used an experiment in two Canadian cities where, in one city, price lists for a market
basket of supermarket goods were provided via newspaper advertising and by direct mail to a
sample of consumers, while this information was not provided in the other city. The study found
that supermarket prices fell in the city with the advertising and mail data compared to the city
without it. Food prices eventually declined by 7%, and the variation also declined. Prices began to
rise when the public information program was ended.
94
Howard P. Marvel, “Gasoline Price Signs and Price Behavior: Comment,” Economic Inquiry, vol. 7, January 1979,
pp. 146-149.
95
Alex R. Maurizi and Thom Kelley, Prices and Consumer Information, American Enterprise Institute, Washington,
DC, 1978.
96
William Luksetich and Harold Lofgreen, “Price Advertising and Liquor Prices,” Industrial Organization Review, vol.
4, 1976, pp. 13-25.
97
Jeffrey Milyo and Joel Waldfogel, “The Effect of Price Advertising on Prices: Evidence in the Wake of 44
Liquormart,” American Economic Review, vol. 89, December 1999, pp. 1081-1096.
98
Amihai Glazer, “Advertising, Information and Prices—a Case Study,” Economic Inquiry, vol. 19, October 1981, pp.
661-671.
99
D. Grant Devine and Bruce W. Marion, “The Influence of Consumer Price Information on Retail Pricing and
Consumer Behavior,” American Journal of Agricultural Economics, vol. 62, May 1979, pp. 228-237.
ȱȱȱ
řŞȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱ¢ȱ ȱ
Examining the effect of information on product quality is difficult. We summarize two related
studies. Mathios100 examined the effect of mandatory nutrition labeling on salad dressings. Low
fat products advertised fat content on a voluntary basis, but high fat products (which varied in fat
content) did not. Following the mandatory labeling, sales in the highest fat products declined
significantly, suggesting that consumers used the information to make more desirable choices. Jin
and Leslie101 studied hygiene report cards for restaurants. In 1997, a Los Angeles television
program showed unsanitary conditions in some restaurants. Los Angeles County officials
responded by requiring restaurants to post hygiene quality grade cards. Incorporated cities in the
county, however, retained the power to pass their own regulations. In cities that delayed requiring
restaurants to post report cards, restaurants could display voluntarily hygiene report cards once an
inspection occurred. The authors, by analyzing variations in the timing of implementation of the
report card requirement, found evidence that the displaying cards increased hygiene scores, but
were concerned that this may have reflected “grade inflation.” However, they also found an
increase in revenues and a decrease in food-borne illnesses in the areas posting hygiene scores,
compared to other areas.
ȱȱȱȱ ȱ
In some markets consumers obtain price information with difficulty or at high cost. For example,
car buyers traditionally have had to negotiate with car dealers in person. Obtaining a price quote
from a dealer can therefore require several hours of effort, from identifying local dealers,
traveling to the dealer’s lot, and negotiating with salesmen and finance specialists. When
obtaining price information is costly, a consumer may settle for a given firm’s price, even though
further search might have identified a firm with lower prices. The economic theory of search
describes a consumer’s optimal strategy when obtaining price quotes is costly. A consumer gets a
price quote, then either decides to search further or to settle for one of the price quotes he has in
hand. An optimal search rule balances the cost of obtaining an additional price quote against the
expected gains of further search.102 If a consumer has previous experience in the market, and
knows something about the distribution of prices, then computing an optimal stopping rule for a
search is straightforward. If the distribution of prices is unknown, then no known optimal
stopping rule exists.
In search theory models, firms cannot price discriminate, but consumers still pay different prices.
On average, consumers who search more pay lower prices. Because consumers have different
costs of search, different firms will offer different prices. Firms with higher prices earn higher
markups on a smaller number of sales, while firms with lower prices have smaller markups but a
higher number of sales. If search costs for consumers fall, then both average prices and price
dispersion fall.
100
Alan D. Mathios, “The Impact of Mandatory Disclosure Laws on Product Choices: An Analysis of the Salad
Dressing Market,” Journal of Law and Economics, vol. 48, October 2000, pp. 651-676.
101
Ginger Zhe Jin and Philip Leslie, “The Effect of Information on Product Quality: Evidence From Restaurant
Hygiene Cards,” Quarterly Journal of Economics, vol. 118, no. 2, 2003, pp. 409-451.
102
Search theory was initiated by George Stigler’s article, “The Economics of Information,” Journal of Political
Economy, vol. 69, June 1961, pp. 213-25.
ȱȱȱ
řşȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
ȱȱȱ ȱ
Many economists expected that the Internet, which enabled the emergence of cheap and efficient
price searching mechanisms, would lead to lower prices. Some studies, conducted when the
Internet use had just started to spread to the general public, found higher prices online, although
later studies tended to show some price reductions. Pricing and marketing techniques have
changed as the Internet has evolved, often in different ways for different markets. In addition,
studies of Internet pricing have become more sophisticated over time. In general, later studies,
and studies of comparison sites, tend to find lower prices as a result of the Internet.
ȱ
Lee103 studied an electronic automobile auction network in Japan and found that prices can be
higher than in more traditional markets, even after controlling for quality. This effect might be
attributable to the reduction in transaction costs and the better matching of desired car type. In
two papers, Settlemeyer, Morton, and Silva-Risso104 examined the effect of the Internet on car
prices and found that prices were lower for direct Internet buying. Buyers referred to offline
dealerships also paid lower prices, apparently because additional information increases
bargaining power and because of the referral service.
ȱȱȱ
Bailey,105 in one early Internet pricing study, found that prices for books, CDs, and software in
1996 and 1997 were higher online than in conventional outlets. Brynjolfsson and Smith,106
however, studying a later period and using a more sophisticated methodology, found prices for
books and CDs on the Internet were 9%—16% lower than prices in conventional outlets.
Although posted Internet prices showed considerable dispersion, so that an unweighted measure
of price variation for Internet sellers exceeded that for conventional sellers, prices weighted by
market share varied less than conventional sellers’ prices. This effect occurred because sales at a
few Internet booksellers, whose prices were relatively close to one another, comprised a large
proportion of book sales. Clay, Krishman, Wolff, and Fernandes,107 in 2001, found that book
prices were no lower on the Web than at physical booksellers. They also found evidence of
product differentiation, given the higher prices charged by Amazon, compared to both Barnes &
Noble online and Borders online. Goolsbee and Chevalier108 found significant price variability for
103
H.G. Lee, “Do Electronic Marketplaces Lower the Price of Goods?” Communications of the ACM, vol. 41, January
1998, pp. 73-80. (Based on a review on the ACM website by S. Srinivasan at http://portal.acm.org/
citation.cfm?id=268122&coll=portal&dl=ACM.)
104
Florence Settlemeyer, Fiona Scott Morton, and Jorge Silva-Risso, “Cowboys or Cowards: Why are Internet Car
Prices Lower?” mimeo, November 2005 (also appeared as National Bureau of Economic Research Working Paper
8667, December 2001); and “How the Internet Lowers Prices: Evidence from Matched Survey and Auto Transaction
Data,” National Bureau of Economic Research Working Paper 11515, August 2005.
105
Joseph Bailey, Electronic Commerce: Prices and Consumer Issues for Three Products: Books, Compact Discs, and
Software (Paris: OECD, 1998).
106
Erik Brynjolfsson and Michael D. Smith, “Frictionless Commerce? A Comparison of Internet and Conventional
Retailers,” Management Science, vol. 46, April 2000, pp. 563-585.
107
Karen Clay, Ramayya Krishman, Eric Wolff, and Danny Fernandes, “Retail Strategies on the Web: Price and Non
Price Competition in the Online Book Industry,” Journal of Industrial Economics, vol. 49, December 2001, pp. 521540.
108
Austan Goolsbee and Judith Chevalier, “Measuring Prices and Price Competition Online: Amazon and Barnes &
(continued...)
ȱȱȱ
ŚŖȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
books on the Internet. Waldfogel and Chen109 found that those who used price comparison sites
reduce their shopping at branded retailers, such as Amazon, by a tenth if performing price
comparison, and by a fifth if comparing both price and quality. Price comparison site users
reduced purchases from Amazon and from offline chains.
ȱȱ
Verlinda and Lane110 found an increase in unrestricted airline fares relative to restricted fares as
Internet price searches increased, but this difference was statistically insignificant. Over the time
period of this study the share of restricted tickets decreased substantially. This trend, on which the
authors did not focus, could be interpreted as an increase in quality, as it allows more travelers
flexibility in their travel plans. In addition, airline fares had long been subject to comparison
through travel agents, which means the increase in information may not have been as great for
this product as for other products. Clemons, Hann, and Hitt found similar tickets on different sites
in 1997 varied on average by 18%.111 By 2002, Chen found these differences had narrowed to
0.3%-2.2% for fares available at multiple travel websites.112 This convergence appears to stem
from several major changes in the air travel market. First, the launch of several online ticket
agencies and airlines’ efforts to promote their own direct ticketing websites have substantially
changed the online travel market. Between 1997 and 2002, use of online travel agencies increased
elevenfold. Second, more consumers buy air tickets on the Internet. According to one recent
estimate, 60% of travelers buy tickets on-line.113
ȱ ȱ
Brown and Goolsbee114 found that the appearance of Internet sites which allowed for comparisons
among term life insurance policies led to significant decreases in prices. This study found that an
increase in the share of individuals using the Internet comparisons of 10% led to a 5% decrease in
price.
¢ȱȱ ȱȱ
The evidence from Internet studies is mixed, and it is, of course, possible that Internet purchasers
are willing to pay higher prices to purchase on the Internet because of the reduction in
transactions cost or other advantages.
(...continued)
Noble,” National Bureau of Economic Research working paper 9085, July 2002.
109
Joel Waldfogel and Lu Chen, “Does Information Undermine Brand? Information Intermediary Use and Preference
for Branded Web Retailers,” National Bureau of Economic Research working paper 9942, September 2003.
110
Jeremy A. Verlinda and Leonard Lane, “The Effect of the Internet on Pricing in the Airline Industry,” working
paper, November 2004, available at http://www.ags.uci.edu/~verlinda/papers/verlinda-lane-final.pdf.
111
E.K. Clemons, I. Hann, and L.M. Hitt, “Price Dispersion and Differentiation in Online Travel: An Empirical
Investigation,” Management Science, vol. 48, no. 4, 2001, pp. 521-39.
112
Jihui Chen, “Difference in Average Prices on the Internet: Evidence from the Online Market for Air Travel,”
Economic Inquiry, vol. 44, no. 4, October 2006, pp. 656-670.
113
Ibid., p.656.
114
Jeffrey R. Brown and Austan Goolsbee, “Does the Internet Make Markets More Competitive? Evidence from the
Life Insurance Industry,” Journal of Political Economy, vol. 110, June 2002, pp. 481-507.
ȱȱȱ
Śŗȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
The characteristics of the Internet as it evolves have complex effects on marketing and pricing
strategies. The Internet is well suited to increasingly sophisticated price comparison tools, which
tend to reduce prices and price dispersion for those who use them. The evidence on the use of
comparison sites (as opposed to direct sales on the Internet), which may be most relevant to the
question at hand, seems to suggest that having access to direct price comparisons reduces prices
when consumers use price comparison sites. Baye and Morgan contend that prices reached via
price comparison sites are lower than prices obtained directly from a vendor’s website.115
On the other hand, some Internet characteristics make entry of new firms difficult. Internet traffic
patterns show strong winner-take-all features: a small number of websites account for a large
proportion of total traffic. Designing, building, and maintaining a major retail website are
expensive tasks. Some Internet sellers have been able to establish strong brand identification that
permits higher prices. Because of these characteristics, in some product markets a few dominant
firms may be able to maintain substantial market power in the Internet Age. Highly visible firms,
however, can be vulnerable to public pressure, as the case of Amazon’s dynamic pricing
experiment illustrates.
ȱȱȱȱ¢DZȱȱ
Most research suggests that when better price information is available prices for goods sold to
consumers fall. The largest and most straightforward body of evidence relates to the effect of
advertising, where nearly all research indicates advertising prices is associated with lower prices.
This reduction in prices suggests that advertising’s increased information on prices and increases
in competition outweigh any tendency to increase prices through increasing demand and brand
identification. Evidence on price comparison sites on the Internet also seems to support this view.
(Again, this conclusion may not apply to current manufacturers’ drug advertising that does not
include price information.)
Evidence for markets in intermediate goods is more complicated. When middlemen are involved
the effects of price transparency depend on the particulars of market structure. Price transparency
gives buyers and sellers important information about the true economic value of goods, services,
or assets, but may also enable traders to observe deviations from collusive practices. Allowing
weaker bargainers to see prices negotiated by stronger bargainers will change incentives facing
buyers and sellers, and can lead to price increases. In financial markets dealers need to trade with
other dealers on a frequent basis to rebalance portfolios and take actions to maintain liquidity,
which leads to a complex relationship among price transparency, trading costs, and market
efficiency. Studies in experimental financial economics suggest that price transparency can either
increase or decrease prices.116 In short, how price transparency affects intermediate goods markets
is an active area of research, and settled conclusions have not yet been reached.
In traditional economic theory consumers react to price differences because lower prices mean
that consumers can buy more goods and services. The unwillingness of consumers to pay higher
115
Michael R. Baye and John Morgan, “Information Gatekeepers and Price Discrimination on the Internet,” Economic
Letters, vol. 76, no. 1 (June), pp. 47-51.
116
For instance, Glosten notes that one experimental study found that “transparency involves lower spreads and more
‘efficient’ prices,” while another study found that “transparency involves lower spreads and less ‘efficient’ prices
(emphasis added).” See Lawrence R. Gloten, “Introductory Comments: Bloomfield and O’Hara, and Flood, Juisman,
Koedijk, and Mahieu,” Review of Financial Studies, vol. 12, no. 1, 1999, pp. 1-3.
ȱȱȱ
ŚŘȱ
ȱ
ȱȱ¢ȱ ȱȱ¢ǵȱ
prices imposes market discipline upon firms. Other mechanisms, however, may act to discipline
firms as well. Firms that charge unusually high prices may face political or legal pressure. For
example, sellers of gasoline may face complaints of price gouging with sharp price increases, as
happened in some states following Katrina.117 Also consumers are willing to punish firms by not
doing business with them, even if this action reduces the welfare of consumers. For example,
many consumers reported that they had resolved not to buy from Amazon.com after experiencing
“dynamic pricing,” even if this meant that they would pass up advantageous offers in the
future.118 NASDAQ spreads narrowed not because of consumer pressure, but because NASDAQ
administrators feared adverse media coverage and lawsuits filed by investors and regulators.
Thus, price transparency may impose discipline upon firms, even if this occurs through nonmarket mechanisms.
ȱȱ ȱ
(name redacted)
Analyst in Economic Policy
[redacted]@crs.loc.gov, 7-....
(name redacted)
Senior Specialist in Economic Policy
[redacted]@crs.loc.gov
, 7-....
117
James McPherson, “Retail Gas Prices Jump, Deliveries Falter as Katrina’s Energy Effects Spread,” September 1,
2005, available at http://www.signonsandiego.com/news/nation/20050901-0605-katrina-gasprices.html.
118
David Streitfeld, “On the Web, Price Tags Blur,” Washington Post, September 27, 2000, p. A1.
ȱȱȱ
Śřȱ
EveryCRSReport.com
The Congressional Research Service (CRS) is a federal legislative branch agency, housed inside the
Library of Congress, charged with providing the United States Congress non-partisan advice on
issues that may come before Congress.
EveryCRSReport.com republishes CRS reports that are available to all Congressional staff. The
reports are not classified, and Members of Congress routinely make individual reports available to
the public.
Prior to our republication, we redacted names, phone numbers and email addresses of analysts
who produced the reports. We also added this page to the report. We have not intentionally made
any other changes to any report published on EveryCRSReport.com.
CRS reports, as a work of the United States government, are not subject to copyright protection in
the United States. Any CRS report may be reproduced and distributed in its entirety without
permission from CRS. However, as a CRS report may include copyrighted images or material from a
third party, you may need to obtain permission of the copyright holder if you wish to copy or
otherwise use copyrighted material.
Information in a CRS report should not be relied upon for purposes other than public
understanding of information that has been provided by CRS to members of Congress in
connection with CRS' institutional role.
EveryCRSReport.com is not a government website and is not affiliated with CRS. We do not claim
copyright on any CRS report we have republished.
This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.