• No results found

349_lect9.pdf

N/A
N/A
Protected

Academic year: 2020

Share "349_lect9.pdf"

Copied!
14
0
0

Loading.... (view fulltext now)

Full text

(1)

Lecture Notes 9: Structure-Conduct-Performance

Having studied the main market structures in detail, we are now in a position to study one of the most important pieces of industrial organization from a policymaking perspective. In a general sense, policymakers and regulators are interested in how the structure of a market impacts the way that buyers and sellers behave and, in turn, the outcomes that the market produces for society. This seems like a straightforward statistics exercise, but there are complications both with measurement and with the way in which we interpret results.

Structure-Conduct-Performance Paradigm

The basic idea is that the features of an industry impact decisionmaking within firms. For example, we have seen that market concentration and substitutability across products impacts market prices. In turn, decisions made by firms impact social welfare. For example, price markups lead to deadweight loss.

The structure-conduct performance paradigm relates the following elements.

Structure – Factors like market concentration, technology or consumer demand patterns that characterize an industry.

Conduct – The way that firms behave. Examples are pricing and output decisions, advertising, or engagement in Research & Development spending.

Performance – The outcomes that a market produces for society. We can think about things like consumer surplus and profits, which measure a market’s success at producing benefits.

(2)

Some Simple Theory

Before we dive into measurement in detail, let’s just review some simple theory on market structures and outcomes. The table below summarizes pricing and profit outcomes that characterize the equilibrium across our four market structures.

Market Pricing SR Profit LR Profit

Perfect Competition P = MC +, −, or 0 0

Monopolistic Competition P > MC +, −, or 0 0

Oligopoly P > MC +, −, or 0 + possible

Monopoly P > MC +, −, or 0 + possible

There are a couple of important points.

First, the existence of short-run profits or losses says nothing. Any firm in any kind of market can have short-run profits or losses.

Second, long-run profits come from barriers to entry, not from market power. Monopolistic competitors differentiate their products and have market power over their customers (i.e. the demand curve they face is not totally elastic), but their profit is nevertheless driven to zero in the long-run by easy entry.

Third, zero profit and allocative efficiency (P=MC) are not the same thing. Monopolistic competitors are driven to zero profit in the long-run, but the zero-profit equilibrium involves price marked up over marginal cost of production. Recall the diagrams below, showing zero-profit equilibrium in perfect competition and in monopolistic competition, respectively. As another way to put it, if we want to check for allocative efficiency we need price/cost data, not data on overall profits.

(3)

Measures of Structure

If we are going to relate the structure of a market to the conduct of firms and ultimately to the performance of the market, we need ways to measure important aspects of structure. We will begin with various measures of market concentration, which are the most commonly used measures of structure.

Market Concentration

Do industries consist of a few large firms or a lot of small firms? In a nutshell, this is the level of concentration in an industry.

The market share of a firm is the percentage of total sales in an industry that are accounted for by a particular firm. For example, if industry boat sales are $50 billion and a firm has $10 billion of boat sales, then the firm’s market concentration is 0.2 = 20%.

The four-firm concentration ratio (C4) is the sum of the market shares of the four largest firms in an industry. Industries that feature a C4 close to 0 are less concentrated. An example is farming. Even the four largest potato farmers in the US only account for a small fraction of total potato sales. Industries that have a C4 close to 1 are more concentrated. An example is aircraft manufacturing, where the C4 is 81%. The four largest aircraft manufacturers account for more than 80% of total aircraft sales.

Not everybody agrees that 4 is the right number. The government also reports C8 and C50 for some industries. It may also be the case that increasing the market share of the largest firm matters more than increasing the market share of the third or fourth-largest firms.

The Herfindahl-Hirschman Index (HHI) is the sum of the squared market shares of all firms in an industry times 10,000. An industry with an HHI closer to 10,000 is a concentrated industry with a few large firms with big market shares. An industry with an HHI closer to 0 features a large number of firms with smaller market shares.

For example, consider a market that has 5 firms, each with a 20% market share.

𝐻𝐻𝐻𝐻𝐻𝐻 = 10000 ⋅ [0.22+ 0.22+ 0.22+ 0.22+ 0.22] = 2000

By contrast, suppose that a market has one firm with an 80% market share, and the other four firms with a 5% market share.

(4)

One advantage of the HHI over concentration ratios is that it incorporates data on all the firms in an industry, rather than only giving information about the top 4 or 8 firms. In truth, most studies obtain similar results no matter which measure of market concentration is used. Both the HHI and concentration ratios are a good indicator of how concentrated an industry is. Nevertheless, there are a few cautions about interpreting concentration measures.

1. Most concentration measures typically include only domestic production. An industry might look concentrated domestically, but actually be quite competitive if it faces robust competition with imports.

2. Most concentration indexes measure overall nationwide concentration. The problem with this is that, even if an industry features many firms with modest overall market shares, firms may have local markets cornered. In other words, even a firm with a small national market share can exercise a high degree of market power over its customers if it doesn’t face much competition within its geographical region. Health care is a good example.

3. Be careful of how the market is defined. For example, the beverage industry is relatively unconcenrated, but that includes all nonalcoholic beverages. Are sodas really the same market as sports drinks? The market for sports drinks is actually quite concentrated. On the other hand, you don’t want to get too specific. Measuring concentration in the market for glass bottles is pretty uninformative because of competition with plastic bottles.

The US Census Bureau regularly conducts an Economic Census, and you can go online and find a wealth of data on concentration measures across industries. Here are some industries. There are no hard rules, but regulators generally consider industries with an HHI greater than about 1800 to be concentrated.

Industry C4 HHI

Cigarettes 0.95 ??? Computers 0.81 2662

Cars 0.81 2324

Snack foods 0.64 2717

Soap 0.63 2808

(5)

A bit of history – Overall, concentration ratios have been trending downward overall for the last few decades. The distribution (i.e. which industries are more concentrated and which are less concentrated) hasn’t changed much since the government started measuring.

Firm Size

A simple measure of an industry’s structure is the overall size of a typical firm or of the largest firm in an industry. This can depend on both the size of the industry and on concentration. There is high variation across industries. For example, the largest merchandising firm is Wal-Mart, with annual sales over $480 billion. By contrast, the largest auto manufacturer is GM, with a little over $200 billion in sales. And the largest manufacturer of building materials is Owens-Corning, with $5.3 billion in sales. Some industries are quite dynamic. The largest telecommunications firm keeps changing.

Demand and Market Conditions

Another key question for the structure of an industry is what kind of demand the firm faces. There are a few key questions.

• How large is demand for the product generally? That could impact the number of firms that the market can support, especially if there are large fixed costs.

• How much information is available to consumers? Industries where consumers can easily get pricing information without much effort tend to be super-competitive, even if concentrated. Airlines are a good example, since the advent of online search websites.

The most important metric regarding consumer demand demand is price elasticity. A key element of a market’s structure, and of the way that firms behave, is the elasticity of demand that a typical firm faces. How much market power does a firm have over its customers?

One common standardized measure of the elasticity that firms face is the Rothschild Index (R). The Rothschild index is the ratio of the market elasticity to the individual firm elasticity.

𝑅𝑅 =𝜀𝜀𝜀𝜀𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑓𝑓𝑓𝑓𝑚𝑚𝑚𝑚

(6)

quite elastic compared to the market demand. But when consumers are not inclined to substitute between different firms in the market, the elasticity faced by a particular firm would be close to the overall market elasticity. In other words, the decline in demand when the firm raises its price is the overall industry-wide decline in demand for the product – no substitution to new firms.

Here is some data on the Rothschild Index across various industries.

Industry 𝜺𝜺𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎 𝜺𝜺𝒇𝒇𝒇𝒇𝒎𝒎𝒎𝒎 R

Tobacco -1.3 -1.3 1 Paper -1.5 -1.7 0.88 Publishing -1.8 -3.2 0.56

Cars -1.0 -1.9 0.53

Clothes -1.1 -4.1 0.27 Construction -1.0 -5.2 0.19 Financial Services -0.1 -5.5 0.02 Agriculture -1.8 -96.2 0.02

When 𝑅𝑅 = 1, the responses of consumers to each firm’s price changes are identical to the overall response in the market. There is no substitution between firms. When 𝑅𝑅 is close to 0, each firm faces a large reduction in demand when it raises its price relative to the overall reduction in market demand. This is because customers switch to new firms. Overall, the Rothschild index is a good way to measure how substitutable consumers consider the products offered by different firms.

Barriers to Entry

Another key aspect of market structure is potential for entry. There is no good single number that captures the strength of entry barriers, but some possibilities include upfront capital startup costs, how heavily the industry is covered by patent protection, economies of scale in the production technology, or advertising expenditures (advertising saturation can make it difficult for new startups to compete).

Technology

(7)

Performance

The general motivation for measuring performance is to determine how close a market is to the competitive ideal. The two main measures of market performance are the rate of return and the Lerner Index (the price/cost markup). We will treat each in turn.

Rate of Return

The big problem for economists in measuring rate of return is that accounting profit (typically reported on financial statements) is not the same as economic profit. In measuring whether a firm is earning an excess return, we need to know the economic profit, not merely the accounting profit.

Remember the difference. Accounting profit reflects total revenue net of explicit costs only. Economic profit also takes into account the opportunity cost of the firm’s resources – which represent a sacrifice, but without a direct outlay. Practically, the biggest discrepancy between the two is the opportunity cost of capital. That is, a firm incurs an opportunity cost by holding capital rather than using it for alternative purposes. The best way to measure this implicit cost is the rental cost – what the owner(s) would have received if the firm’s capital were rented out instead of being tied up at the firm.

If there is a well-developed rental market, the implicit cost of holding the capital is easy to compute. But there may not be. In this case, we have to impute what the rental value of capital would be. If we do it this way, it is critical to look at the current value of the capital (its replacement cost) versus the historical value of the capital (the cost originally paid). For example, a machine might have been purchased for a low cost but is currently worth a lot more. Even though it generated a huge profit for the firm, the rate of return on the machine might be quite low if the firm could rent it out for a high rent versus keeping it in its current use.

In addition to the opportunity cost of the foregone rental value, another cost of holding capital is its depreciation – the wear and tear that it incurs.

A firm holds 𝐾𝐾 units of capital and the price of each unit of capital is 𝑃𝑃𝐾𝐾, so the total value of the capital it holds is 𝑃𝑃𝐾𝐾𝐾𝐾. All in all, if the rental rate is 𝑟𝑟 and the depreciation rate is 𝛿𝛿, then the cost to hold a firm’s capital over the year is:

Capital Cost = 𝑇𝑇𝑇𝑇𝑚𝑚 = (𝑟𝑟 + 𝛿𝛿) ⋅ (𝑃𝑃𝐾𝐾𝐾𝐾)

(8)

The firm’s economic profit is its revenues net of all costs, including capital costs.

Π = 𝑇𝑇𝑅𝑅 − 𝑇𝑇𝑇𝑇𝑙𝑙− 𝑇𝑇𝑇𝑇𝑚𝑚− 𝑇𝑇𝑇𝑇𝑚𝑚

= 𝑇𝑇𝑅𝑅 − 𝑇𝑇𝑇𝑇𝑙𝑙− 𝑇𝑇𝑇𝑇𝑚𝑚− (𝑟𝑟 + 𝛿𝛿)(𝑃𝑃𝐾𝐾𝐾𝐾)

A normal rate of return on capital sets economic profits equal to 0.

Π = 0 𝑇𝑇𝑅𝑅 − 𝑇𝑇𝑇𝑇𝑙𝑙− 𝑇𝑇𝑇𝑇𝑚𝑚− (𝑟𝑟 + 𝛿𝛿)(𝑃𝑃𝐾𝐾𝐾𝐾) = 0

𝑟𝑟𝑃𝑃𝐾𝐾𝐾𝐾 = 𝑇𝑇𝑅𝑅 − 𝑇𝑇𝑇𝑇𝑙𝑙− 𝑇𝑇𝑇𝑇𝑚𝑚− 𝛿𝛿𝑃𝑃𝐾𝐾𝐾𝐾

𝑟𝑟 =𝑇𝑇𝑅𝑅 − 𝑇𝑇𝑇𝑇𝑙𝑙− 𝑇𝑇𝑇𝑇𝑃𝑃 𝑚𝑚− 𝛿𝛿𝑃𝑃𝐾𝐾𝐾𝐾

𝐾𝐾𝐾𝐾

The firm’s earned rate of return on its capital is therefore given by its net income (accounting profits) divided by the value of its assets.

This is a benchmark for measuring excess returns. Importantly, the rate of return depends on the ratio of the excess profits to total capital value. For example, a small profit (in dollars) can lead to large differences in the rate of return in industries with relatively little capital.

Trying to measure rates of return in practice is fraught with difficulties, especially in the face of limited data provided by accounting statements. Here are a few of the main ones.

1. Accountants tend to value capital at book value (historical value). We emphasized above the need to value capital at its current value (replacement cost).

2. Accountants tend to measure depreciation as a constant, linear decline over the useful life of an asset. But some assets might depreciate very little at first (and hence have high replacement value) or conversely might depreciate a lot right away. Thus, the accountant’s linear decline is not always a good metric of the current value of the capital.

3. Advertising and R&D expenditures have long-lasting effects on revenues even if they are paid for immediately. The consequence is that the firm’s rate of return will look too low in the year in which the expenditure was undertaken and will look too high in subsequent years.

(9)

5. You have to adjust for inflation. Returns are worth less at the end of the year than at the beginning of the year.

6. You need to use after-tax rates of return, not before-tax rates of return.

7. You need to adjust rates of return for risk. A high return might not be as attractive as it seems on paper if it involves a lot of risk.

8. Suppose a firm with monopoly power sells its assets to a new owner. The sale price demanded by the current owner is going to reflect the capitalized value of future monopoly profits. Thus, the new owner’s actual return will look misleadingly low because the price he paid for the monopoly’s assets was not the true value of the capital, but inflated to reflect the fact that the original owner captured the future excess returns when he sold off the firm’s assets to the new owner.

9. Stock returns are definitely not a good measure of earned rate of economic return, mainly because of differences in debt/equity ratio across firms. Firms with a high proportion of debt on the books are riskier for stockholders and will have to pay higher stock returns, even if the actual rate of return earned by the firm is comparable to other firms.

Once we can agree on a firm’s rate of return (not an easy task), we have to compare it against some benchmark.

Consider a firm with 100 units of capital, each worth $10. It has revenues of $110, labor and material costs of $10 and faces a depreciation rate of 2%. Using our formula, its earned rate of return is:

𝑟𝑟 =$110 − $10 − 0.02 ⋅ ($10 ⋅ 100)$10 ⋅ $100 = 0.08

If the competitive return in the economy is 5%, then the firm is earning an excess rate of return. Equivalently, if we use the competitive rate of return to value the firm’s capital, then its economic profit is:

Π = $110 − $10 − (0.02 + 0.05) ⋅ ($10 ⋅ 100) = $30

In other words, a positive economic profit is equivalent to earning an excess rate of return on its capital.

(10)

we think of as more competitive. For example, the rate of return for cigarette sales is 14% and for crude petroleum has averaged around 12%. Historically, the rate of return in telecommunications (which used to be highly regulated) has been around 15%, but that is dropping. By contrast, the rate of return in textile manufacturing is around 9% and in agriculture is around 7%.

It is interesting to note that even the lowest rates of return from owning a business are much higher than the rate of return for a risk-free financial asset. For example, T-bills have historically averaged around 3% rate of return. This discrepancy is presumably a risk premium associated with having capital in any business, versus in a risk-free T-bill.

Lerner Index

Besides rate of return, a second measure of market performance and outcomes is the Lerner index. Recall that the Lerner index is the price-cost margin. It is the percentage of the price that represents a markup over marginal cost.

𝐿𝐿 = 𝑃𝑃 − 𝑀𝑀𝑇𝑇𝑃𝑃

The biggest measurement problem with the Lerner index is that marginal cost is very hard to measure. Firms’ accounting statements typically give average costs, not marginal costs. The difference between the two is significant because fixed capital costs are included in the average but not at the margin.

A second, related problem is that the Lerner index can be a misleading indicator of rate of return for firms with extremely high fixed costs. For example, a pharmaceutical firm might be able to produce a pill for a marginal cost of a few cents, but the high price is covering the fixed R&D costs. While it’s true that there is a high deadweight loss because of pricing over marginal cost, the high Lerner index in this case does not imply a huge rate of return.

Relationship between Structure and Performance

Old Models

(11)

The usual place to start, and the “old” way of doing things in IO was to think about a regression model of the following form:

Performance𝑓𝑓 = 𝛽𝛽0+ 𝛽𝛽1⋅ Structure𝑓𝑓 + 𝑢𝑢𝑓𝑓

In other words, we look across a whole bunch of industries, and we test the relationship between the structure of those industries and the performance of those industries. (You could include other control variables in the regression if you wanted to). This is a simple regression exercise. The structure is the independent variable, and performance is the dependent variable, or response variable.

There are a whole mountain of studies of this form. Here are some of their results.

• Rate of return is higher in more concentrated industries. • Profits are higher in industries with high barriers to entry.

• Profits are higher in industries with high R&D and advertising expenditures.

• Profitable industries in the short-run tend to suffer declines in profit as the markets reach long-run equilibrium, but the adjustment is slower in more concentrated industries.

• The connection between concentration and the Lerner index is weak.

• The price-cost margin is higher in industries with more advertising. It’s lower in industries where there is concentration among buyers.

There are also a whole host of industry-specific studies. For two examples, it appears to be true for airlines that routes with higher concentration have higher fares (fares are about 6% higher when C4 doubles), although it seems to have more to do with whether there are one or two dominant firms than on overall concentration. Rail transport features high entry costs, but with just one competitor, rates are on average 18% lower. Adding more competitors has only a small additional impact.

New Models

The “old” approach is basically descriptive. We look at different industries or markets and the regression approach tells us how the structure of those markets is related to performance. Newer economists tend to view these results with some skepticism. Why?

(12)

Here’s a simple example. If firms are earning low profits, then there’s no incentive for new entry. But this relationship is the reverse: Performance  Structure. Another example. An innovative firm is able to earn high profit and expand. The market then becomes more concentrated. Again, we have Performance  Structure.

In other words, it’s certainly true that structure can impact performance. Concentrated industries lead to high markups, high returns and deadweight loss. But it’s also true that market performance itself can impact market structure. Thus, when we just estimate the correlation between market structure and market performance, we don’t know if we are getting the answer that we want (how structure impacts performance) or whether we have a chicken-and-egg problem.

This kind of problem goes beyond studying the impact of concentration measures. For example, it is definitely true that industries with inelastic demand engage in more advertising. But are they engaging in advertising to take advantage of market power with their customers? Or is the inelastic demand a result of heavy advertising? Correlation is not causation. Market structure can affect how firms behave, but how firms behave can itself impact market structure.

The only real solution is to find an exogenous structure change. In other words, we want to study a change in structure that was created by an outside circumstance, not one that arose from within the market itself. For example, maybe a sudden change in government regulations shakes up industry structure. This lets us legitimately study the impact of structure on performance because the change in structure was exogenous – it doesn’t come from within the market, so we avoid a chicken-and-egg problem.

(13)

Exercises

Problem 1

An industry consists of only three firms. Their sales are $200,000, $500,000 and $400,000 respectively. Calculate the HHI and the four-firm concentration ratio.

Problem 2

The market price elasticity for retail gasoline is −0.9. If the Rothschild index is 0.6, what does this imply about the elasticity faced by a typical retailer?

Problem 3

The table below gives information on four industries – The C4 and HHI for the industry, along with the Lerner Index and Rothschild Index of a typical firm. Indicate whether each industry is best characterized by perfect competition, monopolistic competition, oligopoly or monopoly.

Industry C4 HHI Lerner

Index

Rothschild Index

Industry A 0.005 75 0.45 0.34 Industry B 0.0001 55 0.0034 0.0023 Industry C 1 10,000 0.4 1 Industry D 1 5573 0.43 0.76

Problem 4

A firm competes against many other firms in a highly competitive industry. Over the last decade, several firms have entered the industry and as a consequence the firm is no longer earning an excess rate of return. While the four-firm concentration ratio and the Herfindahl-Hirschman index are low, the Rothschild index is close to 1. What market structure characterizes the industry in which this firm is competing in?

Problem 5

(14)

Problem 6

Several years ago, Pfizer and Warner-Lambert agreed to a $90 billion merger, thus creating one of the world’s largest pharmaceutical companies. Pharmaceutical companies tend to spend a greater percentage of sales on R&D activities than other industries. The government encourages these R&D activities by granting companies patents for drugs approved by the Food and Drug Administration. For instance, Pfizer-Warner-Lambert spent large sums of money developing its popular cholesterol-lowering drug, Lipitor, which is currently protected under a patent. Lipitor sells for about $3 per pill. Calculate the Lerner index if the marginal cost of producing Lipitor is $0.30 per pill. Does the Lerner index make sense in this situation? Explain.

Problem 7

References

Related documents

To assess the quality of change point detection algorithm, we ran ADWIN on the original log with following input parameters: feature = J-measure for every activity pair, hypothesis

○ If BP elevated, think primary aldosteronism, Cushing’s, renal artery stenosis, ○ If BP normal, think hypomagnesemia, severe hypoK, Bartter’s, NaHCO3,

percentage of germination, and germination rate, germination index, energy of germination and seedling.. Table.1: Averages of percentage of germination, and

When employees forget their password or forget to login/logout properly, helpdesk calls will increase to request password resets and front desk has to cover non-answered phone

As inter-speaker variability among these the two groups was minimal, ranging from 0% to 2% of lack of concord in the 21-40 group and from 41% to 46% in the 71+ generation, we

First, using ZINC and ChEMBL, we predict molecular target activities for 171 million commercially available compounds at 2629 targets and store them in an accessible database..

Each approach can locate a person capable of completing necessary business tasks. But, a new outsourcing methods has emerged. It allows a business to better leverage the labor

For the Cox-type regression model with bent-line change- points in the underlying regression function we proposed an estimation procedure and we proved that the regression