George S. Ford Chief Economist - PowerPoint Presentation

Slide1

George S. FordChief Economist

Economic Tools for Broadband Policy

PHOENIX

CENTER

1

Slide2

The 800 Pound Gorilla …

2

REGULATION

V.

Limited Competition

Slide3

The 800 Pound Gorilla …

3

complainers

V.Idealogues

Markets Never Work

Markets Never Fail

Slide4

The 800 Pound Gorilla …

4

Those of us who are amazed at how ridiculous the arguments have become.

Slide5

Realistic Expectations

5

We need to form realistic expectations about …

Market Structure (how many firms)

Nature of Competition

Effect of Competition on Investment and Innovation

The Relationship between Communications Infrastructure and Economic Activity

The Efficacy of Regulation

What Markets Do

What the Data Tells Us

Slide6

Market Structure

6

N* is FewN ≈ 2 is not an unrealistic expectation for terrestrial, wireline networksIt appears that N ≈ 3 to 5 is right for wireless communications “Nearby” competitors (wifi, wimax) offer additional limits to market power

Slide7

N*: Equilibrium Number of Firms

7

Firms enter until the next entrant loses moneyWhen entry stops, and exit does not occur, we have equilibriumN* depends on the size of the market, the fixed/sunk cost of entry, and the intensity of competition

Slide8

Numerical Example:Policy Paper No. 21

8

Number of Firms (N)Firm Net RevenuesEntry CostsNet Profits110015852401525N * = 32015541215-35815-76515-107415-11

The Equilibrium Number of Firms is 3. If the 4

th

firm enters, all firms lose money. All make money at 3, so this is the equilibrium.

Slide9

Numerical Example:Policy Paper No. 21

9

Number of Firms (N)Firm Net RevenuesEntry CostsNet Profits12001518528015653401525424159N* = 51615161015-57815-7

The Equilibrium Number of Firms is 5. If the 6

th

firm enters, all firms lose money. All make money at 5, so this is the equilibrium.

Slide10

Numerical Example:Policy Paper No. 21

10

Number of Firms (N)Firm Net RevenuesEntry CostsNet Profits1100595240535320515412575853N* = 6550745-1

The Equilibrium Number of Firms is 6. If the 7

th

firm enters, all firms lose money. All make money at 6, so this is the equilibrium.

Slide11

Competition

11

General view is the price falls as the number of firms increases

This is the Cournot View with Homogeneous Goods

Or, Bertrand with Product Differentiation

The full range of outcomes is possible theoretically

P

MMC0

1 2 3 4 5

Price

N









 Cournot



Bertrand

Collusion

A

Slide12

What About Competition?

12

Intense Price Competition

Moderate Price Competition

Perfect Collusion

N

Entry Costs

Net Rev.

Net Profit

Net Rev.

Net Profit

Net Rev.

Net Profit

1

15

100

85

100

85

100

85

2

15

28

13

40

25

50

35

3

15

12

-3

20

5

33

18

4

15

6

-9

12

-3

25

10

5

15

4

-11

8

-7

20

5

6

15

3

-12

5

-10

17

2

7

15

2

-13

4

-11

14

-1

Slide13

Competition and N

13

There’s a Difference:

Intensity of Competition

Number of Firms

The Intensity of Competition relates to how firms interact

Price reductions from additional firms is based on more firms competing, not necessarily a change in the way they behave with respect to price

Intensity is behavior, not count, and too much intensity leads to very few firms

Slide14

The Evidence

14

Can we form a general expectation about the intensity of rivalry?Empirical studies are mixedExperimental work suggests the tendency is Cournot or better.

Experimental Results:Fouraker and Siegal (1963)TypeOutcomesCollusion3Weak Competition1Cournot7Bertrand5Total16

75% were Cournot or Better.

Slide15

More Like Cournot

More Like Bertrand

Price is inversely related to industry concentrationP = f(HHI)P = f(N)

Price is not related to industry concentration for 3 or more firmsP≠f(HHI)P≠f(N)Price cut occurs at 2nd firm. Difficult to distinguish between Cournot with only 2 firms.Also difficult to differentiate when N gets large

15

Interpreting the Evidence

Slide16

Too Much Competition?

16

Policy Paper No. 24: Network Neutrality and Industry Structure

Argument: If differentiation is prohibited by rule, firms are homogeneous, leaving only price to compete over.

Intense price competition can deter entry, leaving monopoly and no price competition at all.

Paradox

Differentiation softens price competition, but allows for price competition.

Slide17

What’s the Benchmark?

N* = 2

17

Perfect CompetitionMarginal Cost = MCFive FirmsP5Equilibrium Industry StructureP2 is as low as it goesProfit ≈ 0N=3, Profit < 0

P

MP2P5MC0

1 2 3 4 5

Price

N









 Cournot

Slide18

What’s the Benchmark?N* = 2

18

Perfect CompetitionMarginal Cost = MCProfits = 0Marginal Cost Pricing Not PossibleLosses = abcdEquilibrium Industry StructureCan only satisfy zero profit condition with fixed/sunk costs MC Pricing not possible without massive subsidies

P

2MC0

$

q

Firm Demand

Avg. Cost



a

b

c

d

Equilibrium

N=2

Slide19

The 800 Pound Gorilla …

19

REGULATION

V.

Limited Competition

Slide20

Competition and Investment/Innovation

20

Investment and Innovation are higher with competition, up to a pointEffect 1. Monopolies have the profits to fund innovation, but face no pressure to do so.Effect 2. Competition reduces profits, but firms innovate to escape competition (create market power).

Rate of Innovation

Competition

Maximum



*

Aghion

et al, Competition, Imitation and Growth with Step-by-Step Innovation, Review of Economic Studies (2001).

Slide21

Efficacy of Regulation(and it’s role)

21

How good is regulation in a Monopoly setting?

How good is regulation in a Duopoly setting?

How good is regulation in a

Triopoly

setting?

And so forth?

Is limited competition better than regulation?

What can regulation improve in a setting of limited competition?

What the role of regulation if we have N* and its small?

What N* = 2, Profit = 0, yet there is collusion?

Slide22

Congress’ View:The Cable Act of 92

22

1992 Cable Act deregulated a cable system when it had a competitor that passed 50% of its market (and had a penetration rate of at least 15%).

Deregulate at 1.5 firms

Small systems were also unregulated

Benefits < Costs

Unbundling/Special Access

Triggers (a bit nutty)

Slide23

Successes and Failures

23

Cable Regulation – FailureSpecial Access – FailureUnbundling – FailureInterconnection Reform – FailureUSF – Broken? Broadband Ubiquity by 2007 – FailurePayphone Compensation – Repeated FailureWireless E911 - FailureMadison River?Part 68 Rules – Success (Monopoly Period)Number portability wireline, wireless, but not between – SuccessE911, Wireline and VOIP, not wireless – SuccessSpectrum Auctions – SuccessOther ….

How do we pick successful interventions?

Slide24

Special Access Profit Margins

24

Pricing Flexibility/

Deregulation

Slide25

Madison River

25

VOIP-Capable DSL = $30VOIP-Blocked DSL = $20FCC Says No VOIP-Blocking, or No Price DifferencesVOIP-Capable DSL = $25 (or something like that)Now everyone pays $25, where many would prefer to pay $20Under plausible conditions, such a decision is welfare reducing (consumer welfare reducing!)

Difference is opportunity cost of lost profit from phone service.

Slide26

Regulation versus Limited Competition

A little competition probably wins this battle.Regulation is typically for things that the market cannot fix (Externalities, Natural Monopoly, Social Programs).Price and Quality are the purview of the market.Firms can impede competitionPerfection is not possible by regulators or markets, because perfection is subjective. Firms typically meet the average needs (cost-benefit test)Some Won’t be Happy

26

Who’s driving policy today?

Slide27

Be Realistic About Markets

Markets don’t resolve problems in an instantIt is the loss of a large number of customers that reveals a bad business decision. That takes time.Complaints and news articles are the manifestation of this emigration. It is often evidence the market is working, not failing.There will be complaints – people complain incessantly about Wal-Mart. Some people complain about their Toyota or Honda, despite being highly reliable, affordable autos.Complaining is not evidence of market failure.Service prices and quality YOU or I don’t like is not market failure.

27

Slide28

Be Realistic About Markets

Home NetworkingNetwork Neutrality proponents use as exampleMarket eventually solved it without regulationBluetooth CripplingAnother Net Neutrality exampleMarket solved itWIFI CripplingAnother Net Neutrality example Market solved it

28

Firms are out for profits, so expect manifestations of that. But consumers are out for the best deal, and expect that discipline. Watch for long-term problems.

Slide29

A Review of Broadband Comparisons and Rankings

29

What About the Evidence?

Slide30

Broadband Rankings, OECD

200120022003200420052006KoreaKoreaKoreaKoreaIcelandDenmarkCanadaCanadaCanadaDenmarkKoreaNetherlandsSwedenBelgiumIcelandNetherlandsNetherlandsIcelandU.S.IcelandDenmarkIcelandDenmarkKoreaDemarkNetherlandsCanadaSwitzerlandSwitzerlandSwedenBelgiumSwitzerlandFinlandNorwayNetherlandsSwedenBelgiumNorwayFinlandU.S.JapanJapanCanadaSwedenSwitzerlandFinlandSwedenCanadaU.S.NorwayBelgiumBelgiumSwedenJapanUKU.S.UKLuxembourgU.S.FranceJapanU.S.

30

Slide31

Broadband Rankings

What should ourbroadband rankingbe?

31

Slide32

Broadband Rankings

What are the policy ramifications of our rank?

32

Slide33

Broadband Rankings, OECD

2001

20022003200420052006KoreaKoreaKoreaKoreaIcelandDenmarkCanadaCanadaCanadaDenmarkKoreaNetherlandsSwedenBelgiumIcelandNetherlandsNetherlandsIcelandU.S.IcelandDenmarkIcelandDenmarkKoreaDemarkNetherlandsCanadaSwitzerlandSwitzerlandSwedenBelgiumSwitzerlandFinlandNorwayNetherlandsSwedenBelgiumNorwayFinlandU.S.JapanJapanCanadaSwedenSwitzerlandFinlandSwedenCanadaU.S.NorwayBelgiumBelgiumSwedenJapanUKU.S.UKLuxembourgU.S.FranceJapanU.S.

33

Slide34

Normalization of Subscription Counts

OECD ranks subscriptions per capitaPer Capita is a reasonable choice for normalization, but it is not an innocuous adjustment.Any adjustment introduces error; may not make much of a difference, but it may.

Home A: 4 peopleHome B: 3 peopleIf both have a broadband, then Home A has a subscription rate of 25 while Home B has a subscription rate of 33.Household normalization helps, but does not control for businesses.

34

Slide35

United States

Sweden

Persons per Home = 2.7

Persons per Home = 2.0

35

Normalization

The U.S. needs 35% more connections than Sweden to make up the difference.

Slide36

2006 Broadband Ranking2003 Telephones RankingDenmarkLuxembourgNetherlandsSwedenIcelandIcelandKoreaNorwaySwitzerlandDenmarkNorwaySwitzerlandFinlandFinlandSwedenGreeceCanadaUKBelgiumItalyUKGermanyLuxembourgNetherlandsFranceSpainJapanCzech Republic

Some countries tend to rank high in per-capita communications consumption.The U.S. presently ranks 24th in telephone per capita (wireless and wireline).Over the period 1991-1998, we ranked about 7th. After 1998, the U.S. has fallen to quickly to the low 20’s (presumably due to mobile phones and different counting schemes).Who ranks high?

High Rank for Communications

36

Slide37

Broadband Rankings, OECD

2001

20022003200420052006KoreaKoreaKoreaKoreaIcelandDenmarkCanadaCanadaCanadaDenmarkKoreaNetherlandsSwedenBelgiumIcelandNetherlandsNetherlandsIcelandU.S.IcelandDenmarkIcelandDenmarkKoreaDemarkNetherlandsCanadaSwitzerlandSwitzerlandSwedenBelgiumSwitzerlandFinlandNorwayNetherlandsSwedenBelgiumNorwayFinlandU.S.JapanJapanCanadaSwedenSwitzerlandFinlandSwedenCanadaU.S.NorwayBelgiumBelgiumSwedenJapanUKU.S.UKLuxembourgU.S.FranceJapanU.S.

37

Slide38

Time and Subscription

Time

Subscription

Rate

Measure

At

Time t

0

Measure

At

Time t1

A fall in rank may suggest leadership.

Must have some guesstimate of terminal subscription for rank to relevant.

38

Slide39

Household Penetration of Broadband:US and EU OECD Members

Source: EU Commission, E-Communications Household Survey (Apr. 2007). U.S. from Pew Internet and American Life Project (2007).

39

Complaint: Different studies/methodologies. But, where does the OECD data come from?

Slide40

Broadband Nirvana:All Homes and Business Have Broadband

CountrySubscriptionRankCountrySubscriptionRank Sweden54.1%1 Norway40.3%16 Iceland48.9%2 New Zealand39.8%17 Czech Republic47.8%3 Portugal39.2%18 Denmark47.8%4 Japan39.0%19 Finland47.7%5 United Kingdom38.9%20 Germany44.9%6 United States38.0%21 Netherlands43.7%7 Luxembourg37.8%22 Australia31.5%8 Greece36.2%23 Switzerland42.9%9 Slovak Republic35.1%24 France42.4%10 Ireland34.7%25 Canada41.9%11 Poland34.1%26 Hungary41.1%12 Spain33.8%27 Belgium41.%13 Korea25.4%28 Austria40.6%14 Mexico24.7%29 Italy40.4%15 Turkey21.2%30

40

Slide41

Determinants of Broadband Subscription

The Broadband Nirvana ignores the impact of income, age, density, and so forth on subscription.Phoenix Center Policy Paper No. 29 measures the impact of a wide range of determinants of broadband subscription to assess whether or not countries are exceeding, meeting, or falling short of “expectations.”Relevant determinants are: Price (-), Income (+), Income inequality (-), Education (+), Age (-), Density (+), Phones (+), Business Size (-), Time (+).

41

Slide42

Algebra: y = a + bx

Econometrics: Regression

42

Regression Analysis

25

20

15105

0 1 2 3 4 5

y

x











y = 30 – 5x

25

20

15105

0 1 2 3 4 5

y

x











We try to fit the curve to “messier” data

























Slide43

Univariate Regressiony = a + bx

Multivariate Regression:y = a + bx + cz

43

Regression Analysis

25

2015105

0 1 2 3 4 5

y

x

25

2015105

0 1 2 3 4 5

y

x









































































Slide44

Effect Size of Determinants

VariableEffect for a 10% IncreaseVariableExplanatory PowerGINI-8.4%AGE6550GDPCAP+8.1%GDPCAP42AGE65-7.3%BUSSIZE41BUSSIZE-3.5%DENSITY31PHONE+2.5%GINI30PRICE-2.3%PHONE20EDUC+1.8%PRICE18DENSITY+0.8%PHONE218BIGCITY-0.6%EDUC13HHSIZE+0.4%BIGCITY04HHSIZE00

44

AGE65 Japan = 26.9% AGE65 Korea = 12.4%

Slide45

Broadband Policy

45

90% of differences in broadband subscription across OECD is explained by non-policy factors

Slide46

Broadband Performance Index(Alternative Specification)

46

Exceeds Expectations

Below Expectations

Meets Expectations

Slide47

What Should We Rank?

47

Estimate the BPI regression only for the top 15 of OECD countries excluding the U.S.

Coefficients are driven by the most successful countries

U.S. Ranks 10

th

Absent reformulations of the counting schemes across countries (like for wireless), the U.S. should rank about 10

th

in subscriptions per capita

This is roughly consistent with its rank for telephones/pop before wireless

Slide48

From Rank to …

The rank statistics often precedes claims aboutLack of deploymentLack of Bandwidth/SpeedUbiquitous availability might move us up 4 spots in the rankingsAssuming 10% unavailability, we go from a subscription rate of 0.196 to 0.218.Assuming 5% unavailability, we go up 3 spots.

48

Slide49

Does Speed Matter?

Higher speeds will only draw in the marginal usersThose that want speed the most have already subscribed to broadbandSpeed is unlikely to increase our broadband ranking, or reduce it

SpeedValue AValue B14052501036015470205802569030710035811040

At $30, the high-value customer buys broadband at any speed. Speed must be 6Mbps for the low-value customers to buy. POINT: Higher speeds impact only marginal users.

49

Slide50

Speed and Subscription:No Help Here

Speed

Subscription

Rate

After about here, the additional impact of speed on subscription is trivial.

For OECD countries, advertised speeds (provided by ITIF) are unrelated to broadband subscription rates in a well specified model.

50

Slide51

Speed and Time

Time

Speed/Technology

Technology improves over time.

Investments are made at time t.Network remains for x years.Later investments are in better technology.Perpetual game of catch-up or leap frog.

Invest

Use

Invest

Use

Are we paying the price for being a first mover?

51

Slide52

Time and Subscription

Time

Subscription

Rate

Measure

At

Time t

0

Measure

At

Time t1

A fall in rank may suggest leadership.

Must have some guesstimate of terminal subscription for rank to relevant.

52

Slide53

The Demand for Bandwidth: Korea (KTT)

ServiceUpDown% SubsLite640Kpbs4Mpbs84%Premium4Mpbs13Mpbs4%Special4Mpbs20Mpbs5%Ntopia4Mpbs50Mpbs12%

Source: S. Wallsten, Progress on Point 14.13 (June 2007).

53

Slide54

Broadband Speeds

CountryClaimed Speed (Mbps)Actual Speed(Mbps)Japan61.012.0Korea45.63.4Finland21.73.6Sweden18.27.4France17.64.7Netherlands8.85.0Portugal8.13.1Canada7.63.6Poland7.51.6U.S.4.84.7

Sources: Claimed Speed, D. Correa, ITIF (2007). Actual Speed, www.speedtest.net (2007); Korea may be slightly understated due to a lack of a local test server. Sample sizes in Speedtest.net data are often very large.

Of actual speeds, U.S. ranks 8th of 100 countries.

54

Slide55

The Japanese Miracle?

Japan is touted as a “broadband miracle”Claim: 100 Mbps Fiber Service for $30RealityAvg. Download Speed = 10Mbps100Mbps = $53/monthJapan is ranked 14th in OECD, basically with the same subscription rate as U.S.58% DSL; 29% Fiber; 14% CableZero and Low Interest Loans for NetworksTax Incentives (not very important with 0% loans)33% of Municipal System Construction Costs are Subsidized

Is it unbundling or zero-interest loans?

55

Slide56

Investments in Broadband

K = Investment Amountr = Hurdle RateP = Price you can sell it for (price structure)c = incremental costn = number of buyersI = Information availableQ = Quality of ServiceZ = Other stuff like income, education, etc…

K(1 + r) ≤ (P-c).n(P,I,Q,Z)

56

Slide57

Investments in Bandwidth

Cost of the upgrade must exceed the net revenue from the sale of the services made possible.If you upgrade your network to 50Mbps, how many people will buy it?Communications carriers can only spend what investors will give them. Investors demand a return. That’s the world we live in.If you don’t reward investors, the money won’t come.

Cost≤ Benefit

57

Slide58

Investments in Bandwidth

"It is not the multitude of alehouses … that occasions a general disposition to drunkenness among the common people; but that dis-position, arising from other causes, necessarily gives employment to a multitude of alehouses."

Adam Smith, Wealth of Nations (1776)

58

Slide59

Investment in Bandwidth

MARKETS DON’T WORKThat depends on what you want markets to do.Rather than saying the market has failed to provide adequate speed and coverage, why not say the market is providing us information that the demand for speed is low and the demand is inadequate to cover costs in some markets. A proper perspective leads to proper policy responsesMarkets work because they are the fastest way to disseminate and incorporate information. Look to the market as a source of information. An all-knowing being could mimic the market (but I’m not interested in the job).

59

Slide60

Investments in Broadband

K = Investment Amountr = Hurdle RateP = Price you can sell it for (price structure)c = incremental costn = number of buyersI = Information availableQ = Quality of ServiceZ = Other stuff like income, education, etc…

K(1 + r) ≤ (P-c).n(P,I,Q,Z)

60

Slide61

Investments in Broadband

ActionEffectPrice RegulationProbably Less InvestmentNet NeutralityProbably Less InvestmentTax IncentivesMore InvestmentLow Interest LoansMore InvestmentSubsidiesMore Investment (if done right)Connect KentuckyMore InvestmentMore CompetitionDepends (up or down slope?)

K(1 + r) ≤ (P-c).n(P,I,Q,Z)

61

Slide62

Firms Invest Less than Society Desires(Complaints are to be expected)

Private Decision: K(1 + r) ≤ (P-c).n(P,I,Q,Z)

Social Decision: K(1 + r*) ≤ (P-c).n(P,I,Q,Z) where  = ratio of 1 plus the Avg. Consumer Surplus per unit relative to profit and r* ≤ r.

Interestingly, charges different prices to different people/groups reduces the difference.Regulatory risk inflates r. FCC says competition inflates r.

62

Slide63

[Preliminary]BPI for U.S. States

63

What about the US?

Slide64

Effect Size of Determinants

VariableEffect for a 10% IncreaseVariableExplanatory PowerINSCHOOL-22.0%RURALFARM0.57NATIVE-20.0%NATIVE0.40GINI-13.2%RURAL0.37HHSIZE-12.8%GINI0.36ENGLISH+8.7%ENGLISH0.34RETIRE-3.4%HHSIZE0.14PERCAPINC+3.3%RETIRE0.12SELF EMP.+2.6%SELF EMP.0.11RURAL-2.3%INSCHOOL0.08EDUC+2.1%PERCAPINC0.08BUSSIZE+1.5%BUSSIZE0.06RURALFARM-1.4%EDUC0.05

64

Slide65

BPI – Sort of(Preliminary)

65

Slide66