GEORGE FORDCHIEF ECONOMIST

THE PHOENIX CENTER

July 15 , 2009Nat ional Press C lub

The Broadband Adoption Index:

Policy Paper No. 36

Create an economically meaningful and policy relevant index of

broadband adoption that also is capable of handling heterogeneous

connection modalities (e.g. fixed and mobile).

If the $7B achieves universal availability over the next few years, the U.S. will probably

not change rank.

General Form

Target

ActualttBAI

Goal:

1. Provide for meaningful performance evaluation across geo-political units (intra- and internationally).

2. Incorporate the underlying economics of adoption and deployment

3. Accommodate different connection modalities

Population is not a Target

Population is not a target of adoption as implied by OECD statistics

In the U.S., telephones were ubiquitously available and nearly universally adopted, yet TEL/POP was only 0.493.

In Sweden, it was the same, but their TEL/POP was 0.686.

Broadband Nirvana(Phoenix Center Policy Paper No. 29, July 2007)

Country Subscription Rank Country Subscription Rank

Sweden 0.541 1 New Zealand 0.398 16

Iceland 0.489 2 Portugal 0.392 17

Czech Republic 0.478 3 Japan 0.39 18

Denmark 0.478 4 United Kingdom 0.389 19

Finland 0.477 5 United States 0.38 20

Germany 0.449 6 Luxembourg 0.378 21

Netherlands 0.437 7 Greece 0.362 22

Switzerland 0.429 8 Slovak Republic 0.351 23

France 0.424 9 Ireland 0.347 24

Canada 0.419 10 Poland 0.341 25

Hungary 0.411 11 Spain 0.338 26

Belgium 0.41 12 Australia 0.315 27

Austria 0.406 13 Korea 0.254 28

Italy 0.404 14 Mexico 0.247 29

Norway 0.403 15 Turkey 0.212 30

BB/POP tells you Little

0

0.2

0.4

0.6

0.8

1

Shareof

PotentialMarket

Population

1.0

OECD (BB/POP)

Ignores business connections (could assume proportional to households and scale up; no loss of generality).

Hmax = 0.33

Economy APop/HH = 3

(eg, Portugal)

Hmax = 0.50

Economy BPop/HH = 2(eg, Sweden)

Range of Deception(Economy A outperforms B, but BB/POP says otherwise.)

Trends in OECD Rank: The Fall (Connections/Capita)

2001 2002 2003 2004 2005 2006 2007 2008

Korea Korea Korea Korea Iceland Denmark Denmark Denmark

Canada Canada Canada Denmark Korea Netherlands

Netherlands Netherlands

Sweden Belgium Iceland Netherlands

Netherlands

IcelandIceland Iceland

U.S. Iceland Denmark Iceland Denmark Korea Norway Norway

Demark Netherlands

Canada Switzerland

Switzerland

Switzerland Switzerland

Sweden Belgium Switzerland

Finland NorwayFinland Finland

Netherlands

Sweden Belgium Norway FinlandKorea Korea

U.S. Japan Japan Canada Sweden Sweden Sweden

Switzerland

Finland Sweden CanadaLuxembourg Luxembourg

U.S. Norway Belgium Belgium Canada Canada

Sweden Japan UK UK UK

U.S. UK Luxembourg

Belgium Belgium

U.S. France France France

Japan Germany Germany

U.S. U.S. US

Trends in OECD Rank: The Rise (Connections/Capita)

2001 2002 2003 2004 2005 2006 2007 2008

Korea Korea Korea Korea Iceland Denmark Denmark Denmark

Canada Canada Canada Denmark Korea Netherlands

Netherlands Netherlands

Sweden Belgium Iceland Netherlands

Netherlands

IcelandIceland Iceland

U.S. Iceland Denmark Iceland Denmark Korea Norway Norway

Demark Netherlands

Canada Switzerland

Switzerland

Switzerland Switzerland

Sweden Belgium Switzerland

Finland NorwayFinland Finland

Netherlands

Sweden Belgium Norway FinlandKorea Korea

U.S. Japan Japan Canada Sweden Sweden Sweden

Switzerland

Finland Sweden CanadaLuxembourg Luxembourg

U.S. Norway Belgium Belgium Canada Canada

Sweden Japan UK UK UK

U.S. UK Luxembourg

Belgium Belgium

U.S. France France France

Japan Germany Germany

U.S. U.S. US

THE PHOENIX CENTER

10

Trends in OECD Rank: The Rise (Connections/Capita)

2007

Denmark

Netherlands

Norway

Switzerland

Iceland

Finland

Korea

Sweden

Luxembourg

Canada

1996 PSTN Subscriptio

n RankTOP 10

Denmark

Netherlands

Norway

Switzerland

Iceland

Finland

Sweden

Luxembourg

Canada

Telecom Rank not in sequence.

OECD Rank (Fixed Telephones 1996; Broadband Dec 2007)

Name BB #

TEL #

Name BB #

TEL#

Name BB#

TEL#

Denmark 1 2 UK 11 12 Ireland 21 23

Netherlands 2 10 Belgium 12 17 Italy 22 20

Norway 3 7 France 13 9 Czech Republic 23 25

Switzerland 4 5 Germany 14 13 Hungary 24 26

Iceland 5 6 United States 15 16 Portugal 25 24

Sweden 6 1 Australia 16 11 Greece 26 14

Korea 7 21 Japan 17 15 Poland 27 29

Finland 8 8 Austria 18 19 Slovak Republic 28 27

Luxembourg 9 4 New Zealand 19 18 Turkey 29 28

Canada 10 3 Spain 20 22 Mexico 30 30

90% Match 80% Match 80% Match

THE PHOENIX CENTER

12

Food for Thought

Top 10 in broadband; 9 are Top 10 in Wireline Telephone (only 5 in 2001)

Bottom 10 in broadband; 8 are Bottom 10 in Wireline Telephone (7 in 2001)

Of the 14 above the U.S. in broadband, 12 are also above the U.S. in telephone subscriptions

Of the 15 below the U.S. broadband, 12 are also below the U.S. in telephone subscriptions

THE PHOENIX CENTER

13

Convergence to Telephone Rank

Time0 2 4 6 8 10 12 14

-20.00

-15.00

-10.00

-5.00

0.00

5.00

10.00

15.00 US

UK

SWITZ

MEXICO

LUXEM

AUSTRIA

Tele

ph

on

e R

an

k –

Bro

ad

ban

d R

an

k

* Most other countries follow a similar path.

THE PHOENIX CENTER

14

Terminal Expectations:Broadband and Wireline Telephone Ranks

Year(June Data)

Correlation Avg. Difference in Ranks

2002 0.600 5.8

2003 0.642 5.5

2004 0.668 5.1

2005 0.728 4.4

2006 0.772 4.1

2007 0.824 3.3

2008 0.861 3.1

Other Issues with OECD-Style Metric

Today, it includes only Fixed ConnectionsAt the request of a few countries, OECD now

plans to collect and report mobile broadband in the near future. Broadband = 256kbps or more Counted if you buy it or use it (in a specified interval of

time)But how is the data to be used as a measure of

adoption performance?

OECD Style

Index Fixed = Fixed/Population ?Index Mobile = Mobile/Population ?

How do you rank across 2 dimensions? Can we just add the two for a single index? Do we get the SIM card problem for Europe? What about countries that do mostly one and hardly

any of the other? What if the count is high but due to some having

multiple connections rather than many having at least one form of connection?

How do we deal with the fact that fixed is typically shared, mobile is often not (scaling problem).

How to Combine the Two?

Can we just add: Fixed + Mobile Fixed is mostly shared, Mobile is mostly consumed by

individuals Mobile counts will swamp fixed counts

Should we scale Fixed by the share rate Fixed + Mobile/ShareRate (sharerate may be hh size) Assumes Mobile is a low quality fixed

How do homogenize unlike things? We convert the counts to values What is the value of fixed? What is the value of mobile?

One Modality

**

,,t

Target

Actual

ff

tftft qv

qvBAI

vi* = average social value of a connection of modality i at the

“target”qi

* = quantity of connections of modality i at the “target”

What is the Value of Broadband?

Willingness to Pay (w)Social Premia (e)

Externalities Spillovers Etc.

Cost of Production (c)Net Social Value v = w + e - c

Target Adoption

w

c - e

$

q* qv=0

c

q’

Social Value

w

c - e

$

q*

Social Value

qv=0

Maximum Subscription is Not Ideal

As long as c - e > 0, 100% consumption is not ideal.

w

c- e

$

q*

A

qv=0

W

Welfare Loss from excess

consumption.

qi* qi

v=0

Vi*

Viv=0

Vi

SocialValue (V)

Social Value = A - B

Optimal Consumption Depends on Costs

If costs are higher, then optimal quantity is lower.

wc- e

$

q*

w

c- e

$

q* q

Low CostMarket

High CostMarket

q

Optimal Consumption Depends on Demand

If demand is lower, then optimal quantity is lower.

wc- e

$

q*

wc- e

$

q*

High DemandMarket

Low DemandMarket

qq

Simple Graph

w

c - e

$

q*

Avg. Value = v* = V/q*

Soc. Value = v*q* = V

q

V

BAI at Time t

CBABA

A

qv

qvA

qBAv

t

t

**11

11 /)(

Assumption: Marginal, thus average, valuation declines over time. Here, highest valued users adopt first.

wci

$

q*q1

A

BC

qv=0

Two Modalities (f, m)

****

,,,,

mmff

tmtmtftft qvqv

qvqvBAI

Does it simplify?

One Modality

*

,

f

tft q

qBAI

Diminishing Marginal Valuation

Simulation

Two Modalities, f and mf is sharedm is personalcf =40; cm = 20Max value for m is 100Average share rate: k = 2Scale f demand to 200 (= 100·2)Personal Market = 2,000 personsShared Market = 1,000 units (= 2,000/k)m is a mild net substitute for f

Willingness-to-Pay (Demand) System

fm

mf

mm

qq

qp

qp

1000

05.0200)05.0200(

2000100

100

Simulation Algorithm

ci

vi

qiqi* qi

v=0

Compute q*, V*, then scroll through quantities up to qv=0. We compute V at each quantity then compute weights.

Do so in 10 percentage point intervals, so we have a 11x11 matrix of wi’s.

BAI Simulation: Two Modalities

m ↓ f→ 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.1 30.3 43 53.7 62.4 69.2 73.9 76.7 77.4 76.2 73.1

0.2 42.9 54.7 64.6 72.6 78.8 83.1 85.5 86 84.7 81.4

0.3 53.4 64.3 73.4 80.8 86.4 90.2 92.2 92.5 91 87.7

0.4 61.8 71.8 80.2 86.8 91.9 95.2 96.9 96.9 95.2 91.9

0.5 68.1 77.2 84.8 90.8 95.2 98.1 99.4 99.2 97.3 93.9

0.6 72.3 80.6 87.4 92.7 96.6 99 99.9 99.4 97.4 93.9

0.7 74.5 81.8 87.8 92.5 95.8 97.7 98.3 97.5 95.4 91.9

0.8 74.5 81 86.2 90.2 92.9 94.4 94.6 93.5 91.2 87.7

0.9 72.5 78.1 82.5 85.8 87.9 88.9 88.8 87.5 85 81.4

1.0 68.4 73.1 76.7 79.3 80.9 81.4 80.9 79.3 76.7 73.1

BAI Simulation: Two Modalities(Zero costs; no substitution)

m ↓ f→ 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.1 19.0 27.5 35.0 41.5 47.0 51.5 55.0 57.5 59.0 59.5

0.2 27.5 36.0 43.5 50.0 55.5 60.0 63.5 66.0 67.5 68.0

0.3 35.0 43.5 51.0 57.5 63.0 67.5 71.0 73.5 75.0 75.5

0.4 41.5 50.0 57.5 64.0 69.5 74.0 77.5 80.0 81.5 82.0

0.5 47.0 55.5 63.0 69.5 75.0 79.5 83.0 85.5 87.0 87.5

0.6 51.5 60.0 67.5 74.0 79.5 84.0 87.5 90.0 91.5 92.0

0.7 55.0 63.5 71.0 77.5 83.0 87.5 91.0 93.5 95.0 95.5

0.8 57.5 66.0 73.5 80.0 85.5 90.0 93.5 96.0 97.5 98.0

0.9 59.0 67.5 75.0 81.5 87.0 91.5 95.0 97.5 99.0 99.5

1.0 59.5 68.0 75.5 82.0 87.5 92.0 95.5 98.0 99.5 100

BAI Simulation: Alternatives

Scenario 1 Cost of m (cm):

20 25 30 35 40 45 50 55 60

qm*/ qmw=0

0.57 0.52 0.47 0.42 0.36 0.31 0.26 0.21 0.16

qf*/ qfw=0

0.72 0.73 0.74 0.75 0.76 0.76 0.77 0.78 0.78

Scenario 2 Cost of f (cf): 40 45 50 55 60 65 70 75 80

qm*/ qmw=0

0.57 0.58 0.58 0.59 0.60 0.60 0.61 0.62 0.64

qf*/ qfw=0

0.72 0.68 0.65 0.61 0.57 0.54 0.50 0.46 0.41

Scenario 3 Max Value m

100 120 140 160 180 200 220 240 260

qm*/ qmw=0

0.57 0.64 0.70 0.73 0.76 0.79 0.81 0.82 0.84

qf*/ qfw=0

0.72 0.71 0.69 0.69 0.68 0.67 0.66 0.66 0.66

Can this be done?

Econometric Implementation

Data on What is Purchased? How much is Paid? Demographics of Buyers Cost data by modality

BAI can be computed using this data Example provided in Paper No. 36

Target Setting

Some Portugal Targets

Achieve at least 50% household broadband adoption.

Increase public access to public Internet locations (16 per 100 POP).

Increase number of computers in schools to one per five students.

100% of Central Government institutions with broadband access.

100% of hospitals with broadband access.

Target Setting

It is unreasonable to expect a relatively poor and uneducated country to have the same broadband deployment and adoption rates as a relatively rich and educated country.

Comparing such countries on per-capita terms says nothing about the success or failure of broadband policies.

50% adoption in Mexico or Turkey may be stellar, but in the U.S. would be considered a failure.

Target Setting

Variable Coef t-stat

C -9.95 -4.81

LN(PRICE) -0.39 -2.56

LN(GDPCAP) 0.35 2.46

LN(GINI) -0.73 -3.18

LN(AGE65) -0.29 -2.60

LN(URBAN) 0.99 3.89

LN(TEL) 2.81 3.50

LN(TEL)^2 -0.36 -2.73N = 30; June-08 data; R2 = 0.93

-.3

-.2

-.1

.0

.1

.2

.3

-3.2

-2.8

-2.4

-2.0

-1.6

-1.2

-0.8

65 70 75 80 85 90

Residual Actual Fitted

Nearly all (93%) of the differences in fixed connections per capita across countries are explained by few demographic and economic endowments.

Summary

Performance is a value-based conceptAny modality that generates value must be

included in performance measures Per- Capita Normalizations are misguided Anyway, not clear how to do it with multiple modalities

Combining heterogeneous modalities is tricky, but the problem is understood

The underlying economics of deployment and adoption must be considered for good policy Countries vary in their demand and cost profiles Maximal deployment/adoption assumes external effects are

enormous

Summary

Avoid the willy-nilly by focusing the analysis on the components of target setting Willingness to Pay Realistic size of the Social Premia Cost of deployment

44