Approaches to measuring telecommunication services for the CPI - - PowerPoint PPT Presentation

Approaches to measuring telecommunication services for the CPI

Presenter: Michael Abbondante Presenter: Michael Abbondante Authors: Stephen Frost, Margaret Yang & Lewis Conn

Initial thinking & progress report

Context

Initial thinking & progress report Complex area of price measurement Fixed-line telecommunication tightly regulated Mobile telephony less so Mobile telephony less so ABS – matched sample

Real time flow of telephone bill samples providing quantity information

Possible approaches

quantity information Testing different approaches, using bills to

• 1. create usage profiles (notional quantities) priced

using publicly available price schedules.

• 2. repricing each individual transaction (such as a phone

call) recorded using publicly available price schedules.

• 3. calculate unit values

precise service identifier broad type of service duration (sec) total charge customer identifier destination pricing plan state postcode broad type of txn 221590USAGE 24 0.52 62400MNet A VIC XXXX ON_NET_ MOBILE

Bill sample

221590USAGE 24 0.52 62400MNet A VIC XXXX ON_NET_ MOBILE 221590USAGE 27 0.54 62400MNet A VIC XXXX ON_NET_ MOBILE A0LU8USAGE 231 1 62400United King A VIC XXXX IDD_CALL 122100RENTAL

• 33

62400 A VIC XXXX LINE_RENTAL 122100RENTAL 47.85 62400 A VIC XXXX LINE_RENTAL 568400USAGE 10882 12.4 62400 A VIC XXXX LOCAL_CALL 714000USAGE 0.35 62400 A VIC XXXX CALL_RETURN 378400USAGE 83 0.27 62400 A VIC XXXX ONE_THREE_CALL 74000USAGE 0.35 62400 A VIC XXXX CALL_RETURN 375400USAGE 175 0.28 62400 A VIC XXXX ONE_THREE_CALL A00LUDUSAGE 50 0.5 62400Cowes A VIC XXXX STD_CALL 375400USAGE 91 0.27 62400 A VIC XXXX ONE_THREE_CALL 375400USAGE 91 0.27 62400 A VIC XXXX ONE_THREE_CALL 629400USAGE 99 1 62400 A VIC XXXX OFF_NET_MOBILE 629400USAGE 624 4.24 62400 A VIC XXXX OFF_NET_MOBILE 229400USAGE 20 0.5 62400MNet A VIC XXXX ON_NET_ MOBILE 378400USAGE 46 0.28 62400 A VIC XXXX ONE_THREE_CALL 37400USAGE 292 0.28 62400 A VIC XXXX ONE_THREE_CALL

Bill segmentation

Calendar year by type of service (e.g. fixed-line, post- paid mobile)

Approach 1 – Average usage profiles

paid mobile)

Average usage profiles (quantities)

Grouped like bills into a set of usage profiles Weighting information (revenue) for each service type provided (e.g. mobile to mobile, text ) by the carrier Fixed for a given period

Pricing plans (price) Pricing plans (price)

• Priced using publicly available price schedules.

For future consideration

Frequency of updating quantities to reflect behaviour change e.g. internet usage by smart phones.

Chart illustrates a profile where SMS and

Approach 1 – Average usage profiles

5%

Example of a Mobile Profile

profile where SMS and data download are the predominant transactions

5% 25% 18% International calls Data downloads International SMS Message bank Calls diverted to mobile 0% 2% 0% 50% SMS Local & national calls

Chart illustrates a profile where local calls are the predominant

Approach 1 – Average usage profiles

5%

Example of a Fixed Line Profile

are the predominant transactions

50% 25% 5% Local Calls Off-net mobile calls 13 Calls National long distance calls 20% 0% On-net mobile calls

Sample of bills (quantity), repriced each period using latest information – complex in practice Attractive features

Approach 2 – Direct calculation

Attractive features

Observable call destinations and durations Ability to reprice each transaction Ensures representative quantities over time and discounts irrational consumer behaviour

Unattractive features

Data intensive Data intensive assistance and effort from telecommunication provider Commercially sensitive changing plan shares revenue information by service types for weighting Pricing plans available, but difficult to locate

Calculating expenditure – combining the fixed bill quantities with the plan prices Approach 2 – Direct calculation Classifying and plans (price) and bills (quantity)

• Plans (price) classified into high, medium and low-

usage with provider advice

• Bills classified into high (H), medium (M) and low (L)-

activity bills (quantities) by:

Could be based on plan type or Could be based on plan type or Calculate expenditure from applying a mid-range plan

• ver all the bills;

Classify bills into 3 types (H, M or L) by annual expenditure; and fix the activity shares for the life of each sample

• Quantity – telephonic

Price – Usage plans

Direct calculation

Quantity – telephonic transactions (activity) Price – Usage plans High Profiles/bills Plans (H prices) Medium Profiles/bills Plans (M prices) Low Profiles/bills Plans (L prices) Low Profiles/bills Plans (L prices)

Calculating total expenditure High usage bills expenditure

first high-usage plans (H1 prices) multiplied by all high-activity bills;

Approach 2 – Direct calculation

first high-usage plans (H1 prices) multiplied by all high-activity bills; second high-usage plans (H2 prices) multiplied by same high-activity bills and so on… H expenditure calculated using confidential plan shares (weights) to capture compositional shift

Medium usage bills expenditure

the medium-usage plans (the M prices) multiplied by all medium- activity bills activity bills M expenditure calculated using confidential plan shares (weights)

Low usage bills expenditure

the low-usage plans (the L prices) multiplied by all low-activity bills L expenditure calculated using confidential plan shares (weights)

Average H, M and L expenditures added together to provide a total expenditure

Calculating total expenditure for quarter 1

Approach 2 – Direct calculation

Plan share Weight Weight Weight Total bill for Plan Plan share (%) Usage Weight in H Weight in M Weight in L Bill for H Bill for M Bill for L for quarter H1 8H 32% $20,000 H2 12H 48% $22,000 H3 5H 20% $19,000 H4 1H 0% M1 20M 56% $15,000 M2 10M 28% $16,000 M3 6M 17% $13,000 M4 1M 0% M4 1M 0% L1 2L 0% L2 28L 80% $10,000 L3 7L 20% $9,000 Weighted average $20,760 $14,944 $9,800 $45,504

Calculating total expenditure for quarter 2

Approach 2 – Direct calculation

Plan share Weight Weight Weight Total bill for Plan Plan share (%) Usage Weight in H Weight in M Weight in L Bill for H Bill for M Bill for L for quarter H1 8H 27% $20,000 H2 12H 40% $22,000 H3 5H 17% $19,000 H4 5H 17% $24,000 M1 20M 56% $15,000 M2 10M 28% $16,000 M3 6M 17% $13,000 M4 1M 0% M4 1M 0% L1 2L 0% L2 24L 77% $10,000 L3 7L 23% $9,000 Weighted average $21,300 $14,944 $9,774 $46,019

Calculating the price movement

Approach 2 – Direct calculation

Weighted average bill High Medium Low Total bill for quarter Quarter 1 $20,760 $14,944 $9,800 $45,504 Quarter 2 $21,300 $14,944 $9,774 $46,019 Price movement 1.13%

Things to note

Re-weight prices continually according to proportion of consumers on each plan (within high , medium and low segmentation – fixed basket) Plan shares based on the ’number of services in operation’

Approach 2 – Direct calculation

Plan shares based on the ’number of services in operation’

For future consideration

Re-use bill (quantity) data by applying other carriers’ prices (for the same service) Consumers’ calling patterns are similar across carriers customers of major carriers have similar calling patterns, customers of smaller carriers exhibit different behaviour inappropriate to use post-paid mobile bills to price prepaid mobile services - behavioural differences behavioural differences These experimental estimates use one set of bills for all periods desirable to re-sample the bills annually to capture changing consumption patterns Prices should be re-weighted based on the shifting market shares of the carriers themselves Robustness to quality change

A daily electronic feed of the transactions accruing on the sampled accounts.

Approach 3 – Unit values

Product billing identifier defines the service, although identifiers change fairly frequently Potential to use the bills data to construct unit prices sum of all expenditure / sum of all durations by product type e.g. international by country Pros - efficient way of constructing price movements Cons - careful thought regarding nature of the sample: representativeness over time; and across carriers.

Assessment of approaches

Approach 1: Customer profiles - likely to provide good estimates of price change with less effort. Further research on application of pricing plans to customer profiles required pricing plans to customer profiles required Approach 2: Direct calculation - rigorous method, real world quantities observed. Labour and data intensive, difficult to implement in practice Approach 3: Unit values - using the full richness in the dataset shows great promise, further investigation required. shows great promise, further investigation required. Areas for additional investigation Pre-paid mobile pre-paid phone cards VOIP services

Thank you Questions?