QoS Project Group StCom Workshop

1
QoS Project Group StCom Workshop
UPU Global Monitoring System (UPU GMS)- Current
Status -

• Berne
• 12 June 2007

2
System Overview
3
Underlying concept for the Global Monitoring
System (GMS)
UPU Global Monitoring System

• Sole focus on Inbound stretch (no end-to-end
  measurement at the beginning)
• Country classification based on total inbound
  mail volumes and not on status (e.g.
  DC / IC)
• Category 1 lt 2,000 tons/year
• Category 2 1,000 to 1,999 tons/year
• Category 3 500 to 999 tons/year
• Category 4 250 to 499 tons/year
• Category 5 gt 250 tons/year
• Ensure affordability by implementing a
  volume-based approach
• High inbound real mail volumes High test mail
  volumes Higher accuracy
• Lower inbound mail volumes Lower test mail
  volumes Lower accuracy
• Same rules for all (depending on total inbound
  volumes only)
• System which ensures minimum statistical design
  but offers flexibility by allowing customisation
• Use of RFID transponder technology as the common
  technical basis for all
• Cost for the GMS is Inbound based meaning that
  countries pay for items they receive

4
Basic requirements
UPU Global Monitoring System

• Classification of countries into 5 categories
  solely based on inbound mail volumes
• Statistical parameters are seen as the minimum in
  order to be linked to TDs
• Many parameters can be upgraded (boosted) for
  each country if a higher statistical accuracy is
  required by the users

For category 5 the requirement to have 1
permanently measured flow might be relaxed in
cases where this would lead to substantially
higher costs.
5
Example for Category 1
UPU Global Monitoring System

• 18 biggest flows
• Permanently measured
• On each link 250 items/year
• Each flow weighted according to real mail volume

• 6.000 items / year
• /- 1 accuracy
• 4 to 8 cities

4,500 in total
(250 items per link)
Pool I Cat. 1-4 countries
Inbound OE

• Items are being sent randomly from both pools
• Pools technically treated
• as 1 country
• weighting according to
• real mail of all countries in a pool

1,000 in total
Pool II Cat. 5 countries
500 in total
6
Example for Category 4
UPU Global Monitoring System

• 3 biggest flows
• Permanently measured
• On each link 100 items/year
• Each flow weighted according to real mail volume

Country 3

• 900 items / year
• /- 3 accuracy
• 1 to 2 cities

300 in total
(100 items per link)
Pool I Cat. 1-4 countries
Inbound OE

• Items are being sent randomly from both pools
• Pools technically treated
• as 1 country
• weighting according to
• real mail of all countries in a pool

400 in total
Pool II Cat. 5 countries
200 in total
7
Weighting Model Scenario
UPU Global Monitoring System

• The distribution of countries into 2 separate
  pools is important in order to keep a link
  between the actual real mail volumes and the
  weight given to each test item
• Therefore the 2 pools provide a very simple and
  cost-effective weighting system
• However, the weighting model can -within limits-
  be adjusted by TDPG

Calculation of Total Result (cat. 1 country)
Permanent links
Result of flow 1
X
Weight of flow 1 20
80
Result of flow 2
Weight of flow 2 13
Weighted Total Result
up to link 18
up to link 18 1.5
Result of Pool I
14
combined real mail weight of all countries in
pool I
X
Result of Pool II
combined real mail weight of all countries in
pool II
X
6
100
Due to the fact that in many cases the total
mail weight of pool II might be too small to have
an impact on the final weighted result, a
threshold (e.g. of 5) could be considered.
8
Parameters in the GMS
UPU Global Monitoring System
9
Main system parameters currently under discussion
UPU Global Monitoring System

• Number of permanently measured flows
• The number of permanently measured flows is
  currently fixed for each category (see table
  above)
• The proposed number of links results from the
  division of the total amount of items which are
  required to achieve the statistical accuracy by
  the number of items each link should have in
  order to be weighted reliably
• Example category 1 foresees 4,500 items in total
  for permanently measured links and each link
  shall be measured with 250 test items, i.e.
  4,500/25018
• However, a fixed number might lead in some cases
  to unwanted design effects

• Considerations in regards to permanently measured
  flows
• Ideally 80 of the total inbound mail should be
  covered by permanently measured links and 20 by
  the pools I II
• Category 1 countries will probably overachieve
  this target of 80 regularly as categories 3-5
  will have difficulties to reach it
• In addition category 1 countries can not save
  on items as they are needed to achieve the
  statistical accuracy and category 3-5 countries
  would face an increase of items and therefore
  costs if 80 shall be achieved

10
Main system parameters currently under discussion
UPU Global Monitoring System

• Classification of countries into categories 1 to
  5
• The classification of a country is foreseen to be
  based on the total amount of inbound mail per
  year (see table above)
• Based on the classification each country will be
  either in pool I or pool II unless its volumes
  to a country are high enough to be covered
  permanently
• This approach has the advantage of simplicity
  both for the study design (hence the contractor
  and system manager) and the country itself as
  real mail volume information on each link is not
  required in all cases (e.g. for smaller countries)

• Considerations in regards to classification
• In some cases this fixed classification might not
  reflect the actual situation of a country
• Example Country X might be a category 5 country
  due to its rather small inbound volumes but may
  have a high-volume link to a neighbouring
  country. For this link the classification in pool
  I might be more accurate than pool II (which
  would be foreseen under the current model)
• It could therefore be considered to let the
  individual weight of each link determine whether
  the sending country shall be included in the
  inbound countrys Pool I or Pool II

11
City Country Coverage
UPU Global Monitoring System
In its initial phase the GMS design will only
foresee the coverage of selected cities in each
country. The extension to smaller cities and
rural areas is currently under development
(voluntary or mandatory for certain categories).

• City Coverage
• The number of cities which are required to be
  covered varies between the categories
• The exact number and location of cities each
  category has to cover depends on the geographical
  and demographical situation in the country
• The definition of a city includes the larger
  city area (e.g. suburbs)

• Open issues in regards to City Country coverage
• For categories 1 to 4 the coverage of smaller
  cities and rural areas could be justified or even
  requested by countries
• Can countries choose additional cities?

12
City Weighting
UPU Global Monitoring System

• In case several cities are measured within a
  country the question of city weighting arises. In
  principle 3 main approaches are possible
• weighting based on equal weights between
  inbound cities
• weighting purely based on real mail volume
  going to each city
• weighting purely to population in each city
• weighting according to a balanced approach
  (political vs. real mail weight)

• Considerations in regards to city weighting
• take each city not only as the city itself but as
  a good indication also for the quality given to
  the wider surrounding area (e.g. the entire
  state)
• therefore take the population for each city as a
  good indicator and add a cap and floor
  mechanism (e.g. no city less than 5 or more than
  50 weight)

13
Panel Requirements
UPU Global Monitoring System
The recruitment, management and the total number
of panelists are one of the main cost drivers in
the GMS. It is therefore important to find a good
balance between costs on one side and statistical
accuracy, system integrity and a good reflection
of reality on the other side.

• Panel
• The minimum number of panelists per city should
  not be lower than 4
• The panel will only consist of private receivers
  (for system management reasons)

• Open issues in regards to panel
• Will the panel be used for item production /
  induction?
• P.O. box delivery will need to be considered in
  some countries
• Can a lower number than 4 panellists per city be
  accepted if the number of cities are increased
  (by countrys request, f.eg) and/or rural areas
  are included in the GMS?

14
Boosting Options
UPU Global Monitoring System

• Options foreseen (to be extended)
• Increase of test mail volume on particular
  permanently measured links
• Increase of number of permanently measured links
• Increase of test mail volume in the pool

• General rules
• All extensions can only be applied to a whole
  year
• All extensions must follow the same allocation
  pattern as the regular items
• Only extensions are allowed which are
  specifically described in the project manual
• Always possible to upgrade (e.g. from category 4
  to category 3)

15
Estimated Costs
UPU Global Monitoring System
16
UPU Global Monitoring System
Costs for the Global Monitoring System

• The cost estimates encompass 3 distinct areas 1)
  the RFID costs, 2) the measurement system costs,
  and 3) the overhead and management costs.
• Apart from the cost estimates in the examples
  below, there will be additional costs for
  transponder investments and management of the
  GMS. It is premature to give a cost estimate at
  this time however, the impact on the total costs
  will not be dramatic.
• The cost for the planned GMS is expected to be
  considerably lower than the current IC-IC system

Examples for selected cases

• Country in Category 2 (countries with 1,000 to
  1,999 tons/year)
• Installation RFID 20,000 Euro (RFID standard
  system with up to 4 gates including travel and
  man days)
• Set-up Measurement 15.000 Euro
• Total Set-up 35,000 Euro
• Running Measurement 30,000 Euro
• Running RFID Standard System 2,000 Euro (i.e.
  data transfer)
• Total Annual Running Costs 32,000 Euro

Categories 4 5 see next slide
17
UPU Global Monitoring System
Costs for the Global Monitoring System cont.

• Country in Category 4 (countries with 250 to 499
  tons/year)
• Installation RFID 8,000 Euro (RFID single unit
  system with 1 reader site)
• Set-up Measurement 2.000 Euro
• Total Set-up 10,000 Euro
• Running Measurement 10,000 Euro
• Running RFID Standard System 1,000 Euro (i.e.
  data transfer)
• Total Annual Running Costs 11,000 Euro
• Country in Category 5 (countries with less than
  250 tons/year)
• Installation RFID 8,000 Euro (RFID single unit
  system with 1 reader site)
• Set-up Measurement 2.000 Euro
• Total Set-up 10,000 Euro
• Running Measurement 3,000 Euro
• Running RFID Standard System 1,000 Euro (i.e.
  data transfer)
• Total Annual Running Costs 4,000 Euro

18
UPU Global Monitoring System
Next steps
In the next few months the GMS DG will have to
deal with a number of issues, some of which are

• Develop full and detailed documentation of the
  entire GMS design concept, e.g.
• Development of accurate allocation model for test
  items (e.g. which countries/regions should
  send how many items to which countries/cities/area
  s?)
• Full revision of all statistical assumptions and
  models by independent experts
• Development of rules which address potential
  problems or even short-comings of the GMS already
  beforehand (in order to avoid unclear situation
  at a later stage)
• Ensure alignment between GMS proposal and
  requirements from QSPG SC
• Initiate a process which ensures maximum
  involvement and feedback on GMS concept proposal
  from Developing Countries
• Involve expert organisations through a Technical
  Advisory Group in the development of the GMS
  concept to ensure technical feasibility and
  efficiency
• Actively assist the members of the QoS Project
  Group SC with their respective projects

19
Possible Implementation Timetable
UPU Global Monitoring System
20
UPU Global Monitoring System
Implementation - reflections

• Since the UPU ICIC System already provides a
  solution for Industrialised Countries, the main
  focus should be to first offer the GMS to the
  Developing Countries
• As of 2011 (or later) the current ICIC countries
  will join the GMS
• Participation in the starting phase of the GMS
  should be in batches of a limited number of
  countries
• 2008 could be considered for a pilot and 2009 as
  the official start of the GMS
• During the starting phase particular focus will
  be given to provide a stable system as well as to
  address potential problematic issues in regards
  to the diagnostic monitoring (transponder gates),
  the panel and the system management
• Roll-out of system will be spread over several
  years in order to maintain control of system
  (e.g. maximum of 15-25 new countries each year)
• The final stage with having all countries
  integrated into the GMS will realistically not be
  reached before 2014 or 2015

21
Timetable
UPU Global Monitoring System