Swedish Commodity Flow Surveys Evaluated – Statistics Sweden`s

Swedish Commodity
Flow Surveys Evaluated
– Statistics Sweden’s Experiences and Survey
Adjustments Since 2001
Session 10 ICES III Montreal 2007
Lars Werke
Assistant head at the unit for transport statistics
Statistics Sweden
Outline of presentation
• Background and purpose of the Swedish CFS.
• The Swedish CFS in perspective.
• Comparing CFS 2001 and CFS 04/05
• A pragmatic data collection approach
• Challenges ahead and conclusions
Background and purpose
Background and our comissioners
• Swedish Institute for Transport and Communications Analysis
• Infrastructure agencies (SNRA, SCAA, SMA, NRA, Vinnova)
• Responsible for infrastructure planning in Sweden
• Long term Goods Model (SAMGODS)
• 6 Different Economic models for prognosis and scenario analysis
• 2 Supply models , 2 Demand models 2 Effect models
• Both demand models models goods flows in all 289 municipalities
• Before CFS 2001 synthetic O-D matrix (not observed O-D flows)
Background and purpose
The purpose of the survey
1. To give a statistical description of the annual
commodity flow within Sweden and between Sweden
and foreign countries (with respect to weight/value,
means of transportation, cargo type ,cargo size and
industry)
2. Provide the SAMGODS models with the necessary
data
The Swedish CFS in perspective
Statistics Sweden and transport
• No experience with multimodal statistics
• Unimodal: sea goods and road goods
• Regional CFS on county level (ERG project 1988, Bjurklo)
• Contacted US Census Bureau (CFS 1993)
• EU Mystic-project 1998 ”carrier based”- approach
• Decided on adopting Census Bureau approach
• Test survey in 1996 and 1998 and first full-scale in 2001
The Swedish CFS in perspective
Swedish first CFS approach
• Stratified three-stage probability sample.
• Aimed at Mining, Manufacturing and Wholesale sectors
• Stratification: Geography, Local unit size and Commodity group
• Three stage sampling –>Local unit->Time period->Shipments
• Questionnaire different from US, divided in to three parts
Part 1: Outgoing shipment to own county
Part 2: Outgoing shipment to other county and abroad
Part 3: Incoming shipments from abroad
• Parts as a register based survey (Forestry, Agriculture)
Comparing CFS 2001 and CFS 04/05
The CFS 2001 results
Pro
• Response rate 78,4 %
• Outgoing shipments 246 million tonnes. SEK 1 905 billion
• Electronic reporting at 10,9%
• Commissioners content with quality and delivery
Con
• Lack of Control in the final stage of sampling (shipments)
• Instructions (3pl), (Incoterms), (Wholesale agents)
• Cut off limits (gas and oil)
• Undercoverage (sand and gravel)
• Report burden
• We had to revise the statistical report.
Comparing CFS 2001 and CFS 04/05
Preparing for CFS 04/05
• Should cover more sectors of the economy.
• Logistics model developed for SAMGODS.
• Additional variables in the questionnaire (P-C) (PWC).
• Increase electronic reporting.
• Decrease report burden.
• Improve overall design.
A pragmatic data collection approach
Going for the low hanging fruit
• Tried to look at the structure in different sectors
• Transport cost in relation to commodity value/weight ratio.
• Company level: Did all the units belong to a few companies.
• We thought of using different questionnaires.
• Administrative registers ”tracking” some goods were monitored.
• Large shipments by sea ”flagged” (paper mills etc.)
• Training of staff, sector specific, (distribution etc.)
• Developed of new simple control tools.
A pragmatic data collection approach
Altered design
• Added admin. registers to avoid undercoverage (sand)
• A revision of cut off limits with a focus on the wholesale sector.
• Increased the registry based part of the survey.
• Agricultural
• Energy sector
• We suggested specific industry surveys
• Petroleum (Oil & Gas)
• Cars and Trucks (Wholesale)
• Corn & Grain
A pragmatic data collection approach
Altered questionnaire
• Dropped ”ownership of goods” decided on ”handling of goods”
• Redesigned the questionnaire and instructions in to one form
• Added new variables after a test (type of receiving industry)
• We dropped ”hazmat” and advised against inclusion of PWC
• Added control questions , focus on large shipments
• Redesigned commodity code list.
• Encouraged electronic reporting and complete data sets.
A pragmatic data collection approach
Pro
• Response down from rate 78,4% in 2001 to 73,6 % in 04/05
• Outgoing shipments +15 % to 282 Million tonnes
• Outgoing shipments +10 % to SEK 2 093 billion
• Electronic reporting from 10,9% to 18 %
• Comparison to trade stat. export within 2 %
• Covered most of agricultural sector with admin. data
• Commissioners content with quality and delivery
Con
• Lack of Control in the final stage of sampling (shipments)
• Instructions (3pl), (Incoterms), (wholesale agents)
• Cut off limits (gas and oil)
• Undercoverage (sand and gravel)
• Report burden
Challenges ahead and conclusions
Challenges in 2009?
• CFS 04/05 is being evaluated by the commissioners.
• Report burden must decrease to improve response rate.
• Evaluate only one part for outgoing shipments.
• Combine collected admin. data with specific surveys.
• Improve measurement on imports.
• More industry specific surveys.
• Improve overall design.
Challenges ahead and conclusions
Conclusions or why does this work in Sweden
•
Sweden is a small open economy.
•
Government admin. data is easily accessible to
Statistics Sweden.
•
The three stage sample design is a good approach.
•
The increased response burden of the Swedish CFS
called for a new data collection approach that also
improved coverage of the survey.
•
Norway is preparing for a CFS with similar approach
•
EU commission working on ”Intermodal indicators”
Thank you for your attention