1. Session on adjusting for quality change 27 th Voorburg Group Meeting Warsaw, Poland André Loranger October 1, 2012

2. Goal of this session  This session is a continuation of last year’s discussion on quality change. The objective of the session are: 1.Review the issues raised at last year’s meeting. 2.Present and discuss the paper and the outcome of the consultation with price experts. 3.Outline and discuss options for moving forward on quality adjustment issues. 15/09/2015 Statistics Canada Statistique Canada 2

3. Outstanding questions and recommendations, VG 2011  Advertising and Air Transport Main question: Should consumer utility or changes to the production function be the basis for the quality adjustment for SPPIs? Outcome and recommendations: The majority present supported the practice of only quality adjusting for changes in the production function while a minority supported considering consumer utility. 15/09/2015 Statistics Canada Statistique Canada 3

4. Outstanding questions and recommendations, VG 2011  Distributive Trades (wholesale and retail services) Main question: Should the good being traded be quality adjusted when adjusting the service? Outcome and recommendations: No consensus regarding whether quality adjustments should be made to the goods resold when calculating a margin price index. Countries should instead consult with their own SNA for direction.  The Ottawa Group on Price Statistics should be consulted on matters pertaining to quality adjustment 15/09/2015 Statistics Canada Statistique Canada 4

5. Consultation with CPI Experts  The Ottawa Group did not meet over the last year instead a paper summarizing the issues was written and presented at the UNECE-ILO meeting of the Group of Experts on Consumer Price Indices in Geneva, Switzerland this spring. As part of the institutional review process, the paper was reviewed was by several price experts including members of Statistics Canada’s Price Measurement Advisory Committee. 15/09/2015 Statistics Canada Statistique Canada 5

6. Summary of Geneva paper  Provided background on PPIs, the Voorburg Group and VG discussions on quality change  Summarized the issues and discussion from VG 2011 for Air Transport, Advertising and Distributive Trades  Expanded arguments for the production function view  VG 2012 version of the paper also discusses recommendations from Geneva and a potential way forward for the VG 15/09/2015 Statistics Canada Statistique Canada 6

7. Expanding the arguments for the production function view  Utility curve example, achieved vs. expected audience size  Discrepancy between producer and consumer valuations of quality, aircraft seat example  Triplett argument for resource-cost view  Distributive trades, QA for service or good or both 15/09/2015 Statistics Canada Statistique Canada 7

8. Utility and production functions  P is surface that traces out all combinations of 2 characteristics Z 1 and Z 2  q is indifference curve that maps out all combinations of Z 1 and Z 2 that the consumer is indifferent against purchasing  S is production function (combination of inputs and technology  Service delivery occurs at A, the intersection of the optimal production and utility at a given price 15/09/2015 Statistics Canada Statistique Canada 8

9. Utility and production functions “Expected Audience” example  In Q1, P = 1000 Expected audience = 1 M  In Q2, P = 1500, Expected Audience = 1 M  P = f(Expected Audience)  Achieved Audience does not factor into production or consumption decision, although it may lead to greater consumer utility  Don’t adjust for audience size, pure price effect 15/09/2015 Statistics Canada Statistique Canada 9

10. Utility and production functions “Achieved Audience” example  In Q1, P = 1000, Achieved Audience = 1 M, P perAchievedViewer = 0.001  In Q2, P = 1500, Achieved Audience = 1.5 M P perAchievedViewer = 0.001  P = f(Achieved Audience)  Achieved Audience is a price determining characteristic, therefore must adjust for quality  What is the appropriate basis for making the adjustment? Potential discrepancy between producer and user valuations. 15/09/2015 Statistics Canada Statistique Canada 10

11. Discrepancy between producer and consumer valuations of the QA Aircraft seat example 15/09/2015 Statistics Canada Statistique Canada 11 Observed price (£) SPPI Quality adjust- ment (£) Quality Adjusted Price (£) Price index SPPI Observed Turnover (£) Derived (constant price) output Change in volume 2009140 1005,000,000 - 2010126141401005,250,000 5.0% Observed price (£) CPI Quality adjust- ment (£) Quality Adjusted Price (£) Price index CPI Observed Consumer Expenditure (£) Derived (constant price) expenditure Change in volume 2009140 1005,000,000 - 2010126281541105,250,0004,772,727-4.5%  Change in configuration of aircraft, increase number of seats (smaller), change in production function  Leads to a decrease in operating costs and a decrease in price  To deliver same level of quality, firm needs to increase price by 14 (QA)  Leads to increase in volume  Consumers place more value on leg room than the QA estimated from the producer side. A QA of 28 is required to get back to same level of quality.  Translates into a 10% increase in CPI  Leads to decrease in volume

12. Air Transport, Additional thoughts  Unit value (Revenue per flight) is one of the recommended approaches for measuring price change in this industry  Production function approach to QA fits well with this model. Quality change based on changes to inputs (seating configuration, flight crew, etc., fuel, etc.) Fix quality of the flight, not the individual consumer. 15/09/2015 Statistics Canada Statistique Canada 12

13. Discrepancy between producer and consumer valuations of the QA Aircraft seat example, continued 15/09/2015 Statistics Canada Statistique Canada 13  The supply-demand identity for a given product or service in the SNA is: Supply = output + imports + transport margin + trade margin + taxes – subs. on products = intermediate consumption + final consumption expenditure + gross capital formation + exports = Demand  Assuming that consumers purchase 100% of the output and that there are no margins or taxes, it follows that: output = final consumption expenditure  Identity cannot be resolved in real terms if PPI and CPI are different

14. Additional thoughts  From Jack Triplett, “Concepts of Quality in Input and Output Price Measures. A Resolution of the User Value Resource-Cost Debate.” There are two different uses of the data (input measures and output measures) and that input and output price indexes imply different theoretical price index treatments. In a resource cost view, the cost of making a machine is the proper basis for making quality adjustments, not the productivity of using these machines to produce other goods.  “The principal conceptual basis for the output PPI is the fixed-input output price index (FIOPI). The output PPI thus aims to measure an output price index constructed on the assumption that inputs and technology are fixed.” PPI manual, 7.44  Resource-cost view is conceptually consistent with the Fixed-input output price index (FIOPI), therefore quality adjustment based on the production function would seem to be appropriate for SPPI 15/09/2015 Statistics Canada Statistique Canada 14

15. Wholesale and Retail Services Quality adjusting goods or service portion?  SNA output definition excludes the good The recording in the SNA of transactions for wholesalers and retailers does not mirror the way in which those involved view them. The purchases of goods for resale by wholesalers and retailers are not recorded by these units explicitly, and they are viewed as selling, not the goods, but the services of storing and displaying a selection of goods in convenient locations and making them easily available for customers. This partitioning measures output for traders by the value of the margins realized on goods they purchase for resale. (SNA 2008, 3.68)  If output definition excludes the good then QA should not be carried out on the underlying good. 15/09/2015 Statistics Canada Statistique Canada 15

16. Wholesale and Retail Services Quality adjusting goods or service portion? 15/09/2015 Statistics Canada Statistique Canada 16 Output+Trade Margin=Final Consumption V10050150 P10 (PPI)5 (15-10) (Margin Price) 15 (CPI) Q10 Example:  V = value, P= price, Q = quantity, V = P x Q  10 widgets produced by manufacturer and sold to a consumer by a wholesaler/retailer  Produce price of widgets = $10  Final purchase price of widgets = $15  Example shows that margins are additive in terms of value and price  The margins and margins prices are mark-ups on the output and output prices. They exclude the value of the good.  Quality adjustment for the margin index should exclude any adjustment for good because the good is netted out of the margin calculation.

17. Recommendations from Geneva  Interactions between various international groups are seen as positive as they encourage dialogue and different perspectives “It was found useful to exchange experiences on the approaches used for SPPI and CPI. One example of an area where exchange of views would be useful is the different perspective on quality adjustment in PPIs based on production functions and technology and CPIs based on utility functions.”  Very little specific feedback was received from the Expert Group on CPI. Possible reasons include: Complex topic Short discussion Paper and presentation not clear 15/09/2015 Statistics Canada Statistique Canada 17

18. Way forward for VG  Further work on this topic is required  Dichotomy between producer view and consumer view  Despite the arguments put forward here, for some services, intuitively it seems reasonable to adjust for quality based on consumer utility rather than the production function. 15/09/2015 Statistics Canada Statistique Canada 18

19. Way forward for VG 15/09/2015 Statistics Canada Statistique Canada 19 Service (ISIC)Disposition of use (consumer expenditure, intermediate use) Approach (consumer utility or production function) Method  In deciding which approach is best, we need to consider how the service is consumed and by whom  Basic proposal If consumption of output mainly by final consumer, adopt utility approach for QA If consumption of output mainly intermediate use, adopt production function view Incorporate the analysis and findings in sector paper to eventually cover all ISIC industries under VG mandate

20. 15/09/2015 Statistics Canada Statistique Canada 20 Discussion