## cr17302

## Source details

**Canonical URL:** [cr17302](https://www.imf.org/-/media/files/publications/cr/2017/cr17302.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2017/cr17302.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2017/cr17302.pdf.json)

---

### Introduction and mission objectives
- TA mission to Kampala during January 11–29, 2016 to provide advice and training on development and compilation of export-import price indices (XMPI), and review rebasing work on the CIPI and the H&R-PPI.
- New XMPI design: hybrid indices combining quarterly Customs unit values for relatively homogeneous commodities and survey-based specification prices collected in price surveys of importers (and then exporters) for more heterogeneous commodities.
- Explicit mission objectives:
  - Review and finalize the structure and weighting pattern of the XMPI and the H&R-PPI.
  - Review progress on determining optimum data sources for weights and prices for the CIPI.
  - Analyze all survey-based prices collected to date, including adequacy of product specifications and collection strategy.
  - Discuss frequency of weight updates (including the CPI basket).
  - Review index compilation systems and procedures.
  - Prepare dissemination strategies and documentation.
  - Provide additional training to staff.

### Export and Import Price Indices — Statistics prerequisites and design
- XMPI design and implementation:
  - Hybrid approach: Customs unit values for homogeneous items; survey-based prices from XMPS for heterogeneous items.
  - UBOS commenced development of the MPI including pre-testing a survey initialization interview program to collect quarterly prices.
  - Pre-test sample: 40 importers; 30 of 40 provided product specifications and prices for March, June, and September quarters of 2015; 10 refusals.
  - Retrospective quarterly prices to March quarter 2015 will be obtained during initialization interviews to supplement Customs unit values; retrospective survey prices beyond about three or four quarters generally not feasible.
  - Documentation focuses on MPI development; same techniques apply to XPI.

- Index reference bases and periodicity:
  - Reference base period decided for XMPI: 2009/10 to be consistent with other economic indices.
  - EA indices at the 8-digit level of the HS.
  - Planned compilation: 2009/10 based XMPI compiled quarterly from September 2009, and annually from 2009/10.
  - Publication goal: publish indices at the 2-digit HS Chapter level together with total indices; annual and quarterly indices down to the 8-digit HS commodity level to be provided to National Accounts as needed.
  - Rebase frequency: fully rebased at least each five years; annual monitoring recommended.

### XMPI — Accuracy and reliability: methodology, weights, and source data
- Pricing points and formulas:
  - XPI pricing point: Free on Board (FOB).
  - MPI pricing point: Cost Insurance and Freight (CIF).
  - Unweighted EA indices: Jevons price index formula (geometric mean).
  - Aggregation: EA indices aggregated with fixed weights using the Laspeyres formula.
- Weighting base and basket:
  - Weighting base: 3-year average for 2011 to 2013 used to derive MPI structure, composition and weighting pattern.
  - 151 8-digit HS commodities selected as the index basket for MPI.
- Source data, unit values and survey split:
  - Analysis over 2011–2013 of quarterly unit values for the 151 selected HS commodities; smoothing of outliers applied where needed.
  - Findings: 45 of the 151 8-digit HS commodities had credible Customs-based price indicators; 106 of the 151 required survey-based price indicators.
  - Although 106 of 151 HS 8-digit commodities (over 70 percent) will have survey-based prices, only about 40 percent of the value of the Chapters will be based on survey data.
- HS Chapter weights to All Groups and primary data sources (identified):
  - Petroleum products 28% — Customs
  - Machinery, computers, etc. 11% — Survey
  - Vehicles 11% — Survey
  - Electrical machinery 10% — Survey
  - Iron and steel 5% — Customs and Survey
  - Animal fats and oils 5% — Customs
  - Cereals 5% — Customs
  - Plastics 5% — Customs and Survey
  - Sugar 3% — Customs
  - Cement and lime 3% — Customs
- Back-price construction and compilation sequencing:
  - For periods prior to March quarter 2015, back prices need to be based on Customs unit values for all selected Chapters and commodities.
  - Initial MPI to be compiled retrospectively to September quarter 2009 based solely on adjusted Customs unit value data; when survey data for the 106 commodities becomes available, recompile incorporating survey data from March quarter 2015.
- Imputation and sampling:
  - Imputation methods discussed: carry-forward, using movements in related products' prices, or projecting recent trend.
  - Initial MPS sample: about 500 importers (excessive); reduced to 342 importers; over-sampling intended because of expected refusal rates (pre-test: 10 of 40 refused).
  - Annual review of importers sample recommended to add new importers and remove those who have contracted.

### XMPI — Serviceability and operational needs
- Compilation and dissemination:
  - Goal to publish indices at the 2-digit HS Chapter level and total indices, subject to data analysis.
  - Annual monitoring and rebase every five years, with review for earlier rebasing if significant changes occur.
- Operational needs:
  - Complete development of fully operational linked spreadsheet compilation system for initial and ongoing quarterly MPI and XPI compilation.
  - Refresher training of field interviewers prior to initialization interviews (scheduled 15 March to 15 May 2016).

### Producer Price Index — Manufacturing (PPI-M): prerequisites and scope
- Short-term UBOS priority: complete rebasing of the PPI-M.
- Sample and field work:
  - Sample selected: 173 manufacturing establishments using cut-off sampling techniques (June–July 2014).
  - Initialization interviews and ongoing monthly price collections commenced for 81 of those establishments, aiming to commence with prices back to September quarter 2013 (time reference period for the new index).
  - Coverage of the 81 establishments: provide coverage of some 70 percent of manufacturing gross output, though biased towards large establishments.
  - Cooperation issues: major problems obtaining cooperation from many manufacturers, especially small and medium sized establishments; 92 additional establishments in original sample remain to be interviewed (173–81).
- Reference base, scope, and pricing point:
  - Time reference period: September quarter 2013; new PPS commencing with prices for the month of July 2013.
  - Scope extended to actual transaction prices for gross output of manufacturing and utilities (electricity and water); medium term extension planned to mining and quarrying.
  - Pricing point: ex-factory with valuation basis of basic prices, exclusive of trade and transport margins and VAT.
  - Separate prices to be collected/published for exports and domestic usage; separate structures and weights for local sales, exports and total output.

### PPI-M — Methodology, aggregation, and weighting
- Aggregation and formulas:
  - First level aggregation: 4-digit ISIC class level.
  - EA indices: Jevons price index formula.
  - Higher-level aggregation: Laspeyres formula using derived weighting system.
- Weighting base and rebasing frequency:
  - Weighting base period: 2009/10 (reference period for the most recent SUT).
  - Proposed full rebasing on a 5-yearly basis using the latest finalized SUT.
  - Note: EAC recommendation for member countries to move towards base period 2015; CPI rebase periodicity moving to a 3-yearly cycle.
- Determination of basket and sample:
  - Final 2009/10 index weights calculated for total domestic supply (gross output) based on final 2009/10 SUT.
  - Twenty eight 4-digit ISIC class industries selected as the index “basket.”
  - Sample of establishments selected to represent each of the 28 classes; business rules applied to select establishments across large, medium and small classes (see sampling rules below).

### PPI-M — Source data, sampling, and data quality concerns
- PPS periodicity and sample:
  - Mid-month prices are collected monthly (monthly periodicity and frequency).
  - Regimen of 28 classes categorized into:
    - 6 large classes (5 percent or more of the total weight)
    - 16 medium classes (between 1 and 5 percent)
    - 6 small classes (less than 1 percent)
  - Business rule sampling targets: 90 percent applied to large industries, 70 percent to medium industries supplemented as necessary, and 70 percent applied to small industries.
- Sampling status and targets:
  - Current sample: 173 establishments selected; 81 establishments providing monthly data; mission recommends conducting initialization interviews with remaining 92 establishments.
  - Target: publish rebased PPI-M with available data and progressively update as additional establishments provide data; target release date: June 30, 2016.
- Data credibility issues identified during initialization:
  - (i) incorporation of V.A.T. when pricing basis is basic prices;
  - (ii) need to impute missing prices;
  - (iii) market designation (exports vs local sales) sometimes missing;
  - (iv) monthly prices sometimes constant for two or three years;
  - (v) very large price rises and falls (in excess of 100 percent);
  - (vi) highly volatile price series;
  - (vii) apparent data entry errors;
  - (viii) excessive number of product specifications for some establishments (e.g., over 20).
  - These queries highlighted on spreadsheets require resolution.
- Processing system:
  - Current PPI processing system to be used as foundation; processing must be transferred from Excel to the “Access” system (manages data capture and product and establishment samples).
- Dissemination and linkage:
  - Aim to disseminate the new index on a very timely basis (ideally prior to CPI) to serve as a forward-looking indicator for inflation analysis and to serve National Accounts needs.
  - Annual indices calculated as the simple average of monthly indices.
  - Rebased indices should be linked to the current index in total and at a broad industry level; ultimately coordinate PPI, CPI and national accounts on a common reference base period.

### Hotels and Restaurants PPI — prerequisites, methodology and implementation
- Priorities and staffing:
  - Short-term UBOS priority: complete rebasing of H&R-PPI.
  - Limited progress due to shifting priorities and staff changes; Business and Industry Directorate positioned to move project forward during 2016.
  - Work plan needs preparation; staff and financial resources need allocation to complete structure, weights, compilation system, PPS redesign, sample selection, initialization interviews, and ongoing price collection.
- Conceptual basis and pricing:
  - Concept ratified: relate to actual transaction prices for sales of Accommodation and Food and Beverage Service Activities.
  - Pricing point: ex hotel/restaurant; pricing basis: basic prices exclusive of VAT and after discounts; include subsidies.
  - Because of seasonal/irregular nature, base period prices will be “normalized” by taking a 12 month average; reference base period proposed: 2015.
  - Proposed retrospective monthly prices back to January 2015 during initialization interviews; compile monthly from January 2015.
- Aggregation and weights:
  - Processing system based on spreadsheets developed for rebased PPI-M.
  - EA indices: Jevons price index formula; aggregation upward using Laspeyres formula and derived weighting system.
  - Weighting base period: 2009/10 SUT table data for Accommodation (ISIC Division 55) and Food and Beverage Service Activities (ISIC Division 56).
  - 4-digit ISIC class weights derived using UBI gross output data.
  - Rebase cycle: 5-yearly.

- Proposed H&R-PPI weight breakdown (proposed calculations to be verified):
  - Accommodation (Div. 55)      29.3%
  - Short-term Accommodation Activities (Class 5510)   29.3%
  - Food and Beverage Service Activities   70.7%
  - Restaurants etc. (Class 5610)    41.1%
  - Event Catering (Class 5621)    11.0%
  - Beverage Service Activities (Class 5630)  18.6%

- Source data, sample and collection:
  - Current H&R-PPS collects mid-month price data quarterly along with sales and other data; proposal to redesign and simplify PPS to collect only price data and move to monthly collection.
  - New sample of hotels and restaurants to be selected based on 2009/10 UBI gross output by ISIC class using scientific sampling procedures (as for PPI-M).
  - Program of initialization interviews required to commence monthly collection; UBOS to meet with Ugandan Hotel Owners Association (UHOA) to obtain support and determine data requirements.
- Publication goals:
  - Plan to publish monthly indices at the 4-digit ISIC class level.
  - Link to old indices at ISIC Division and Total level to provide long term series.
  - Target publication and linkage of old and new series: Oct. 7, 2016.

### Training, work programs, and operational timetables
- Training delivered: price index concepts, compilation methodology including splicing and linking techniques, and price survey procedures; refresher training of field staff planned (e.g., Feb.15 refresher for XMPI).
- Work programs drafted and agreed for:
  - Development of XMPI.
  - Completion of PPI-M rebase.
  - Completion of H&R-PPI rebase.
- Operational priorities:
  - Complete development of fully operational linked spreadsheet compilation system for initial and ongoing quarterly MPI and XPI compilation.
  - Transfer PPI-M processing from Excel to Access.
  - Refresher training of field interviewers prior to initialization interviews.

- Key project timetables and milestones (target date: 2016) — selected items from appendices:
  - XMPI Project (Appendix I) key dates:
    - Train field staff (refresher) Feb.15
    - Finalize spreadsheet processing system Mar. 31
    - Compile unit value-based MPI to March 2015 Apr. 15
    - Intensive field interviews Mar.15–May 15; complete by May 15
    - Replace unit values from March 2015 with survey prices and link by May 30
    - Recompile and publish MPI Jun. 30
    - Ongoing quarterly compilation Ongoing
  - PPI-M Project (Appendix II) key dates:
    - Finalize sampling methodology Jan. 31
    - Conduct interviews with 92 additional manufacturers aiming for back prices to July 2013 by Mar. 31
    - Move processing from Excel to Access May 31
    - Publish rebased PPI-M Jun. 30
  - H&R-PPI Project (Appendix III) key dates:
    - Finalize index structure and weights Feb. 5
    - Design H&R-PPS incl. price form Feb. 26
    - Obtain UHOA support Apr. 8; refresher training Apr. 8
    - Conduct initialization interviews May 15
    - Link old and new series and publish new linked indices Oct. 7

### Sample review, product selection principles, and ongoing maintenance
- Annual review of the sample of establishments recommended to introduce new or expanding businesses and remove those that have contracted.
- Price collection initialization/review:
  - Interviews to initiate new collections or review/update products priced in existing surveys.
- Product selection principles for monthly pricing:
  - (i) sampled products need to be representative of wider range of products in terms of price change over time;
  - (ii) usually establishments’ largest product sales by value;
  - (iii) manageable number of products per commodity for workload reasons;
  - (v) full specification of all price-determining characteristics, unit of quantity, etc., required to price to constant quality over time.
- Periodicity and timeliness for H&R-PPI:
  - Proposed rebased H&R-PPI have reference base of 2015 and be compiled monthly from January 2015.

*IMF staff mission report: Executive Summary and selected excerpts (cr17302).*

### Executive Summary ......................................................................................................

### Executive Summary

### Introduction
- TA mission to Kampala during January 11–29, 2016 to provide advice and training on development and compilation of export-import price indices (XMPI), and review rebasing work on the CIPI and the H&R-PPI.
- New XMPI will be hybrid indices combining quarterly Customs unit values for relatively homogeneous commodities and survey-based specification prices collected in price surveys of importers (and then exporters) for more heterogeneous commodities.
- Short-term UBOS priorities: complete rebasing of PPI-M and H&R-PPI; rebasing of CIPI has lower priority.
- For the H&R-PPI, the conceptual basis was reviewed and confirmed; the reference base period will be 2015.
- Training was provided on price index concepts and compilation methodology (including splicing and linking techniques), and price survey procedures.

### Mission objectives (explicit)
- Review and finalize the structure and weighting pattern of the XMPI and the H&R-PPI.
- Review progress on determining optimum data sources for weights and prices for the construction inputs price index (CIPI).
- Analyze all survey-based prices collected to date, including adequacy of product specifications and the collection strategy.
- Discuss frequency of weight updates (including the CPI basket).
- Review index compilation systems and procedures.
- Prepare dissemination strategies and documentation.
- Provide additional training to staff.

### Export and Import Price Indices — I. Statistics Prerequisites
- XMPI design: hybrid indices using Customs unit values for homogeneous items and survey-based prices from XMPS for heterogeneous items.
- UBOS commenced development of the MPI including pre-testing a survey initialization interview program to collect quarterly prices.
- Pre-test: sample of 40 importers interviewed to test strategy and respondent cooperation; 30 of 40 provided product specifications and prices for March, June, and September quarters of 2015; 10 refusals.
- Collected pre-test data loaded onto Excel spreadsheet compilation system; product specifications and prices were analyzed and, with few exceptions, assessed as sound.
- Work required: complete fully operational linked spreadsheet system to support initial and ongoing quarterly compilation of MPI and XPI.
- Retrospective quarterly prices to March quarter 2015 will be obtained during initialization interviews to supplement Customs unit values; not feasible to obtain retrospective prices over a longer period.
- XMPI documentation focuses on MPI development; same techniques will apply to XPI, which is much less complex.

### Export and Import Price Indices — II. Accuracy and Reliability
A. Statistical Techniques — Conceptual basis and methodology
- XMPI will relate to actual transaction prices for exports and imports of merchandise trade.
- Pricing points: XPI at Free on Board (FOB); MPI at Cost Insurance and Freight (CIF).
- Reference base period decided: 2009/10 to be consistent with other economic indices and to provide a reasonable time series.
- For commodities requiring survey prices, retrospective indices will be compiled from:
  - back prices obtained during initialization interviews (to March quarter 2015), and
  - unit values for the period from March quarter 2015 back to Sept. quarter 2009, adjusted to remove outliers and modified in light of survey price performance.
- EA indices will be at the 8-digit level of the HS.
- Unweighted indices: Jevons price index formula (geometric mean of price indicators).
- Aggregation: EA indices aggregated with fixed weights using the Laspeyres formula.
- Weighting base: 3-year average for 2011 to 2013 used to derive MPI structure, composition and weighting pattern.
- 151 8-digit HS commodities were selected as the index basket for MPI; UBOS to repeat process for XPI.
- MPI structure and weighting pattern reviewed and assessed as representative of current trading patterns.

B. Source Data — Price indicators, sampling, and imputation
- Price indicators required for each selected 8-digit import commodity:
  - Customs unit values used for relatively homogeneous items where unit value variations credibly measure pure price change.
  - Quarterly MPS (survey prices) needed for non-homogeneous commodities with non-credible unit value patterns.
- Analysis improved unit value quality by eliminating country mix from 8-digit unit values, i.e., using individual country unit values or removing outlier countries.
- Quarterly unit values for each of the 151 selected HS commodities analyzed over 2011 to 2013 to assess suitability; smoothing of outliers applied where needed.
- Assessment of individual country-of-origin unit values sometimes rendered Customs data suitable, reducing dependence on survey data.
- Findings from analysis:
  - 45 of the 151 8-digit HS commodities had credible Customs-based price indicators.
  - 106 of the 151 required survey-based price indicators.
- Trend analysis performed back to September quarter 2009 for the 45 commodities selected for unit value-based indicators; further smoothing or removal of country data undertaken and needs final review.
- Initial MPI to be compiled retrospectively to September quarter 2009 based solely on adjusted Customs unit value data; when survey data for the 106 commodities becomes available, recompile incorporating survey data from March quarter 2015.
- Pre-testing indicates retrospective survey prices generally not feasible beyond about three or four quarters.
- HS Chapter weights to All Groups and primary data sources for price indicators (as identified):
  - Petroleum products 28% — Customs
  - Machinery, computers, etc. 11% — Survey
  - Vehicles 11% — Survey
  - Electrical machinery 10% — Survey
  - Iron and steel 5% — Customs and Survey
  - Animal fats and oils 5% — Customs
  - Cereals 5% — Customs
  - Plastics 5% — Customs and Survey
  - Sugar 3% — Customs
  - Cement and lime 3% — Customs
- Observations:
  - Although 106 of 151 HS 8-digit commodities (over 70 percent) will have survey-based prices, only about 40 percent of the value of the Chapters will be based on survey data.
- Back-price construction: for periods prior to March quarter 2015, back prices need to be based on Customs unit values for all selected Chapters and commodities; cannot rely only on Customs data for Chapters whose future primary data source will be Customs data because they may not be representative of total exports.
- Imputation methods discussed for missing observations (survey prices or Customs unit values), e.g., carry-forward, using movements in related products' prices, or projecting recent trend.
- Sampling for MPS:
  - Initial selection resulted in about 500 importers which was excessive and needed reduction by at least 50 percent.
  - UBOS reduced initial sample to 342 importers; decision to over-sample to have substitutes because of expected high refusal rates (10 of 40 refused in pre-test).
  - Annual review of the sample of importers recommended to add new importers and remove those who have contracted.
- Price collection initialization/review:
  - Principles for selection of specific products for quarterly price collection outlined in previous mission report.
  - UBOS pre-test used a quarterly collection form designed on the Tanzanian model.
  - Of 40 importers in pre-test, 10 refused; product specifications and back prices for three quarters were loaded to the spreadsheet system.
  - Means to improve response rates and business cooperation discussed; UBOS plans to approach the Ugandan Manufacturers’ Association for PPI-M support.
  - Analysis of product specifications and prices concluded:
    - Majority of product descriptions adequate to indicate pure price change; unit of quantity and country of origin generally specified; a small number identified as inadequate.
    - Most reported prices were credible indicators of price change; small number volatile requiring smoothing or querying; some imputation of missing prices required.
    - A few importers had very lumpy import patterns; consideration should be given to replacing them.
  - Refresher training of field interviewing staff planned prior to initialization interviews scheduled 15 March to 15 May 2016.

### Export and Import Price Indices — III. Serviceability
A. Periodicity and Timeliness
- Planned compilation: 2009/10 based XMPI compiled quarterly from September 2009, and annually from 2009/10.
- Publication goals and dissemination policy:
  - Goal to publish indices at the 2-digit HS Chapter level together with total indices, subject to data analysis.
  - Annual and quarterly indices down to the 8-digit HS commodity level to be provided to National Accounts on an as-needs basis, with any necessary additional comments.
- Rebase frequency:
  - Indices should be fully rebased at least each five years.
  - Annual monitoring of import and export patterns recommended to assess whether significant changes warrant earlier rebasing.

### Producer Price Index — manufacturing (PPI-M)
- Short-term UBOS priority: complete rebasing of the PPI for manufacturing (PPI-M).
- Resource cuts had delayed the process; new corresponding release dates set:
  - PPI-M rebase release date: June 30, 2016.
- Work program drafted and agreed for completion of PPI-M rebase.
- Further training provided on price index concepts, compilation methodology (including splicing and linking techniques), and price survey procedures.

### Hotels and Restaurants Producer Price Index (H&R-PPI)
- Short-term UBOS priority: complete rebasing of H&R-PPI.
- H&R-PPI conceptual basis reviewed and confirmed.
- New index structure and weighting pattern derived based on data from the 2009 Supply and Use Tables (SUT) and Uganda Business Inquiry (UBI) gross output.
- Reference base period for H&R-PPI: 2015.
- Resource cuts delayed process; new corresponding release date for H&R-PPI rebase: October 7, 2016.
- Rebase of the CIPI rescheduled to late 2016 (lower priority compared with PPI-M and H&R-PPI).
- Work program drafted and agreed for completion of H&R-PPI rebase.

### Work programs, training, and systems
- Work programs drafted and agreed for:
  - Development of XMPI.
  - Completion of the PPI-M rebase.
  - Completion of the H&R-PPI rebase.
- Further training delivered on price index concepts and compilation methodology (including splicing and linking techniques) and price survey procedures.
- Operational needs:
  - Complete development of fully operational linked spreadsheet compilation system for initial and ongoing quarterly MPI and XPI compilation.
  - Refresher training of field interviewers prior to initialization interviews.

*IMF staff mission report: Executive Summary (cr17302).*

### 48.      Training on price index concepts and methodology and price collection initialization

### 48.      Training on price index concepts and methodology and price collection initialization

### Work program
- A work program for the development of the indices is provided in Appendix I.
- During the current mission, survey methodology, transaction pricing, product sampling and index aggregation processes were discussed, including:
  - derivation of EA indices based on both survey prices and unit values;
  - application of base weights to calculate indices at higher levels of aggregation.

### PRODUCER PRICE INDEX MANUFACTURING (PPI-M) — I. STATISTICS PREREQUISITES
- A sample of 173 manufacturing establishments was selected for PPI-M using cut-off sampling techniques (June–July 2014).
- UBOS Statistical Assistant field staff have been conducting initialization interviews and ongoing monthly price collections have commenced for 81 of those establishments, as far as possible commencing with prices back to September quarter 2013 (the time reference period for the new index).
- UBOS has experienced major problems in obtaining cooperation from many manufacturers, especially small and medium sized establishments.
- To progress the development of the index, a concerted effort will need to be undertaken to increase the sample size (see paras. 86–92).

### PPI-M — II. ACCURACY AND RELIABILITY — A. Statistical Techniques (Conceptual basis)
- Concepts and definitions for the rebased PPI were reviewed and ratified; they are consistent with the 2008 SNA and the international PPI Manual.
- Key elements:
  - (i) Scope extended to actual transaction prices for the gross output of the manufacturing and utility (i.e., electricity and water) industries. In the medium term, planned extension to cover output of the mining and quarrying sector.
  - (ii) Pricing point is ex-factory etc., with a valuation basis of basic prices, exclusive of trade and transport margins and VAT.
  - (iii) Scope relates to both domestic usage and exports; separate prices will need to be collected for exports and domestic usage as appropriate. Separate indices should be published for local sales and exports as well as total output (see paragraphs 78–85).
- Reference base period:
  - Time reference period for the new index will be September quarter 2013 and the new PPS is commencing with prices for the month of July 2013.
- Aggregation formula and compilation methodology:
  - First level of aggregation will be to the 4-digit ISIC class level.
  - EA indices will be calculated using the Jevons price index formula (ratio of the unweighted geometric mean prices).
  - EA indices will be progressively combined to higher levels (3-digit, 2-digit, 1-digit, all groups) using the Laspeyres formula and the derived weighting system.
  - Current PPI processing system used as foundation; processing must be transferred from Excel to the “Access” system (manages data capture and product and establishment samples).
- Weighting base period and frequency of rebasing:
  - Weighting base period is 2009/10 (reference period for the most recent SUT).
  - Proposed full rebasing on a 5-yearly basis, using the latest finalized SUT.
  - The East Africa Community (EAC) has recommended member countries move towards a common base period of 2015.
  - CPI periodic rebase moving to a 3-yearly cycle; PPI-M has commenced being rebased on a 5-yearly cycle using results of the 5-yearly SUT and the UBI.
- Determination of structure, composition and weighting pattern:
  - Final 2009/10 index weights were calculated for total domestic supply (gross output) based on the final 2009/10 SUT.
  - Twenty eight 4-digit ISIC class industries were selected as the index “basket.”
  - A sample of establishments to represent each of the 28 classes was selected.
  - Broad credibility checks applied to weights by reference to local knowledge and Ugandan Revenue Authority (URA) information on the “largest taxpayers.”
  - Current manufacturing PPI publication has separate tables with index numbers for (i) domestic sales (ii) exports and (iii) total output; separate structures and weighting patterns exist for domestic usage, exports and total output.
  - Bank of Uganda (BOU) requires the three separate tables on an ongoing basis; therefore separate structures and weighting patterns will be developed for exports and domestic usage.
  - Calculation of separate exports and domestic sales structures: decided to use a macro approach linking HS export commodity codes (HSCC) to the SUT 4-digit ISIC industry classes and match 2009/10 data; local sales generally derived as industry output minus exports, with adjustments to remove double counting of local sales that are later exported.

### PPI-M — II. ACCURACY AND RELIABILITY — B. Source Data (Price indicators)
- The current PPS has been greatly simplified; a new PPS operates with a separate collection form relating solely to PPI mid-month product prices.
- Mid-month prices are now collected monthly rather than quarterly; both periodicity and frequency of the PPS are monthly.
- Sampling:
  - Sampling criteria were reviewed and a fresh sample selected using a combination of cut-off and random sampling techniques.
  - Regimen of 28 classes categorized into 6 large classes (5 percent or more of the total weight), 16 medium classes (between 1 and 5 percent) and 6 small classes (less than 1 percent).
  - Business rule: 90 percent applied to large industries, 70 percent to medium industries supplemented by selection of smaller establishments using a skip interval approach as necessary, and 70 percent applied to small industries.
  - Resulting sample of 173 manufacturing establishments (including supplementation by URA “largest taxpayers” and local knowledge).
- To date, ongoing monthly price collections established with 81 establishments (mainly large); the 81 establishments provide coverage of some 70 percent of manufacturing gross output, though biased towards large establishments.
- 2013 Ugandan Auditor General audit recommended scientific sampling to measure sampling errors. Mission is confident of representativeness of cut-off and random based sample used in other East African countries; unless UBOS commits to scientific sampling for the rebased PPI, work should proceed to conduct initialization interviews with the remaining 92 establishments in the original sample (i.e., 173–81).
- Compilation should proceed as soon as possible using price data from the new sample of 81 establishments, progressively updated with incorporation of prices from additional establishments as initialization proceeds.
- Even if an establishment only provides retrospective prices for part of the period back to July 2013, earlier prices can be incorporated using splicing techniques and revising the indices.
- UBOS will contact the Ugandan Manufacturers’ Association (UMA) to enlist support and encourage member cooperation; UBOS will support the interviewing team and optimize resources.
- Target: publish the rebased PPI-M within a reasonable time frame; target date of 30 June 2016 has been set for release of the rebased index, initially incorporating price data from as many as possible of the additional 92 establishments. After initial publication, work should continue to enroll additional establishments, including substitutes if necessary, and progressively revise the indices as sample size improves.
- An annual review of the sample of establishments should be undertaken to introduce new businesses or expanding ones and remove establishments that have contracted (can be undertaken by reference to the Uganda Business Register).
- Mission assessed product specification descriptions obtained during initialization interviews and was satisfied they would support pricing to constant quality.
- Credibility analysis of the time series of prices resulted in many queries; main issues include:
  - (i) incorporation of V.A.T. when the pricing basis is basic prices;
  - (ii) need to impute missing prices;
  - (iii) designation of the market (i.e., exports and local sales) is sometimes missing;
  - (iv) monthly prices sometimes constant for two or three years;
  - (v) very large price rises and falls (in excess of 100 percent);
  - (vi) highly volatile price series;
  - (vii) apparent data entry errors;
  - (viii) an excessive number of product specifications for some establishments (e.g., over 20).
- These queries have been highlighted on the spreadsheet and require resolution.

### PPI-M — III. SERVICEABILITY — A. Periodicity and Timeliness
- PPS periodicity and frequency: monthly, supporting UBOS goal of subscribing to IMF Special Data Dissemination Service (SDDS).
- Aim: disseminate the new index on a very timely basis (ideally prior to the release of the CPI) to serve as a forward-looking indicator for inflation analysis and to serve National Accounts needs.
- Annual indices will be calculated as the simple average of the monthly indices.
- During initialization interviews, retrospective monthly prices are being collected from July 2013, or as far back as practical.

### PPI-M — III. SERVICEABILITY — Publication goals and dissemination policy
- Separate indices will be published on an ongoing basis for total gross output, domestic sales, and exports, broken down to the 2-digit Division level of ISIC and higher levels of aggregation.
- If samples are sufficient for the earlier period, monthly data should be published back to July 2013; if samples are not sufficient, less detail should be released.
- Rebased indices should be linked as far as possible to the current index, at least in total and at a broad industry level.
- After a time series has been built up, the index should be re-referenced to a calendar year and ultimately the PPI, CPI and national accounts should be coordinated on a common reference base period.

### HOTELS AND RESTAURANTS PRODUCER PRICE INDEX — I. STATISTICS PREREQUISITES
- Limited progress due to shifting priorities and staff changes; Business and Industry Directorate is now in a position to move the project forward during 2016.
- A work plan needs to be prepared.
- Staff and financial resources need to be allocated to:
  - complete development of the rebased index structure and weighting pattern;
  - develop the compilation spreadsheet system;
  - redesign the H&R-Producer Price Survey (H&R-PPS);
  - select the sample of hotels and restaurants;
  - commence the new price survey through initialization interviews;
  - conduct ongoing price collection and index compilation.

### HOTELS AND RESTAURANTS PPI — II. ACCURACY AND RELIABILITY — A. Statistical Techniques (Conceptual basis)
- Conceptual basis ratified: relate to actual transaction prices for sales of Accommodation and Food and Beverage Service Activities.
- Pricing point: ex hotel/restaurant; pricing basis: basic prices exclusive of value added tax (VAT) and after any discounts, and include any subsidies.
- Reference base period:
  - Because of seasonal or irregular nature of many commodity prices, base period prices will be “normalized” by taking a 12 month average; reference base period will relate to a full year.
  - Proposed to obtain retrospective monthly prices back to January 2015 during initialization interviews and adopt reference period 2015.
- Aggregation formula and compilation methodology:
  - Processing system for H&R-PPI will be based on spreadsheets developed for compilation of rebased PPI-M.
  - First level of aggregation: combine individual product prices collected in H&R PPS to the 4-digit ISIC class level.
  - EA indices calculated using the Jevons price index formula; EA indices then progressively combined to higher levels and the total index using the Laspeyres formula and derived weighting system.
  - This approach differs from current index (which aggregates to a commodity level before weights); new approach is consistent with PPI-M and industry weights are readily available.
- Weighting base period and frequency of rebasing:
  - Index weights will be based on 2009/10 SUT table data for “Accommodation” (ISIC Division 55) and “Food and Beverage Service Activities” (ISIC Division 56).
  - 4-digit ISIC class weights will be derived using UBI gross output data.
  - The index should be rebased 5-yearly.
- Methodology for determining structure, composition and weighting pattern — proposed calculations:
  - Hotels and Restaurants PPI
    - Accommodation (Div. 55)      29.3%
    - Short-term Accommodation Activities (Class 5510)   29.3%
    - Food and Beverage Service Activities   70.7%
    - Restaurants etc. (Class 5610)    41.1%
    - Event Catering (Class 5621)    11.0%
    - Beverage Service Activities (Class 5630)  18.6%
  - These calculations should be verified.

### HOTELS AND RESTAURANTS PPI — II. ACCURACY AND RELIABILITY — B. Source data (Price indicators)
- Current H&R-PPS seeks mid-month price data for specific, representative products for a range of commodities.
- Collection is conducted quarterly and asks for mid-month price data as well as sales and other data.
- It is proposed that the PPS form be reviewed, redesigned and simplified; consideration should be given to collecting only price data to streamline collection and processing.
- A new sample of hotels and restaurants needs to be selected based on 2009/10 UBI data for establishment gross output by ISIC class, using scientific sampling procedures (as for PPI-M); work was initiated during the current mission.
- A program of initialization interviews needs to be conducted with establishments new to the collection to commence monthly collection of price data.
- UBOS will meet with the Ugandan Hotel Owners Association (UHOA) to obtain their support and that of their members and determine their data requirements.

*cr17302 - 48.      Training on price index concepts and methodology and price collection initialization*

### 109.     An annual review of the sample of establishments should be undertaken to enable the

### 109.     An annual review of the sample of establishments should be undertaken to enable the

### Sample review of establishments
- An annual review of the sample of establishments should be undertaken to enable the introduction of new businesses or existing ones that have expanded, and the removal of businesses that have contracted.

### Price collection initialization/review
- To establish direct price collections from the selected hotels and restaurants, interviews need to be conducted to either (i) initiate a new collection from establishments new to the price survey or (ii) review and update the sample of products priced from establishments already in the survey.

### Product selection principles for monthly pricing
- In selecting the sample of products for monthly pricing the following principles should be applied:
  - (i) the sampled products need to be representative of a wider range of products in terms of price change over time;
  - (ii) they will usually be the establishments’ largest product sales by value;
  - (iii) for workload reasons, a manageable number of products should be selected to represent a commodity; and
  - (v) in order to price to constant quality over time, and reflect pure price change, full specification of all the price-determining characteristics of the product, unit of quantity, etc. are required.

### Periodicity and timeliness
- It is proposed that the rebased H&R-PPI have a reference base of 2015 and be compiled monthly from January 2015.

### Publication goals and dissemination
- It is planned to publish monthly indices at the 4-digit ISIC class level in the index structure (see paras. 114–116 below).
- It should be linked to the old indices at the ISIC Division and Total level of aggregation to provide long term series.

### Rebase cycle and training
- The indices should be fully rebased each five years.
- Price collection staff need to be provided with refresher training on the index concepts and briefed on the index construction methodology.

### Work program
- A broad work program is provided in Appendix III.

### APPENDIX I: XMPI PROJECT TASKS AND TIMETABLE (Target date: 2016)
- 1. Train field staff (refresher) Feb.15
- 2. Finalize development of linked spreadsheet processing system for:
  - 2.1 Calculation of EA indices based on the Jevons formula
  - 2.2 Aggregation to higher levels using Laspeyres formula Mar. 31
- 3. Load all weights onto the system Mar. 31
- 4. Complete analysis/adjustment of U.V. series for all index commodities back to Sept. 2009 Mar. 31
- 5. Compile, and analyze, unit value-based MPI to March 2015 Apr. 15
- 6. Undertake intensive program of field interviews (Mar.15–May 15) May 15
- 7. Replace unit values from March 2015 with survey prices for 106 commodities:
  - 7.1 Convert to UGS
  - 7.2 Link survey prices to unit value series for EAs at Mar. 2015 May 30
- 8. Recompile and publish MPI Jun. 30
- 9. Ongoing quarterly compilation Ongoign
- 10. Commence progressively repeating above for XPI Jun. 30

### APPENDIX II: PPI-M PROJECT TASKS AND TIMETABLE (Target date: 2016)
- 1. Finalize management decision on sampling methodology Jan. 31
- 2. Conduct interview program with 92 additional manufacturers:
  - 1.1 Obtain UMA support
  - 1.2 Allocate very high priority
  - 1.3 Maximize staff resources (incl. re-retraining) and management support
  - 1.4 Select substitute establishments as necessary
  - 1.5 Aim for back prices to July 2013 or as far as possible Mar. 31
- 3. Split 2009/10 SUT ISIC class output between Local Sales and Exports Feb. 15
- 4. Compile separate Local sales, Exports and Total weighting patterns Feb. 15
- 5. Incorporate Utilities in (i) the index structure and (ii) the index compilation Feb. 15 Mar. 31
- 6. Rationalize major shifts between the old and new weighting structures Feb. 15
- 7. Resolve the queries raised in relation to the time series of product prices Feb. 15
- 8. Ongoing monthly compilation of rebased index and progressive introduction of new establishment data and revisions as necessary Ongoing
- 9. Prepare dissemination strategy and publication manuscript May 15
- 10. Move processing from Excel to Access system May 31
- 11. Link old/new indices May 31
- 12. Share draft indices internally for quality assurance May 31
- 13. Publish rebased PPI-M Jun. 30
- 14. Undertake 2015 PPI-M rebase (EAU)
- 15. Incorporate Mining and Quarrying industries

### APPENDIX III: H&R-PPI PROJECT TASKS AND TIMETABLE (Target Date: 2016)
- 1. Finalize development of index structure and weights Feb. 5
- 2. Select samples of establishments for each ISIC class Feb. 5
- 3. Design H&R-PPS, incl. price collection form Feb. 26
- 4. Obtain UHOA support Apr. 8
- 5. Conduct refresher training for field staff Apr. 8
- 6. Conduct initialization interviews May 15
- 7. Load all descriptors and weights onto spreadsheet system May 30
- 8. Ongoing price collection and index compilation from January 2015 Ongoing
- 9. Ongoing analysis of index Ongoing
- 10. Link old and new series and publish new linked indices Oct. 7
- 11. Ongoing price collection, index compilation and publication Ongoing

*IMF document cr17302 excerpt (selected paragraphs and appendices).*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17302.pdf_