## cr17301

## Source details

**Canonical URL:** [cr17301](https://www.imf.org/-/media/files/publications/cr/2017/cr17301.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2017/cr17301.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2017/cr17301.pdf.json)

---

### Executive Summary — Mission overview
- Technical assistance mission to Kampala during the period May 4–22, 2015 to provide advice on developing export and import price indices (XMPI) and rebasing the construction industry price index (CIPI) and the hotels and restaurants producer price index (H&R PPI).
- Mission objectives included:
  - Review of the import price index (MPI) commodity selection criteria and compilation methodology.
  - Development of the Import Price Survey (MPS) and Export Price Survey (XPS) methodology.
  - Development of methodology for integrating MPS and XPS data with Customs data.
  - Finalization of XMPI compilation procedures, including imputation of missing prices.
  - Determination of strategy for rebase of CIPI and H&R PPI.
  - Training of staff on above topics.
  - Preparation of documentation on methodological and computational procedures.
- Training provided on price index concepts and compilation methodology, and price survey procedures.

### XMPI — Design, implementation, and processing systems
- Conceptual design and reference periods:
  - XMPI will be hybrid indices combining quarterly Customs unit values for relatively homogeneous HS commodities and survey-based prices from the MPS for more heterogeneous commodities.
  - Pricing point: XPI f.o.b.; MPI c.i.f.
  - Reference base period selected: 2009 (noted also as 2009–10 in discussion to be consistent with other economic indices).
- Implementation status and schedule:
  - Limited progress due to Trade Unit staff involvement in the 2014 Population Census and resource constraints; program of initialization interviews for the MPS had not yet commenced.
  - MPS initialization interviews planned for mid-July to end-September 2015, including collection of retrospective quarterly prices back to June quarter 2013 to supplement Customs unit values.
  - Hybrid indices to be compiled as soon as practical on a reference base of 2009.
- Processing systems:
  - Basic Excel spreadsheet prototypes used for calculation of price relatives at the eight-digit HS level using the Jevons geometric mean formula and for progressive aggregation with fixed weights using the Laspeyres formula.
  - An operational system of linked spreadsheets needs to be developed based on the prototypes to support initial and ongoing quarterly compilation of the MPI and XPI.
- Basket, structure, and weights:
  - Structure and weighting pattern derived using smoothed import values over the years 2011–13; a three-year average for 2011–13 was judged the most credible distribution for weighting.
  - One hundred and fifty-one eight-digit HS commodities were selected as the MPI index basket.
  - Same top-down methodology and procedures to be repeated by UBOS for the XPI; a much smaller sample of HS commodities will be required for the XPI because of relative trading patterns.
- Time series construction for survey-priced commodities:
  - Retrospective indices for commodities requiring survey prices will be compiled using a mixture of:
    - Back prices obtained during initialization interviews (to June quarter 2013); and
    - Customs unit values for the period from June quarter 2013 back to September quarter 2009, adjusted to remove outliers and modified in light of survey price performance.
  - It is not considered feasible to obtain retrospective prices over a longer period.
- Rationale and design decisions:
  - Elementary aggregate (EA) indices will be at the eight-digit HS level; unweighted EA indices calculated using the Jevons formula (geometric mean of price indicators).
  - EA indices will be aggregated with fixed weights using the Laspeyres formula.
  - An unweighted EA formula selected because maintaining lower-level product and establishment weights continuously is not feasible.
  - A common reference period is required for XPI and MPI to facilitate calculation of the Terms of Trade.

### XMPI — Accuracy, commodity selection, and survey linkage
- Conceptual basis: XMPI relate to actual transaction prices for exports and imports of merchandise trade.
- Weighting base period and rebasing:
  - Weights derived from smoothed import values for 2011–13; updating to include 2014 was rejected due to problems with 2014 Customs data associated with change in definition of c.i.f.
- Experience with unit values and outcome for 151 HS commodities:
  - Experience indicates data quality from unit values is potentially improved if country mix is eliminated from the eight-digit unit values (i.e., use individual country unit values or remove outlier countries).
  - Quarterly unit values for each of the 151 selected HS commodities were analyzed over the three years 2011–13 to assess suitability as price indicators; smoothing of a few quarters’ outlier values in some series improved credibility.
  - For some initially unsuitable commodities, selecting unit values for major country/countries of origin or removing/smoothing outlier country data rendered Customs data suitable and reduced dependence on price survey data.
  - Outcome:
    - 45 were designated as having credible price indicators based on unit values.
    - 106 required survey-based price indicators.
  - For the 45 commodities with unit value-based indicators, trends were analyzed back to the March quarter 2009, with further smoothing or removal of particular country data undertaken; this work needs review and finalization.
  - For the 106 commodities requiring survey prices, retrospective indices will be compiled using:
    - back prices obtained during initialization interviews (back to June quarter 2013) and
    - unit values for the period from June quarter 2013 back to March quarter 2009, adjusted to remove outliers and smoothed and modified in light of survey price performance.
  - Linking unit value-based and survey-based EA commodity prices requires retrospective application of a factor equal to the ratio of the survey-based EA price to the unit value-based EA price for an overlap quarter.
- Imputation and missing observations:
  - Alternative imputation methods discussed include carry-forward, using movements in prices of related products from same or other importers, or projecting the recent trend.

### MPS sample selection, initialization, and ongoing procedures
- Sample selection rules for MPS (for each of the 106 HS commodities requiring survey prices):
  - i. at least two importers are required for each commodity, if possible;
  - ii. ideally significant coverage (in value terms) should be achieved (i.e., greater than 50 percent if feasible, particularly for large commodities);
  - iii. if very low concentration (i.e., dominated by a large number of small importers), select the largest importer then apply a skip interval to select a small, representative sample of small importers;
  - iv. priority should be given to achieving good coverage and representation of the high value commodities; and
  - v. for relatively small commodities with low concentration, delete the eight-digit item from the regimen and pro-rate its weight amongst the remaining eight-digit items in the four-digit heading.
- Initial Trade Unit staff selection resulted in about 500 importers; this is excessive and needs rigorous review and reduction by at least 50 percent, preferably to less than 200 importers.
- Review of initial sample selection should include:
  - removing importers where there is a large number for an HS commodity;
  - aiming for lower coverage of the medium and small HS commodities; and
  - eliminating small HS commodities that have no dominant importers and reallocating their weight on a prorated basis to other commodities in the same four-digit HS heading.
- An annual review of the sample of importers should be undertaken to introduce new importers or existing ones that have expanded, and remove importers who have contracted.
- Principles for selecting products for quarterly price collection:
  - i. sampled products should be representative of a wider range of products in terms of price change over time;
  - ii. they will usually be the importers’ largest imports by value;
  - iii. if possible, they should be regularly imported;
  - iv. for workload reasons, a manageable number of products (say, restricted to three or four if possible) should be selected to represent an eight-digit HS item from each importer;
  - v. full specification of all price-determining characteristics, unit of quantity and transaction details such as country of origin and currency of the transaction are required to price to constant quality; prices are to be converted in the office to Ugandan shillings (U Sh) for index compilation.
- Supporting documentation (to be tailored for Ugandan conditions) includes MPS Field Interviewers’ Instructions, Introductory Letter to Importers, MPS Interview Guide, and an Example of a Generic MPS Questionnaire (Appendix V).
- Main steps for initialization and ongoing collection:
  - contacting each importer to arrange an interview;
  - conducting interviews and selecting some three or four detailed representative product specifications per eight-digit HS code suitable for repeat pricing;
  - undertaking follow-up action, including re-interviews as necessary;
  - creating a tailored collection form for each importer;
  - on an ongoing quarterly basis: delivering tailored collection forms, undertaking collection control, and editing and querying prices;
  - regularly confirming and updating the representativeness of the sample of products with each importer;
  - loading the prices, converting to Ugandan shillings, and compiling the indices.
- Training provided covered price index concepts and methodology, survey methodology, transaction pricing, product sampling and index aggregation, derivation of EA indices based on both survey prices and unit values, and application of base weights to calculate indices at higher levels of aggregation.
- A work program for development of the indices is provided in Appendix I; completion of the 2014 Population Census is expected to improve availability of funding and staff resources to progress the XMPI project.

### XMPI serviceability, periodicity, publication, and rebasing
- Planned compilation and publication schedule:
  - The 2009–10 based XMPI is planned to be compiled quarterly from September 2009, and annually from 2009–10.
- Publication goals:
  - Publish indices at the two-digit HS Chapter level together with total indices, subject to analysis of the data.
  - Annual and quarterly indices down to the eight-digit HS commodity level should be provided to national accounts on an as-needed basis, with any necessary additional comments about the data.
- Rebase policy:
  - Indices should be fully rebased at least every five years. Each year, monitor import/export patterns to assess whether significant changes warrant earlier rebasing.

### Construction Industry Price Index (CIPI) — prerequisites, methodology, and recommendations
- Rebase prerequisites and consultation:
  - Extensive consultation with key stakeholders required to determine primary purpose of the index and to formulate its conceptual basis, including scope.
  - Scope of CIPI needs review; scope of Civil Works should be extended to include additional activities, such as the generation and distribution of electricity and gas.
- Data source identification and source issues:
  - Identify most appropriate data sources for price indicators for materials, labor, and leasing charges.
  - Determine optimum data sources for derivation of representative industry and commodity weights.
  - Determine whether producer price survey (PPS) prices (ex-factory) at basic prices are adequate indicators for input prices at purchasers’ prices required for the CIPI.
  - Analyze whether PPS prices relate to sales to construction companies versus wholesalers; consider adding specifications to PPS forms (e.g., prices of sales to construction companies) or collecting prices from building materials wholesalers to reflect VAT, freight, and wholesale margins.
  - Equipment leasing charges currently collected from construction companies; consider collecting from equipment owners.
- Aggregation and compilation methodology:
  - Basic heading indices (monthly) compiled using price data from a range of sources; EAs compiled using the Jevons formula from PPI, CPI, and External Trade Statistics.
  - Prices of bitumen and diesel collected directly from oil companies; wages and equipment hire charges from construction companies.
  - Basic heading indices aggregated to composite indices (e.g., building and civil works) using the modified Laspeyres formula.
- Past weighting methodology and current structure:
  - Upper level weights for present CIPI were calculated in 2007 using data from Ministry of Works and Transport and Road Agency Formation Unit investment/expenditure records, Kampala City Council building approval records, Directorate of Water Development and National Water and Sewerage Corporation records, and UBOS national accounts.
  - Table 1. Current CIPI Structure and Weighting Pattern:
    - Whole Sector (100%)
      - All Buildings (80%) — Residential (30%), Nonresidential (50%), Formal (15%), Own account (15%)
      - Civil Works (20%) — Roads paved (10%), Roads gravel (5%), Water (5%)
- Recommendations and comparative approaches:
  - For recalculation of weights for a rebase, carefully consider methodology and adequacy of data sources with consultation with major users.
  - Kenya experience (mid-2013 IMF mission) recommended input survey of construction contractors to obtain 2013 expenditure data on materials, labor, and leasing of machinery/equipment to calculate detailed weighting patterns.
  - Recommended study tour to KNBS to evaluate methodology and compare with previous UBOS methodology before final decision on rebase methodology.
  - Rwanda experience (March 2015 IMF mission): NISR planned to derive industry and commodity weights from annual Integrated Business Enterprise Survey revenue and expense data; follow up with NISR and assess whether Ugandan industrial surveys (UBI and Annual Business Inquiry) classify expenses in sufficient detail to support this approach.
- Initialization and serviceability:
  - Depending on review outcomes, new price collections may need to be initialized.
  - CIPI should continue to be compiled monthly and ideally published on a timely basis each month.
  - Rebased index should be linked to the current index at the publication level to provide long-term series.
  - At some point, the index should be arithmetically re-referenced to a common period used for all UBOS economic indices.
- Work program elements for CIPI rebase:
  - i. Research and consultation to determine the most appropriate data sources for price indicators for materials at purchasers’ prices, labor and leasing charges.
  - ii. Investigation and consultation to determine optimum data sources and methodology for deriving representative industry and commodity weights.
  - iii. A review of the scope of the CIPI (e.g., extend Civil Works scope to electricity and gas?).
  - iv. Conduct a study tour to the KNBS, assuming progress on their CIPI, and consult with NISR.
  - v. Review compilation methodology and compare with the PPI-M spreadsheet system.
- Rebase frequency:
  - The index should be rebased every five years.

### Hotels and Restaurants Producer Price Index (H&R PPI) — design, weighting, and timetable
- Funding and work plan:
  - Funding has been approved for the rebase of the H&R PPI during 2015–16; a work plan needs preparation.
  - Work program target date: 2015.
- Weighting, index structure, and sample design:
  - Detailed information on gross output for the commodities in the index will need to be obtained from the sample of hotels and restaurants to assist in the derivation of the index weighting pattern. (para 90)
  - Short-term objective: focus on developing a rebased national index; separate regional indices could be developed in the longer term if user demand exists, but would require a much larger sample and separate regional weighting patterns. (para 91)
  - Higher level index weights will be based on 2009 Supply Use Tables data for “Accommodation” and “Catering Services.” More detailed weights will be derived from commodity data obtained from the sample of hotels and restaurants selected for the ongoing monthly price survey for a recent year (say, 2014). (para 99)
  - Detailed commodity weights will be derived from sales (revenue) data collected from the sample of hotels and restaurants selected for the price survey. (paras 100, 105)
  - Proposed index structure (subject to consultation to allow linking of old and new indices): 
    - 1. Accommodation
      - 1.1 Hotel accommodation (rooms, suites, etc.)
      - 1.2 Other hotel services (e.g., conference halls)
    - 2. Catering services
      - 2.1 Food and snacks
        - 2.1.1 Buffet
        - 2.1.2 A’ la carte
        - 2.1.3 Snacks
      - 2.2 Drinks
        - 2.2.1 Beer
        - 2.2.2 Soft drinks
        - 2.2.3 Wines and Spirits
- UBI gross output data (four-digit ISIC class level) dominant industries and gross output:
  - i. ISIC 5510 Accommodation – room services: U Sh222 billion
  - ii. ISIC 5610 Food :    U Sh22.5 billion
  - iii. ISIC 5630 Beverages :   U Sh8.5 billion
- Selection approach for sample of establishments (example for ISIC 5510):
  - Establishments with gross output of 1 percent or more were selected; 26 establishments accounted for over 80 percent of total ISIC class gross output. To represent small establishments, a further 10 establishments were selected using a skip interval. This analysis to be extended to ISIC classes 5610 and 5630.
  - Ensure “reasonable” geographic representation across the four regions and reconcile new sample with the old sample to avoid unnecessarily discarding existing establishments.
  - Annual review of sample recommended to introduce new businesses, include expanding ones, and remove contracting businesses.
- Conceptual basis, reference base, and formulas:
  - Conceptual basis: index will continue to relate to actual transaction prices for sales of accommodation and catering services; pricing point ex-hotel/restaurant; pricing basis basic prices exclusive of VAT and after any discounts, and include any subsidies. (para 92)
  - Reference base period: because of seasonality/irregular commodity prices, normalize base period prices by taking a twelve-month average; reference base period will relate to a full year, not a single quarter or month. (para 93)
  - Proposed recent year reference base to minimize need for retrospective prices; 2014–15 may be suitable depending on timing of initialization interviews and feasibility of obtaining retrospective prices. (para 95)
  - Aggregation and compilation methodology:
    - First-level aggregation: combine individual product prices from H&R PPS to index commodity level.
    - Elementary aggregate (EA) indices to be calculated using the Jevons price index formula.
    - EA indices to be progressively combined to higher levels and total index using the Laspeyres formula and weighting system derived with described methodology.
    - An unweighted EA formula selected due to infeasibility of maintaining lower-level product and establishment weights; IMF recommends a geometric mean formula over arithmetic mean formulae for the unweighted EA.
- Source data, price indicators, and collection procedures:
  - Current H&R PPS seeks mid-month price data for specific, representative products; collection is conducted quarterly and current form also seeks sales and other data not available until after quarter end.
  - Proposed redesign: review, redesign, and simplify form to relate to price data only, collected monthly. Pricing point is the 15th of each month; data can be obtained immediately after pricing point and timely monthly indices compiled and published.
  - New sample selection: based on 2009–10 UBI data for establishment gross output by ISIC class, using a combination of cut-off and judgmental sampling.
  - Price collection initialization/review: interviews required to (i) initiate 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.
  - Principles for selecting sampled products for monthly pricing:
    - i. Sampled products must be representative of a wider range of products in terms of price change over time.
    - ii. Usually the establishments’ largest product sales by value.
    - iii. For workload reasons, a manageable number of products should be selected to represent a commodity.
    - iv. To price to constant quality over time, full specification of all price-determining characteristics of the product, unit of quantity, etc. are required.
- Price recording conventions (from MPI/MPS guidance applicable to price surveys):
  - Prices must be actual transaction prices, valued on a CIF basis (inclusive of overseas freight and insurance). (para 7.1)
  - Preferably prices as at the 15th of the middle month of the quarter, or the nearest previous trading day. (para 7.2)
  - If shipments are infrequent select a shipment for a day as close as possible, but prior to, the 15th of the middle month. (para 7.3)
  - Seek prices for selected products as at the 15th of middle month of each previous quarter from June 2013 to 2015 during/after the interview. (para 7.4)
  - Operational procedures: create tailored form for each establishment, send for final clearance, deliver the form each quarter for price collection, and undertake follow-up/re-interviewing as necessary. (paras 7.7–7.9, 7.6)
- Serviceability, publication, linkage, and rebasing:
  - Proposed reference base and periodicity: rebased H&R PPI reference base of either 2014 or 2015 depending on timing of initialization/review interviews; compiled monthly either from January 2014 or January 2015. (para 117)
  - Publication goals: publish monthly indices at the commodity level in the index structure (see paragraph 102). (para 118)
  - Linkage requirement: new indices should be linked to the old indices at each level of aggregation to provide long-term series. (para 119)
  - Rebase frequency: indices should be fully rebased every five years. (para 120)
- Project timetable (Appendix VI) — Target date: 2015
  - 1. Finalize development of index structure. May 29
  - 2. Design H&R PPS, incl. price collection form and commodity weights form. June 5
  - 3. Calculate high level industry weights from UBI gross output data. June 5
  - 4. Select samples of hotels and restaurants for each ISIC class. June 5
  - 5. Conduct initialization interviews incl. collection of commodity sales data. End-September
  - 6. Calculate industry and commodity weights. Early October
  - 7. Loads all descriptors and weights onto spreadsheet system. Mid-October
  - 8. Ongoing price collection and index compilation. Ongoing
  - 9. Link old and new series. March 2016
  - 10. Publish new linked indices. March 2016
  - 11. Ongoing price collection, index compilation, and publication. Ongoing

### PPI–Manufacturing (PPI-M) — progress and linkage
- A short summary of progress on the rebase of PPI-M was presented.
- Tasks required before the new index can be published were identified; additional technical assistance provided, particularly on linking new and old indices to provide long-term series.
- PPI–Manufacturing tasks and timetable highlights:
  - i. complete compilation of indices for period September quarter 2013 to December quarter 2014;
  - ii. prepare publication manuscript format and content for December 2014 for internal UBOS analysis;
  - i. complete data gathering in the field for March quarter 2015;
  - ii. finalize data loading and index compilation for March quarter 2015;
  - iii. initiate the price collection for June quarter 2015;
  - iv. load all data and compile June quarter 2015 indices;
  - v. finalize publication manuscript for June quarter 2015;
  - vi. publish new rebased PPI-M from September quarter 2013 to June quarter 2015 in July 2015. (para 124)
- Linking methodology for PPI-M:
  - Link new September quarter 2013 rebased indices to old September quarter 2004 indices retrospectively using link factors as the ratio of the new index to the old index for an overlap quarter; link factors should be based on September quarter 2013 indices and applied at each level of the index structure in the publication, separately for domestic sales and exports. (paras 125–126)
- Longer-term scope extension:
  - Extend PPI scope to include utilities and agriculture; Agriculture Section has commenced initial investigations into development of an agriculture PPI. (para 127)

### Operational guidance and Appendix materials
- MPI/MPS fieldwork and interviewer instructions emphasize initial contact, telephone contact script, interview content, product selection principles, transaction price conventions (c.i.f.), timing (15th of middle month), and logistics for quarterly collection. (Appendix II)
- Appendix V contains an Example of Generic MPS Questionnaire with data fields:
  - HS Code No.; Full description of items imported; Unit of Quantity; Country of origin; Currency of transaction; c.i.f. Import Price per unit as at 15th of the month.
- Appendix VI provides the H&R PPI Project Tasks and Timetable with specific target dates (see Project timetable above).

*IMF Country Report cr17301 (excerpts provided in the content unit).*

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

### Executive Summary

### Mission overview
- Technical assistance mission to Kampala during the period May 4–22, 2015 to provide advice on developing export and import price indices (XMPI) and rebasing the construction industry price index (CIPI) and the hotels and restaurants producer price index (H&R PPI).
- Mission objectives included:
  - Review of the import price index (MPI) commodity selection criteria and compilation methodology.
  - Development of the Import Price Survey (MPS) and Export Price Survey (XPS) methodology.
  - Development of methodology for integrating MPS and XPS data with Customs data.
  - Finalization of XMPI compilation procedures, including imputation of missing prices.
  - Determination of strategy for rebase of CIPI and H&R PPI.
  - Training of staff on above topics.
  - Preparation of documentation on methodological and computational procedures.
- Training provided on price index concepts and compilation methodology, and price survey procedures.

### XMPI — Export and Import Price Indices: statistics prerequisites and implementation
- Conceptual design:
  - XMPI will be hybrid indices combining quarterly Customs unit values for relatively homogeneous HS commodities and survey-based prices from the MPS for more heterogeneous commodities.
  - Pricing point: XPI f.o.b.; MPI c.i.f.
  - Reference base period selected: 2009 (noted also as 2009–10 in discussion to be consistent with other economic indices).
- Implementation status and schedule:
  - Limited progress due to Trade Unit staff involvement in the 2014 Population Census and resource constraints; program of initialization interviews for the MPS had not yet commenced.
  - MPS initialization interviews planned for mid-July to end-September 2015, including collection of retrospective quarterly prices back to June quarter 2013 to supplement Customs unit values.
  - Hybrid indices to be compiled as soon as practical on a reference base of 2009.
- Processing systems:
  - Basic Excel spreadsheet prototypes used for calculation of price relatives at the eight-digit HS level using the Jevons geometric mean formula and for progressive aggregation with fixed weights using the Laspeyres formula.
  - An operational system of linked spreadsheets needs to be developed based on the prototypes to support initial and ongoing quarterly compilation of the MPI and XPI.
- Basket, structure, and weights:
  - Structure and weighting pattern derived using smoothed import values over the years 2011–13; a three-year average for 2011–13 was judged the most credible distribution for weighting.
  - One hundred and fifty-one eight-digit HS commodities were selected as the MPI index basket.
  - The same top-down methodology and procedures will be repeated by UBOS for the XPI; a much smaller sample of HS commodities will be required for the XPI because of relative trading patterns.
- Time series construction for survey-priced commodities:
  - Retrospective indices for commodities requiring survey prices will be compiled using a mixture of:
    - Back prices obtained during initialization interviews (to June quarter 2013); and
    - Customs unit values for the period from June quarter 2013 back to September quarter 2009, adjusted to remove outliers and modified in light of survey price performance.
  - It is not considered feasible to obtain retrospective prices over a longer period.
- Rationale and design decisions:
  - Elementary aggregate (EA) indices will be at the eight-digit HS level; unweighted EA indices calculated using the Jevons formula (geometric mean of price indicators).
  - EA indices will be aggregated with fixed weights using the Laspeyres formula.
  - An unweighted EA formula selected because maintaining lower-level product and establishment weights continuously is not feasible.
  - A common reference period is required for XPI and MPI to facilitate calculation of the Terms of Trade.
- Data sources:
  - Use Customs unit values for relatively homogeneous items where quarter-to-quarter unit value variations credibly measure pure price change.
  - For non-homogeneous commodities with noncredible unit value patterns, obtain transaction prices through the quarterly MPS.

### XMPI — accuracy, reliability, and methodological notes
- Conceptual basis: XMPI relate to actual transaction prices for exports and imports of merchandise trade.
- Reference period considerations:
  - Selection favored 2009–10 to be consistent with other economic indices and provide a sufficiently long time series for inflation analysis and national accounts deflation over using a more recent period (e.g., 2013–14 or 2014–15).
- Weighting base period and rebasing:
  - Weights derived from smoothed import values for 2011–13; updating to include 2014 was rejected due to problems with 2014 Customs data associated with change in definition of c.i.f.
- Commodity representation:
  - Top-down approach used to allow unpriced maximized items to be indirectly represented by allocating their weights to similar priced items.
  - Detailed methodology for derivation of structure, composition, and weights described in previous report.

### CIPI — Construction Industry Price Index: prerequisites and approach to rebase
- Rebase prerequisites and consultation:
  - Extensive consultation with key stakeholders required to determine primary purpose of the index and to formulate its conceptual basis, including scope.
  - Scope of CIPI needs review; scope of Civil Works should be extended to include additional activities, such as the generation and distribution of electricity and gas.
- Data source identification:
  - Identify most appropriate data sources for price indicators for materials, labor, and leasing charges.
  - Determine optimum data sources for derivation of representative industry and commodity weights.
- Compilation methodology:
  - After consultations and investigations to determine index scope and optimum data sources, compilation methodology should be reviewed and a detailed timetable developed.

### H&R PPI — Hotels and Restaurants Producer Price Index: rebase tasks and survey design
- Survey redesign and initialization:
  - Rebase will involve redesign of the price survey including the ongoing price collection form and a form for initial collection of commodity sales values for a recent year to support derivation of commodity weights.
  - A sample of hotels and restaurants needs selection for major ISIC classes, based on Uganda Business Inquiry gross output data by establishment, using a combination of cut-off sampling and application of a skip interval to randomly select a sample of smaller establishments.
  - Initialization interviews required to establish ongoing price collection and obtain a recent year’s commodity sales data on a one-off basis to support calculation of industry and commodity weights.
- Index compilation and linking:
  - Industry and commodity weights can then be calculated and loaded onto the spreadsheet system (based on the PPI-M system).
  - Ongoing compilation of the rebased index can be undertaken and long-term linked series published.
- Work program:
  - Work programs prepared for the XMPI, including programs of survey initialization interviews, and for the H&R PPI.

### Producer Price Index – Manufacturing (PPI-M) progress
- A short summary of progress on the rebase of PPI-M was presented.
- Tasks required before the new index can be published were identified.
- Additional technical assistance was provided, particularly on linking the new and old indices to provide long-term series.

*Executive Summary — IMF technical assistance report (May 4–22, 2015).*

### 38. Experience has shown that the data quality associated with the use of unit values is

### cr17301 - 38. Experience has shown that the data quality associated with the use of unit values is

### Unit values, survey linkage, and retrospective compilation
- Experience indicates data quality from unit values is potentially improved if country mix is eliminated from the eight-digit unit values (i.e., use individual country unit values or remove outlier countries).
- The quarterly unit values for each of the 151 selected HS commodities were analyzed over the three years 2011–13 to assess suitability as price indicators; smoothing of a few quarters’ outlier values in some series improved credibility.
- Individual country of origin unit values were assessed; for some initially unsuitable commodities, selecting unit values for major country/countries of origin or removing/smoothing outlier country data rendered the Customs data suitable and reduced dependence on price survey data.
- Outcome for the 151 eight-digit HS commodities:
  - 45 were designated as having credible price indicators based on unit values.
  - 106 required survey-based price indicators.
- For the 45 commodities with unit value-based price indicators, trends were analyzed back to the March quarter 2009, with further smoothing or removal of particular country data undertaken; this work needs review and finalization.
- For the 106 commodities requiring survey prices, retrospective indices will be compiled using:
  - (i) back prices obtained during initialization interviews (back to June quarter 2013) and
  - (ii) unit values for the period from June quarter 2013 back to March quarter 2009, adjusted to remove outliers and smoothed and modified in light of survey price performance.
- Linking unit value-based and survey-based EA commodity prices requires retrospective application of a factor equal to the ratio of the survey-based EA price to the unit value-based EA price for an overlap quarter.
- Alternative imputation methods for missing observations discussed include carry-forward, using movements in prices of related products from same or other importers, or projecting the recent trend.

### Sample selection for a Merchandise Price Survey (MPS)
- For the conduct of an MPS, a sample of importers needs to be selected from Customs data for each of the 106 HS commodities in the survey.
- Broad business rules for sampling (combination of cut-off and judgmental sampling):
  - i. at least two importers are required for each commodity, if possible;
  - ii. ideally significant coverage (in value terms) should be achieved (i.e., greater than 50 percent if feasible, particularly for large commodities);
  - iii. if very low concentration (i.e., dominated by a large number of small importers), select the largest importer then apply a skip interval to select a small, representative sample of small importers;
  - iv. priority should be given to achieving good coverage and representation of the high value commodities; and
  - v. for relatively small commodities with low concentration, delete the eight-digit item from the regimen and pro-rate its weight amongst the remaining eight-digit items in the four-digit heading.
- Initial Trade Unit staff selection resulted in about 500 importers; this is excessive and needs rigorous review and reduction by at least 50 percent, preferably to less than 200 importers.
- Review of initial sample selection should include:
  - (i) removing importers where there is a large number for an HS commodity;
  - (ii) aiming for lower coverage of the medium and small HS commodities; and
  - (iii) eliminating small HS commodities that have no dominant importers and reallocating their weight on a prorated basis to other commodities in the same four-digit HS heading.
- An annual review of the sample of importers should be undertaken to introduce new importers or existing ones that have expanded, and remove importers who have contracted.

### Price collection initialization and ongoing procedures
- Principles for selecting samples of specific products for quarterly price collection:
  - i. sampled products should be representative of a wider range of products in terms of price change over time;
  - ii. they will usually be the importers’ largest imports by value;
  - iii. if possible, they should be regularly imported;
  - iv. for workload reasons, a manageable number of products (say, restricted to three or four if possible) should be selected to represent an eight-digit HS item from each importer;
  - v. full specification of all price-determining characteristics, unit of quantity and transaction details such as country of origin and currency of the transaction are required to price to constant quality; prices are to be converted in the office to Ugandan shillings (U Sh) for index compilation.
- Supporting documentation prepared (to be modified for Ugandan conditions) includes:
  - i. “MPS Field Interviewers’ Instructions” (Appendix II);
  - ii. “Introductory Letter to Importers” (Appendix III);
  - iii. “MPS Interview Guide” (Appendix IV);
  - iv. “Example of a Generic MPS Questionnaire” (Appendix V) to be tailored to UBOS standards.
- Main steps for initialization and ongoing collection:
  - i. contacting each importer to arrange an interview;
  - ii. conducting interviews and selecting some three or four detailed representative product specifications per eight-digit HS code suitable for repeat pricing, ensuring full specification of price-determining characteristics;
  - iii. undertaking follow-up action, including re-interviews as necessary;
  - iv. creating a tailored collection form for each importer;
  - v. on an ongoing quarterly basis: delivering tailored collection forms, undertaking collection control, and editing and querying prices;
  - vi. regularly confirming and updating the representativeness of the sample of products with each importer;
  - vii. loading the prices, converting to Ugandan shillings, and compiling the indices.
- Training provided covered price index concepts and methodology, survey methodology, transaction pricing, product sampling and index aggregation, derivation of EA indices based on both survey prices and unit values, and application of base weights to calculate indices at higher levels of aggregation.
- A work program for development of the indices is provided in Appendix I; completion of the 2014 Population Census is expected to improve availability of funding and staff resources to progress the XMPI project.

### Serviceability, periodicity, and dissemination (XMPI)
- Planned compilation and publication schedule:
  - The 2009–10 based XMPI is planned to be compiled quarterly from September 2009, and annually from 2009–10.
- Publication goals:
  - Publish indices at the two-digit HS Chapter level together with total indices, subject to analysis of the data.
  - Annual and quarterly indices down to the eight-digit HS commodity level should be provided to national accounts on an as-needed basis, with any necessary additional comments about the data.
- Rebase policy:
  - Indices should be fully rebased at least every five years. Each year, monitor import/export patterns to assess whether significant changes warrant earlier rebasing.

---

### Construction Industry Price Index (CIPI) — Statistics prerequisites and methodology
- Prerequisites and resource implications:
  - Once decisions on data sources for new index weights and future price indicators are made following industry and government consultation, a work program and resource estimate can be developed.
  - If a survey of the value of materials used by construction companies is required for weights, and new monthly or quarterly surveys of material prices are required for ongoing price indicators at correct conceptual basis, substantial financial and staff resources will be needed for the rebase.
- Conceptual and source issues to assess:
  - Determine whether producer price survey (PPS) prices relating to manufacturers’ output prices (ex-factory) at basic prices are adequate indicators for input prices at purchasers’ prices required for the CIPI.
  - Analyze whether the PPS prices relate to the appropriate transaction for the CIPI (i.e., who construction companies purchase from, and the pricing basis such as ex-factory or ex-wholesaler).
  - Assess suitability of import prices on a c.i.f. basis as indicators of construction industry purchasers’ prices; MPS prices could be used if suitable.
  - Review the scope of the index, e.g., whether to include other major civil works such as electricity and gas generation and distribution.
- Reference base period:
  - Cannot be determined until issues relating to future data sources for weights and price indicators are resolved.
- Aggregation and compilation methodology:
  - Basic heading indices (monthly) are compiled using price data from a range of sources; EAs are compiled using the Jevons formula from PPI, CPI, and External Trade Statistics.
  - Prices of bitumen and diesel are collected directly from oil companies; wages and equipment hire charges from construction companies.
  - Price relatives for individual product prices from oil companies and wages are calculated and aggregated to basic heading level using the Jevons formula.
  - Basic heading indices are aggregated to composite indices (e.g., building and civil works) using the modified Laspeyres formula.
- Rebase frequency:
  - The index should be rebased every five years.
- Past weighting methodology and current structure:
  - Upper level weights for the present CIPI were calculated in 2007 using data from Ministry of Works and Transport and Road Agency Formation Unit investment/expenditure records, Kampala City Council building approval records, Directorate of Water Development and National Water and Sewerage Corporation records, and UBOS national accounts.
  - Table 1. Current CIPI Structure and Weighting Pattern (as provided):
    - Whole Sector (100%)
      - All Buildings (80%) — Residential (30%), Nonresidential (50%), Formal (15%), Own account (15%)
      - Civil Works (20%) — Roads paved (10%), Roads gravel (5%), Water (5%)
    - Source: Ministry of Works and Transport and Road Agency Formation Unit, Uganda.
  - Basic heading weights were derived from Bills of Quantities of typical projects; quantities multiplied by prevailing prices at the time (2005–06). Bills obtained from contractors, quantity surveyors, and contract committees.
- Recommendations and comparative approaches:
  - For recalculation of weights for a rebase, carefully consider methodology and adequacy of data sources with consultation with major users.
  - Kenya experience (mid-2013 IMF mission): recommended input survey of construction contractors to obtain 2013 expenditure data on materials, labor, and leasing of machinery/equipment to calculate detailed weighting patterns.
  - Recommended study tour to KNBS to evaluate methodology and compare with previous UBOS methodology before final decision on rebase methodology.
  - Rwanda experience (March 2015 IMF mission): NISR planned to derive industry and commodity weights from annual Integrated Business Enterprise Survey revenue and expense data; follow up with NISR and assess whether Ugandan industrial surveys (UBI and Annual Business Inquiry) classify expenses in sufficient detail to support this approach.
- Source data and price indicators for CIPI:
  - Currently main source of materials prices for CIPI is basic prices (ex-factory) from the PPS.
  - Research needed to determine whether PPS ex-factory prices are adequate indicators for purchasers’ prices required by CIPI.
  - Analyze whether PPS prices represent the appropriate transactions (sales to construction companies versus wholesalers), and consider adding specifications to PPS forms (e.g., prices of sales to construction companies) if buyers face different prices.
  - May need to collect prices directly from building materials wholesalers to reflect VAT, freight, and wholesale margins.
  - Equipment leasing charges currently collected from construction companies; may be more appropriate to collect from equipment owners.
  - A thorough review of data and data sources for CIPI prices is recommended due to sensitivity of indices to inaccuracies.
  - Explore source data issues during proposed study tour to KNBS and consultation with NISR; share experiences with other African countries where possible.
- Initialization and serviceability:
  - Depending on review outcomes, new price collections may need to be initialized.
  - CIPI should continue to be compiled monthly and ideally published on a timely basis each month.
  - Rebased index should be linked to the current index at the publication level to provide long-term series.
  - At some point, the index should be arithmetically re-referenced to a common period used for all UBOS economic indices.
- Work program elements for CIPI rebase:
  - i. Research and consultation to determine the most appropriate data sources for price indicators for materials at purchasers’ prices, labor and leasing charges.
  - ii. Investigation and consultation to determine optimum data sources and methodology for deriving representative industry and commodity weights.
  - iii. A review of the scope of the CIPI (e.g., extend Civil Works scope to electricity and gas?).
  - iv. Conduct a study tour to the KNBS, assuming progress on their CIPI, and consult with NISR.
  - v. Review compilation methodology and compare with the PPI-M spreadsheet system.

### Hotels and Restaurants Producer Price Index (H&R PPI) — prerequisites
- Funding has been approved for the rebase of the H&R PPI during 2015–16; a work plan needs preparation.
- Required actions and resources:
  - Staff and financial resources are required to complete index structure and weighting pattern development, develop the compilation spreadsheet system, redesign the H&R Producer Price Survey (H&R PPS), select the sample of hotels and restaurants, commence price survey through initialization interviews, and conduct ongoing price collection and index compilation.

*IMF report excerpt (cr17301).*

### 90. Detailed information on gross output for the commodities in the index will need to be

### cr17301 - 90. Detailed information on gross output for the commodities in the index will need to be

### Weighting, index structure, and sample design
- Detailed information on gross output for the commodities in the index will need to be obtained from the sample of hotels and restaurants to assist in the derivation of the index weighting pattern. (para 90)
- Short-term objective: focus on developing a rebased national index; separate regional indices could be developed in the longer term if user demand exists, but would require a much larger sample and separate regional weighting patterns. (para 91)
- Higher level index weights will be based on 2009 Supply Use Tables data for “Accommodation” and “Catering Services.” More detailed weights will be derived from commodity data obtained from the sample of hotels and restaurants selected for the ongoing monthly price survey for a recent year (say, 2014). (para 99)
- Detailed commodity weights will be derived from sales (revenue) data collected from the sample of hotels and restaurants selected for the price survey. (paras 100, 105)
- Proposed index structure (current structure to be retained subject to consultation to allow linking of old and new indices): (para 102)
  - Industry Group and commodity
    - 1. Accommodation
      - 1.1 Hotel accommodation (rooms, suites, etc.)
      - 1.2 Other hotel services (e.g., conference halls)
    - 2. Catering services
      - 2.1 Food and snacks
        - 2.1.1 Buffet
        - 2.1.2 A’ la carte
        - 2.1.3 Snacks
      - 2.2 Drinks
        - 2.2.1 Beer
        - 2.2.2 Soft drinks
        - 2.2.3 Wines and Spirits
- Supply Use Tables provided gross output data only for “Accommodation” and “Catering.” (para 103)
- UBI gross output data (four-digit ISIC class level) dominant industries and gross output: (para 104)
  - i. ISIC 5510 Accommodation – room services: U Sh222 billion
  - ii. ISIC 5610 Food :    U Sh22.5 billion
  - iii. ISIC 5630 Beverages :   U Sh8.5 billion
- Selection approach for sample of establishments (example for ISIC 5510): establishments with gross output of 1 percent or more were selected; 26 establishments accounted for over 80 percent of total ISIC class gross output. To represent small establishments, a further 10 establishments were selected using a skip interval. This analysis to be extended to ISIC classes 5610 and 5630. (paras 110–112)
- Ensure “reasonable” geographic representation across the four regions and reconcile new sample with the old sample to avoid unnecessarily discarding existing establishments. (para 113)
- Annual review of sample recommended to introduce new businesses, include expanding ones, and remove contracting businesses. (para 114)

### Accuracy and reliability — conceptual basis, base period, and formulas
- Conceptual basis: index will continue to relate to actual transaction prices for sales of accommodation and catering services; pricing point ex-hotel/restaurant; pricing basis basic prices exclusive of VAT and after any discounts, and include any subsidies. (para 92)
- Reference base period: because of seasonality/irregular commodity prices, normalize base period prices by taking a twelve-month average; reference base period will relate to a full year, not a single quarter or month. (para 93)
- Proposed recent year reference base to minimize need for retrospective prices; 2014–15 may be suitable depending on timing of initialization interviews and feasibility of obtaining retrospective prices. (para 95)
- Current collection practice: monthly prices collected each quarter and quarterly average prices computed. Funding permitting, propose monthly collection and timely monthly indices compiled and published. (para 94)
- Aggregation and compilation methodology: (paras 96–98)
  - First-level aggregation: combine individual product prices from H&R PPS to index commodity level.
  - Elementary aggregate (EA) indices to be calculated using the Jevons price index formula.
  - EA indices to be progressively combined to higher levels and total index using the Laspeyres formula and weighting system derived with described methodology.
  - An unweighted EA formula selected due to infeasibility of maintaining lower-level product and establishment weights; IMF recommends a geometric mean formula over arithmetic mean formulae for the unweighted EA.

### Source data, price indicators, and price collection procedures
- Current H&R PPS seeks mid-month price data for specific, representative products defined in paragraph 102. Collection is conducted quarterly and current form also seeks sales and other data not available until after quarter end. (paras 106–107)
- Proposed redesign of PPS form: review, redesign, and simplify form to relate to price data only, collected monthly. Pricing point is the 15th of each month; data can be obtained immediately after pricing point and timely monthly indices compiled and published. (para 108)
- New sample selection: based on 2009–10 UBI data for establishment gross output by ISIC class, using a combination of cut-off and judgmental sampling. (para 109)
- Price collection initialization/review: interviews required to (i) initiate 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. (para 115)
- Principles for selecting sampled products for monthly pricing: (para 116)
  - i. Sampled products must be representative of a wider range of products in terms of price change over time.
  - ii. Usually the establishments’ largest product sales by value.
  - iii. For workload reasons, a manageable number of products should be selected to represent a commodity.
  - iv. To price to constant quality over time, full specification of all price-determining characteristics of the product, unit of quantity, etc. are required.
- Price recording conventions (from MPI/MPS guidance applicable to price surveys in the document): (Appendices II–IV; paras 7.1–7.4)
  - Prices must be actual transaction prices, valued on a CIF basis (inclusive of overseas freight and insurance). (para 7.1)
  - Preferably prices as at the 15th of the middle month of the quarter, or the nearest previous trading day. (para 7.2)
  - If shipments are infrequent select a shipment for a day as close as possible, but prior to, the 15th of the middle month. (para 7.3)
  - Seek prices for selected products as at the 15th of middle month of each previous quarter from June 2013 to 2015 during/after the interview. (para 7.4)
- Operational procedures: create tailored form for each establishment, send for final clearance, deliver the form each quarter for price collection, and undertake follow-up/re-interviewing as necessary. (paras 7.7–7.9, 7.6)

### Serviceability, publication, and rebasing
- Proposed reference base and periodicity: rebased H&R PPI reference base of either 2014 or 2015 depending on timing of initialization/review interviews; compiled monthly either from January 2014 or January 2015. (para 117)
- Publication goals: publish monthly indices at the commodity level in the index structure (see paragraph 102). (para 118)
- Linkage requirement: new indices should be linked to the old indices at each level of aggregation to provide long-term series. (para 119)
- Rebase frequency: indices should be fully rebased every five years. (para 120)

### Training and work program
- Price collection staff training required on index concepts and index construction methodology. (para 121)
- A broad work program is provided in Appendix VI. (para 122)
- PPI–Manufacturing related tasks and timetable highlights (paras 123–126): main tasks to publish rebased PPI-M include:
  - i. complete compilation of indices for period September quarter 2013 to December quarter 2014;
  - ii. prepare publication manuscript format and content for December 2014 for internal UBOS analysis (main features, conceptual notes, graphs, tables of index numbers and percentage changes, appendix with long-term linked indices at broad level);
  - i. complete data gathering in the field for March quarter 2015;
  - ii. finalize data loading and index compilation for March quarter 2015;
  - iii. initiate the price collection for June quarter 2015;
  - iv. load all data and compile June quarter 2015 indices;
  - v. finalize publication manuscript for June quarter 2015;
  - vi. publish new rebased PPI-M from September quarter 2013 to June quarter 2015 in July 2015. (para 124)
- Linking methodology for PPI-M: link new September quarter 2013 rebased indices to old September quarter 2004 indices retrospectively using link factors as the ratio of the new index to the old index for an overlap quarter; link factors should be based on September quarter 2013 indices and applied at each level of the index structure in the publication, separately for domestic sales and exports. (paras 125–126)
- Longer-term scope extension: extend PPI scope to include utilities and agriculture; Agriculture Section has commenced initial investigations into development of an agriculture PPI. (para 127)

### Operational guidance from appendices (MPI/MPS fieldwork and interviewer instructions)
- XMPI project tasks and timetable target year: 2015. Key tasks include developing linked spreadsheet processing system, calculation of EA commodity indices using Jevons Geometric Mean formula, aggregation using Laspeyres formula with fixed weights, selection of samples, pilot survey, replacement of unit values with survey prices, recompile MPI by December 15, and ongoing quarterly compilation. (Appendix I)
- MPS field interviewers’ instructions emphasize initial contact, telephone contact script, interview content, product selection principles, transaction price conventions (c.i.f.), timing (15th of middle month), and logistics for quarterly collection. (Appendix II)
- Introductory letter and interview guide templates provided for contacting importers and establishing tailored reporting forms and product specifications. (Appendices III–IV)

*Italic: IMF Country Report cr17301 (excerpts provided in the content unit).*

### Appendix V. Example of Generic MPS Questionnaire

### Appendix V. Example of Generic MPS Questionnaire

### Purpose, Authority, and Confidentiality
- PURPOSE: The prices collected in this survey are used to calculate the Import Price Index, which measures changes in prices of import into Tanzania.
- COLLECTION AUTHORITY: The information asked for is collected under the authority of the Statistic Act 2002. In accordance with this Act you are required to complete the questionnaire and return it to the undersigned by the due date. The data reported on this questionnaire will be treated in strict confidence, used for statistical purposes only.
- CONFIDENTIALITY OF INFORMATION: Your completed form remains confidential to the National Bureau of Statistics.
- HELP AVAILABLE: If you have problems completing this form, or feel you may have difficulties in meeting the due date, please contact:

### Instructions for Price Reporting
- 1. In the table below, please record import prices (c.i.f.) per unit in terms of currency of the transaction (e.g., US. dollar, Tanzanian shilling, euro, etc.).
- 2. Import prices recorded should be the actual prices paid for a shipment on the 15th of the middle month of the Quarter (e.g., 15th May for the June quarter). If this is not feasible, prices for a shipment as close as possible, but prior to the 15th of the middle month of the specified quarter, should be recorded.

### Import Prices (c.i.f.) Data Fields (to be completed for the Quarter)
- HS Code No.
- Full description of items imported
- Unit of Quantity (eg., kg, litres, etc.)
- Country of origin
- Currency of transaction (e.g., U.S. dollar, Tanzanian shilling, euro, etc.)
- c.i.f. Import Price per unit as at 15th of the month

(Example numbered rows provided in the form)
- 01.
- 02.
- 03.
- 04.
- 05.
- 06.

### Follow-up Question and Respondent Details
- 1. Have the prices of these products changed since the previous quarter? Yes / No
- If YES, please provide reason(s) for the change(s)?
  - ___________________________________________________________________________
  - _________________________________________________
  - _____________________________________
- Respondent identification fields to be completed:
  - Name ...................................
  - Title ...........................
  - Telephone .............................
  - Mobile .................................
  - Email:
  - Signature..............................
  - Date.....................................

### Closing
- THANK YOU FOR YOUR COOPERATION

---

### Appendix VI. H&R PPI Project Tasks and Timetable

- Target date: 2015

Project tasks and dates:
- 1. Finalize development of index structure. May 29
- 2. Design H&R PPS, incl. price collection form and commodity weights form. June 5
- 3. Calculate high level industry weights from UBI gross output data. June 5
- 4. Select samples of hotels and restaurants for each ISIC class. June 5
- 5. Conduct initialization interviews incl. collection of commodity sales data. End-September
- 6. Calculate industry and commodity weights. Early October
- 7. Loads all descriptors and weights onto spreadsheet system. Mid-October
- 8. Ongoing price collection and index compilation. Ongoing
- 9. Link old and new series. March 2016
- 10. Publish new linked indices. March 2016
- 11. Ongoing price collection, index compilation, and publication. Ongoing

*Appendix V. Example of Generic MPS Questionnaire (cr17301) and Appendix VI. H&R PPI Project Tasks and Timetable*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17301.pdf_