## cr17323

## Source details

**Canonical URL:** [cr17323](https://www.imf.org/-/media/files/publications/cr/2017/cr17323.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2017/cr17323.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2017/cr17323.pdf.json)

---

### Mission scope and context
- Mission dates: April 17–28, 2017.
- Prepared by: Dale A. Smith (August 2017).
- Purpose: Technical assistance to the National Statistics Office of Mongolia (NSOM) to update and improve:
  - Consumer Price Index (CPI)
  - Industrial Producer Price Index (IPPI)
  - Restaurant Index
  - Export Import Unit Value Indexes (XMUVI)
  - Review of index compilation system
- Background: Fourth technical assistance mission following April 2012, September 2013, and October 2014 missions; built on recommendations from national accounts mission in September 2016.

### Executive summary — key findings and priority recommendations
- CPI weights for automobiles and motorcycles likely overstate significance because they include secondhand purchases between households; future HIES must identify such sales to derive net household expenditure.
- Recommendation to adjust expenditure shares in the updated CPI using alternative data sources and improve monthly price-change compilation for automobiles and motorcycles.
- Rebased IPPI selection of enterprises and products must ensure appropriate export coverage; review 2015 customs data is recommended; collect separate prices for domestic transactions and exports.
- For XMUVI, develop alternative methodologies to exclude quality-change effects for exported copper concentrate; compile analytical series and assess suitability.
- NSOM has implemented an internationally standard methodology in the system used to compile the updated CPI; imputation procedures still require review.
- NSOM agreed to implement the internationally standard methodology when the rebased IPPI is published in 2018.

Priority Recommendations:
- Adjust the expenditure share for automobiles and motorcycles in the updated CPI based on alternative data sources for sales of automobiles and motorcycles by households.
- Ensure the future HIES compiles data for sales of automobiles and motorcycles by households.
- Institute a compilation procedure for the automobile and motorcycle components of the CPI that holds the age, rather than the model year, constant.
- Ensure the IPPI’s selection of enterprises and products reflect the significance of exports to production.
- Compile analytical XMUVI series for copper concentrate using alternative methods, and assess their suitability.
- Review the imputation procedures in the index compilation system employed by the updated CPI to ensure they are in line with international guidelines.
- Compile the rebased IPPI with the system used to compile the updated CPI.

### Consumer Price Index (CPI)
Findings and context:
- Current CPI last revised in January 2012 based on the 2010 HIES.
  - Weight reference period: calendar year 2010.
  - Index reference period: December 2010.
- Updated CPI weight reference period: calendar year 2015 (based on 2015 HIES).
- The updated CPI published in April 2017 used an index reference period of April 2017 (note in source).
- Updated CPI must be recompiled back to January 2015 to enable linking with the current CPI.
- Overlap between current and updated indexes: 27 months (January 2015 to March 2017); mission recommends using calendar year 2015 as the overlap period.

Recommendations for linking and timing:
- Commence the updated CPI in January 2015 to enable linking with the current CPI.
- Use calendar year 2015 as the overlap period when compiling linked CPI time series.

Automobiles and motorcycles — weights and pricing methodology:
- Expenditure shares for automobiles and motorcycles should be derived using net purchases; HIES must collect both purchases and sales by households.
- Current CPI overstates expenditure share because purchases between households were included in expenditure (2010 HIES lacked sales items).
- 2015 HIES likewise lacks data items for sales by households; adjustment to expenditure shares in the updated CPI is recommended using alternative data sources (such as national accounts).
- Pricing practice to date:
  - Single brand/model year priced over time (e.g., a single Toyota Prius model year).
  - Neither model year nor brand is changed over time.
- Recommended revised methodology for vehicle pricing:
  - Expand coverage to include four different brands/types of vehicles and four different model years for each brand/type.
  - Institute ongoing sample rotation to hold age constant rather than model year: replace the oldest model year for one brand/type each quarter with a corresponding brand/type that is four years newer, so over a twelve-month period the oldest model year for each brand/type would be updated.

CPI vehicle component recommendations:
- Adjust the expenditure share for automobiles and motorcycles in the updated CPI based on alternative data sources for sales of automobiles and motorcycles by households.
- Ensure the future HIES compiles data for sales of automobiles and motorcycles by households.
- Increase coverage of automobile and motorcycle components to four brands/types and four model years per brand/type.
- Institute a compilation procedure that holds vehicle age constant via quarterly rotation (replace the oldest model year for one brand/type each quarter with a model four years newer).

### Industrial Producer Price Index (IPPI)
Current compilation and scope:
- Monthly IPPI compiled since January 2010 for products primary to mining, manufacturing, electricity and water industries.
- Indexes compiled for 89 products in 52 four-digit ISIC 3.1 categories.
- Monthly price data for 511 specifications collected from 151 enterprises.
- Weight reference period: calendar year 2010.
- NSOM plans to publish a rebased IPPI (to 2015) in January 2018.

Key recommendations for rebasing and export coverage:
- Selection of enterprises and products in the rebased IPPI should ensure appropriate coverage of exports; selection being carried out at four-digit level of ISIC v.4.
- Include a review of 2015 customs data to ensure relevant enterprises and products are selected in proportion to their significance to overall production.
- Collect separate prices for domestic transactions and exports where product characteristics vary by purchaser (quality, terms of sale, etc.) to support disaggregation of IPPI between domestic transactions and exports.

Rebasing procedure:
- Recompile rebased IPPI back to January 2015 using weights derived from 2015 production data to enable linking with current IPPI.
- Use calendar year 2015 as the overlap period when compiling linked IPPI time series.

IPPI recommendations:
- Ensure the IPPI’s selection of enterprises and products reflect the significance of exports to production; collect separate prices for domestic transactions and exports to support disaggregation and compilation of an export price index based on constant quality products with fixed specifications rather than Customs unit values.
- Commence the rebased IPPI in January 2015.
- Use calendar year 2015 as the overlap period when compiling linked IPPI time series.

### Restaurant Index
Findings and data:
- Restaurant Index under development.
- Sample selected from 2012–2014 revenue data: 125 enterprises reporting large volume of sales in all three years.
- Attempt to collect product detail in 2015 was incomplete; revised collection undertaken in 2016 expected to provide desired product detail.

Recommendations:
- Construct Restaurant Index weights and sample selection using 2016 sales data for the selected 125 enterprises; supplement with 2015 data where 2016 is unavailable.
- Commence the Restaurant Index from January 2015.

### Export Import Unit Value Indexes (XMUVI)
Current practice and problems:
- NSOM compiles and disseminates various XMUVI series.
- Sample selection derived from Customs data for 51 months (2010, 2011, 2012, 2013, and the first quarter of 2014); products with transactions in 25 or more of the 53 months were selected.
- Exports: 77 products in 15 groups comprising 97 percent of the value of exports were selected.
- Imports: 248 products in 19 groups comprising 77 percent of the value of imports were selected.
- Unit values are problematic when products are not homogeneous because changes reflect both product differences and price change; previous missions recommended restricting unit values to homogeneous goods to reduce volatility.
- Pilot specification-based price survey (September 2014):
  - Export prices collected from 2 enterprises (covering 1 of the 77 exported products: raw plastic materials).
  - Import prices collected from 40 enterprises (covering 64 of 248 products).
  - Bulk of XMUVI data still comprises unit values from Customs.

Index construction practice and concerns:
- NSOM compiled XMUVI as Fisher indexes with 12 month moving averages to reduce volatility.
- Mission observation: resulting indexes remain volatile and 12 month moving averages:
  - (a) atypical of other countries;
  - (b) complicated;
  - (c) unlikely to meet requirements of national accountants.

Recommendations and specific product issues:
- Replace unit value data obtained from Customs with export prices collected by the IPPI.
- Include export prices in the IPPI to support compilation of an export price index based upon constant quality products with fixed specifications rather than Customs unit values.
- For exported copper concentrate, trace mineral proportions (gold, silver, molybdenum) do not remain constant over time; further efforts required to exclude the impact of quality change from the XMUVI.
- Compile analytical XMUVI series for copper concentrate using alternative methods, and assess their suitability.

### Compilation systems and general recommendations
Findings:
- The index compilation systems currently employed by the NSOM are suboptimal and comprise the most substantial impediment to the orderly compilation of price indexes which meet international guidelines and reflect best practice.
- The current systems are largely Excel-based and employ the current-to-base period formulation of the Laspeyres index formula and do not allow for the automatic imputation of missing prices.
- The Excel-based index compilation system developed for the updated CPI was reviewed and confirmed to use the two-stage modified Laspeyres formulation.
- The mission did not have time to review the system’s imputation procedures to ensure all imputations are self-correcting. Self-correcting imputations are essential to ensure the imputed values are brought back to the next directly observed price and described in international guidelines.
- All other indexes use an index compilation system which is not based on the two-stage modified Laspeyres formulation.
- It was agreed to compile the rebased IPPI with the two-stage modified Laspeyres formulation system used by the updated CPI when implementing the rebased IPPI in 2018.

Recommendations:
- Review the imputation procedures in the index compilation system employed by the updated CPI to ensure they are in line with international guidelines.
- Compile the rebased IPPI with the two-stage modified Laspeyres formulation system used to compile the updated CPI.

*Source: IMF Country Report No. 17/323 — MONGOLIA: Technical Assistance Report — Report on the Prices Statistics Mission (April 17–28, 2017). (cr17323)*

### Section 1

### REPORT ON THE PRICES STATISTICS MISSION (Section 1)

### Mission scope and context
- Mission dates: April 17–28, 2017.
- Prepared by: Dale A. Smith (August 2017).
- Purpose: Technical assistance to the National Statistics Office of Mongolia (NSOM) to update and improve:
  - Consumer Price Index (CPI)
  - Industrial Producer Price Index (IPPI)
  - Restaurant Index
  - Export Import Unit Value Indexes (XMUVI)
  - Review of index compilation system
- Background: Fourth technical assistance mission following April 2012, September 2013, and October 2014 missions; built on recommendations from national accounts mission in September 2016.

### Executive summary — key findings and priority recommendations
- CPI weights for automobiles and motorcycles likely overstate significance because they include secondhand purchases between households; future HIES must identify such sales to derive net household expenditure.
- Recommendation to adjust expenditure shares in the updated CPI using alternative data sources and improve monthly price-change compilation for automobiles and motorcycles.
- Rebased IPPI selection of enterprises and products must ensure appropriate export coverage; review 2015 customs data is recommended; collect separate prices for domestic transactions and exports.
- For XMUVI, develop alternative methodologies to exclude quality-change effects for exported copper concentrate; compile analytical series and assess suitability.
- NSOM has implemented an internationally standard methodology in the system used to compile the updated CPI; imputation procedures still require review.
- NSOM agreed to implement the internationally standard methodology when the rebased IPPI is published in 2018.

Priority Recommendations (from Executive Summary)
- Adjust the expenditure share for automobiles and motorcycles in the updated CPI based on alternative data sources for sales of automobiles and motorcycles by households.
- Ensure the future HIES compiles data for sales of automobiles and motorcycles by households.
- Institute a compilation procedure for the automobile and motorcycle components of the CPI that holds the age, rather than the model year, constant.
- Ensure the IPPI’s selection of enterprises and products reflect the significance of exports to production.
- Compile analytical XMUVI series for copper concentrate using alternative methods, and assess their suitability.
- Review the imputation procedures in the index compilation system employed by the updated CPI to ensure they are in line with international guidelines.
- Compile the rebased IPPI with the system used to compile the updated CPI.

### Consumer Price Index (CPI)
Findings and context:
- Current CPI last revised in January 2012 based on the 2010 HIES.
  - Weight reference period: calendar year 2010.
  - Index reference period: December 2010.
- Updated CPI weight reference period: calendar year 2015 (based on 2015 HIES).
- The updated CPI published in April 2017 used an index reference period of April 2017 (note in source).
- Updated CPI must be recompiled back to January 2015 to enable linking with the current CPI.
- Overlap between current and updated indexes: 27 months (January 2015 to March 2017); mission recommends using calendar year 2015 as the overlap period.

Recommendations for linking and timing:
- Commence the updated CPI in January 2015 to enable linking with the current CPI.
- Use calendar year 2015 as the overlap period when compiling linked CPI time series.

Automobiles and motorcycles — weights and pricing methodology:
- Expenditure shares for automobiles and motorcycles should be derived using net purchases; HIES must collect both purchases and sales by households.
- Current CPI overstates expenditure share because purchases between households were included in expenditure (2010 HIES lacked sales items).
- 2015 HIES likewise lacks data items for sales by households; adjustment to expenditure shares in the updated CPI is recommended using alternative data sources (such as national accounts).
- Pricing practice to date:
  - Single brand/model year priced over time (e.g., a single Toyota Prius model year).
  - Neither model year nor brand is changed over time.
- Recommended revised methodology for vehicle pricing:
  - Expand coverage to include four different brands/types of vehicles and four different model years for each brand/type.
  - Institute ongoing sample rotation to hold age constant rather than model year: replace the oldest model year for one brand/type each quarter with a corresponding brand/type that is four years newer, so over a twelve-month period the oldest model year for each brand/type would be updated.

Recommendations (CPI vehicle components):
- Adjust the expenditure share for automobiles and motorcycles in the updated CPI based on alternative data sources for sales of automobiles and motorcycles by households.
- Ensure the future HIES compiles data for sales of automobiles and motorcycles by households.
- Increase coverage of automobile and motorcycle components to four brands/types and four model years per brand/type.
- Institute a compilation procedure that holds vehicle age constant via quarterly rotation (replace the oldest model year for one brand/type each quarter with a model four years newer).

### Industrial Producer Price Index (IPPI)
Current compilation and scope:
- Monthly IPPI compiled since January 2010 for products primary to mining, manufacturing, electricity and water industries.
- Indexes compiled for 89 products in 52 four-digit ISIC 3.1 categories.
- Monthly price data for 511 specifications collected from 151 enterprises.
- Weight reference period: calendar year 2010.
- NSOM plans to publish a rebased IPPI (to 2015) in January 2018.

Key recommendations for rebasing and export coverage:
- Selection of enterprises and products in the rebased IPPI should ensure appropriate coverage of exports; selection being carried out at four-digit level of ISIC v.4.
- Include a review of 2015 customs data to ensure relevant enterprises and products are selected in proportion to their significance to overall production.
- Collect separate prices for domestic transactions and exports where product characteristics vary by purchaser (quality, terms of sale, etc.) to support disaggregation of IPPI between domestic transactions and exports.
- Rebasing procedure:
  - Recompile rebased IPPI back to January 2015 using weights derived from 2015 production data to enable linking with current IPPI.
  - Use calendar year 2015 as the overlap period when compiling linked IPPI time series.

Recommendations (IPPI):
- Ensure the IPPI’s selection of enterprises and products reflect the significance of exports to production; collect separate prices for domestic transactions and exports to support disaggregation and compilation of an export price index based on constant quality products with fixed specifications rather than Customs unit values.
- Commence the rebased IPPI in January 2015.
- Use calendar year 2015 as the overlap period when compiling linked IPPI time series.

### Restaurant Index
Findings and data:
- Restaurant Index under development.
- Sample selected from 2012–2014 revenue data: 125 enterprises reporting large volume of sales in all three years.
- Attempt to collect product detail in 2015 was incomplete; revised collection undertaken in 2016 expected to provide desired product detail.

Recommendations:
- Construct Restaurant Index weights and sample selection using 2016 sales data for the selected 125 enterprises; supplement with 2015 data where 2016 is unavailable.
- Commence the Restaurant Index from January 2015.

### Export Import Unit Value Indexes (XMUVI)
Current practice and problems:
- NSOM compiles and disseminates various XMUVI series.
- Sample selection derived from Customs data for 51 months (2010, 2011, 2012, 2013, and the first quarter of 2014); products with transactions in 25 or more of the 53 months were selected.
- Exports: 77 products in 15 groups comprising 97 percent of the value of exports were selected.
- Imports: 248 products in 19 groups comprising 77 percent of the value of imports were selected.
- Unit values are problematic when products are not homogeneous because changes reflect both product differences and price change; previous missions recommended restricting unit values to homogeneous goods to reduce volatility.
- Pilot specification-based price survey (September 2014):
  - Export prices collected from 2 enterprises (covering 1 of the 77 exported products: raw plastic materials).
  - Import prices collected from 40 enterprises (covering 64 of 248 products).
  - Bulk of XMUVI data still comprises unit values from Customs.

Index construction practice and concerns:
- NSOM compiled XMUVI as Fisher indexes with 12 month moving averages to reduce volatility.
- Mission observation: resulting indexes remain volatile and 12 month moving averages:
  - (a) atypical of other countries;
  - (b) complicated;
  - (c) unlikely to meet requirements of national accountants.

Recommendations and specific product issues:
- Reiterate previous recommendations to use specification-based price data obtained directly from exporters and importers.
- Include export prices in the IPPI to support compilation of an export price index based upon constant quality products with fixed specifications rather than Customs unit values.
- For exported copper concentrate (a significant product), trace mineral proportions (gold, silver, molybdenum) do not remain constant over time; further efforts required to exclude the impact of quality change from the XMUVI.
- Compile analytical XMUVI series for copper concentrate using alternative methods, and assess their suitability.

### Compilation systems and general recommendations
- NSOM has implemented an internationally standard methodology within the system used to compile the updated CPI (as recommended by previous missions).
- Mission did not have sufficient time to review the system’s imputation procedures and encouraged NSOM to complete this activity.
- NSOM agreed to implement the internationally standard methodology when the rebased IPPI is published in 2018.
- Review imputation procedures in the index compilation system used for the updated CPI to ensure alignment with international guidelines.

*Source: IMF Country Report No. 17/323 — MONGOLIA: Technical Assistance Report — Report on the Prices Statistics Mission (April 17–28, 2017).*

### Section 2

### cr17323 - Section 2

### Specific price series issues and recommendations
- Several possible methodologies were discussed, including construction of a hedonic price index.
- Recommendations:
  - Replace unit value data obtained from customs with export prices collected by the IPPI.
  - Compile analytical XMUVI series for copper concentrate using alternative methods, and assess their suitability.

### Compilation systems and recommendations for all price indexes
- Paragraph 23:
  - The index compilation systems currently employed by the NSOM are suboptimal and comprise the most substantial impediment to the orderly compilation price indexes which meet international guidelines and reflect best practice.
  - The current systems are largely Excel-based and employ the current-to-base period formulation of the Laspeyres index formula and do not allow for the automatic imputation of missing prices.
  - These issues are discussed in previous mission reports which recommended adopting the two-stage modified Laspeyres formulation for compiling all the price index series.

- Paragraph 24:
  - The Excel-based index compilation system developed for the updated CPI was reviewed and confirmed to use the two-stage modified Laspeyres formulation.
  - The mission did not have time to review of the system’s imputation procedures to ensure all imputations are self-correcting.
  - Self-correcting imputations are essential to ensure the imputed values are brought back to the next directly observed price and described in international guidelines.

- Paragraph 25:
  - All other indexes use an index compilation system which is not based on the two-stage modified Laspeyres formulation.
  - The mission discussed the possibility of compiling the rebased IPPI with the two-stage modified Laspeyres formulation system used by the updated CPI.
  - It was agreed this change would be undertaken when implementing the rebased IPPI in 2018.

- Recommendations:
  - Review the imputation procedures in the index compilation system employed by the updated CPI to ensure they are in line with international guidelines.
  - Compile the rebased IPPI with the two-stage modified Laspeyres formulation system used to compile the updated CPI.

*Source: cr17323 - Section 2*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17323.pdf_