## REPORT ON THE RESIDENTIAL PROPERTY PRICE INDEX MISSION (NOVEMBER 30–DECEMBER 4, 2020)

## Source details

**Canonical URL:** [REPORT ON THE RESIDENTIAL PROPERTY PRICE INDEX MISSION (NOVEMBER 30–DECEMBER 4, 2020)](https://www.imf.org/-/media/files/publications/cr/2022/english/1geoea2022003.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2022/english/1geoea2022003.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2022/english/1geoea2022003.pdf.json)

---

### Summary of mission outcomes and priority recommendations
- Purpose: assist the National Statistics Office (GeoStat) in progressing on the compilation of a residential property price index (RPPI); mission conducted under a three-year project funded by the Government of The Netherlands; focused on finalizing work for launching the index from 2021.
- Importance: RPPI identified by the G-20 Data Gaps Initiative and guidance on financial soundness indicators as critical for financial stability policy analysis and macroprudential measures; RPPI is also a macro-economic indicator of growth and key to understanding financial market conditions.
- Implementation status:
  - File management system put in place.
  - Data from two websites being collected by web scraping.
  - Experimental compilation with three hedonic methods conducted quarterly.
- GeoStat plans:
  - Compile a quarterly RPPI covering new flats and new detached houses for Tbilisi.
  - Obtain data from residential dwellings announcements via web scraping from the two main Georgian websites.
  - First release planned for April 25, 2021 (three weeks after end of period); mission recommended publishing a technical note and metadata with the first release.
  - Publish other real estate indicators (average price per square meter and per district and rates of change; real RPPI deflated with the CPI; RPPI deflated with the USD exchange rate; number of transactions and rate of change using NARP data).
  - Future improvements: expand coverage to existing dwellings; expand to country coverage; include data from the second website.
- Institutional data source (NAPR): data transmitted regularly but contains few variables and cannot distinguish new vs existing dwellings; workshop in Batumi improved cooperation but a firm written commitment at top management level is needed to improve administrative data.
- CPI modernization: work on scanner data and web scraping; recommendation to continue work to implement automatic data transmission.

Priority recommendations (from Table 1):
- March 2021 — Firm written commitment (Memorandum of Understanding) at top management level to improve data collection from NAPR. — Responsible: GeoStat and NAPR
- April 2021 — Publish a technical note and metadata with the first release. — Responsible: GeoStat
- April 2021 — Compile and publish other real estate indicators. — Responsible: GeoStat

### Detailed technical assessment and recommendations
- Outcomes and milestones (selected from action plan):
  - January 2021 — Assess the relevance of the new variables and consider dropping the non-relevant.
  - March 2021 — Firm written commitment (Memorandum of Understanding) at top management level to improve data collection from NARP.
  - April 2021 — Release the index compiled only with data from the first website.
  - Continuous — Continue web scraping data from the second website.
  - April 2021 — Publish a technical note and metadata with the first release.
  - April 2021 — Compile and publish other real estate indicators.
  - April 2022 — Compile the index with data from two websites from 2022.
  - April 2022 — Expand the coverage to existing dwellings.
  - April 2023 — Expand the coverage to country wide.
  - Ongoing — Continue work to implement automatic scanner data transmission.

- Recommended actions (consolidated):
  - Assess relevance of new variables and drop non-relevant ones.
  - Establish a firm top-management level commitment (Memorandum of Understanding) to improve data collection from NARP.
  - Release the index compiled only with data from the first website (start from 2020).
  - Continue web scraping data from the second website and add to compilation from 2022.
  - Publish a technical note and metadata with the first release.
  - Compile and publish selected other real estate indicators.
  - Expand coverage to existing dwellings and countrywide in phased manner.
  - Continue implementation of automatic scanner data transmission for CPI modernization.

### A. Residential Property Price Index — technical findings and guidance
- Mission history: fifth price statistics mission to GeoStat; prior missions in April–May 2018, December 2018, third mission (date not specified), and fourth mission focused on producer price index.
- Data sources and coverage:
  - GeoStat collects web-scraped announcement data from two main websites covering all dwelling types and the whole country; initial compilation will begin with new dwellings in Tbilisi only.
  - For one website: data available since 4Q2018; since 3Q2019 seven more variables were added; tests show improvement for all sub-indices though some variables have no impact—relevance should be assessed and non-relevant dropped.
  - For the second website: collection began 4Q2019; structural changes required adaptation of scraping code; resource constraints delayed retrieval until 4Q2020.
  - NAPR administrative data continue to be transmitted regularly but include few variables and cannot distinguish new vs existing dwellings; unusable for index calculation but usable for weights? The mission concluded weights cannot be derived because distinction between new and existing dwellings cannot be made.
- Methodology and experimental results:
  - Experimental compilation performed with three hedonic methods (characteristics hedonic, characteristics-imputations hedonic, and imputations hedonic); the three indices follow the same trend; characteristics and imputations hedonic methods produced exactly the same results.
  - GeoStat revised codes to implement improvements to the RPPI practical compilation guide.
- Release strategy and data governance:
  - First release on April 25, 2021 to be compiled only with data from the first website and starting from 2020 (data before 2020 judged to have less quality).
  - Exclude second website data during 2021 because it does not cover the full year 2020 needed for weights; continue scraping it for inclusion from 2022 to mitigate dependence on a single data source.
  - Publication should include a technical note and metadata; a template from the Batumi workshop was revised and prepared.
  - GeoStat to prepare a more detailed Sources and Methods document addressing worldwide status, historical developments, uses and relevance, links with economic growth, monetary policy and financial regulations; reasons for selecting the characteristics hedonic method and results from other methods during the experimental phase; plans for future development.
- Aggregation and weights considerations for two-website compilation (from 2022):
  - Two options: merge datasets and compile a single index; or compile sub-indices per website and aggregate.
  - Prefer compiling sub-indices per website and aggregating with a Laspeyres type index due to difficulty identifying duplicates across datasets.
  - Data source for weights at website sub-index level: turnover reported for this activity may be inappropriate because enterprises report advertising turnover together with other products; therefore weights should be obtained from one-year data of advertised prices.

### B. Modernization of the CPI
- Scanner data:
  - Scanner data obtained since April from two retailers, covering the first 10 days of each month.
  - Mapping of GTINs to COICOP in progress with promising results.
  - Around 500 new codes per month require manual mapping; retailers providing their own classification eases mapping.
- Automatic transmission:
  - GeoStat faced some reluctance from retailers for automatic data transmission.
  - Mission recommended preparing and sharing preliminary results with providers to improve engagement and timely data transmission.
- Web scraping and hedonic methods:
  - Being explored to improve the secondhand cars sub-index.
  - Key variables recommended for secondhand cars hedonic regressions: vintage, kilometrage, general condition of the vehicle, make and model, horsepower, and consumption.
- Recommended action:
  - Continue work to implement automatic data transmission.

### C. Officials met during the mission
- Giorgi Tetrauli — Head of Price Statistics Department, GeoStat

*Prepared By Vanda Guerreiro. Report completed on December 2020.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1geoea2022003.pdf_