21
response error, and non-sampling error; the results of the assessments are monitored and made available to guide statistical processes.
• Accuracy of the data from surveys is routinely assessed.
Technical reports are prepared and published on website indicating
extent of sampling errors, biases and non-response.
Periodically survey is conducted to verify the accuracy of the data and
during validation missing and outliers value are highlighted for further
action to be taken.
A matrix of price event is introduced into data collection forms so that
earlier detection of outliers is made. Second outlier detection is done by
CPI compiler by flagging out price changes exceed a pre-set threshold,
NBS does not use automatic deletion of outlier because some outliers are
true price value.
Item with missing price due to seasonality or temporally absence are imputed to a maximum of three (3) months.
3.3 Statistical techniques Relevance
3.3.1 Data compilation employs sound statistical techniques to deal with data sources
• Data compilation procedures are sound . CPI Compiler is used to simplify the task to minimize processing errors such as coding, editing, and tabulation errors. Monthly reports on errors observed during data collection, data entry, and data validation stages together with any compilation issues are circulated to relevant members of staff with a view to take action to rectify such errors. Unusual values are identified during validation; imputation method is used for non- response cases.
• Appropriate statistical methods are used to handle missing prices and the introduction of new products that are within the scope of the CPI.
Appropriate statistical methods are used to handle missing prices and to facilitate the introduction of new products that are within the scope of the CPI. Missing data is calculated using a well-recognized statistical technique recommended by the CPI manual. Prices of missing products
22
are imputed on a monthly basis using month to month price changes of a higher group. Prices are imputed for seasonal products using the price relative of the sub group in which the items reside. Item with missing price due to seasonality or temporally absence are imputed to a maximum of 3 months. For seasonal products imputation method is done until a product exist in the market. New product qualify in basket if it gains a share of 0.001 and more and it can qualify to be in the basket due to its necessity ( e.g. Salt, match box, razor blade etc).
3.3.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques.
• Appropriate techniques are used to address specific issues of measurement for CPI weights.
The methodological soundness dimension is assessed against the guidelines outlined in the System of National Accounts 2008 (2008 SNA) and the Consumer Price Index (CPI) Manual – Concept and Methods 2020. The concepts and definitions from the 2008 SNA are used as guidelines with regard to coverage and valuation. The methods and procedures from the Consumer Price Index (CPI) Manual – Concept and Methods 2020 are used as guidelines for compiling the CPI, for owner- occupied dwellings, consumption is valued as rentals that tenants would pay for similar accommodation. Goods and services produced for own final consumption are valued at market prices at the time produced. Furthermore, non-monetary expenditure is not covered by the CPI.
• Calculation of the elementary (item) level indices
The arithmetic mean of price relatives (Carli) is not used to calculate the elementary (item) level indexes due to its bias, that the ratio of arithmetic mean prices (Dutot) is only be used for homogeneous item and that the geometric mean of price relatives (equivalent to the ratio of geometric means) is adopted as the preferred measure.
• Aggregation of the higher-level indices
23
Currently, the weight reference period precedes the price reference period. When fixed-quantity basket index is desired, the weights are adjusted for price change to align with the price reference period when they were introduced. Price index is calculated at each higher level of aggregation on the basis of the weights and price indices for its components, that is, the lower-level or elementary price indices are aggregated to form the higher indices by weighting together lower-level indices through progressive levels of aggregation, as defined by the classification structure.
• Weight reference period
New weight is adopted after every five (5) years as the chain link technique is used to recompute the old index and if there is any changes during rebasing that change is documented.
During rebasing new products are introduced into the samples, new formulas, and classification systems are also updated and the weights at the lower level are re-established on the basis of more recent information.
3.4 Assessment and validation of intermediate data and statistical outputs
3.4.1 Intermediate results are validated against other information where applicable
• Data compiled from the main data sources are validated against other independent data sources.
The CPI is compared with comparable estimates from other major price indices mainly Producer Price Index (PPI) during price stabilisation committees’ meetings. Comparison with export and import price indices are not currently being done, but there are plans in place to start such comparisons.
3.4.2 Statistical discrepancies in intermediate data are assessed and investigated
• Unusual index movements arising from potential problems in price data are investigated.
The reasons for the unusual index movement are evaluated and published in the CPI report.
24
3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated.
• Statistical discrepancies, if any, between CPI for geographical areas and the CPI for products are investigated and measures taken to remove or explain them.
The CPI data is thoroughly reviewed and edited to correct any errors or inconsistencies. Adjustments are undertaken to deal with discrepancies in the data before finalizing.
3.5 Revision studies
3.5.1 Studies and analyses of revisions and/or updates are carried out and used internally to inform statistical processes (see also 4.3.3).
• Revision and/or update studies are undertaken on a regular basis.
Given the nature of consumer behaviour, items that disappear in the CPI basket are usually substituted with new like items. However, the substitution effects are not studied. Studies of long-term trends in the revision pattern are planned to be conducted in future to identify systematic bias in revisions. In an event of error in CPI aggregation, detailed studies are normally conducted to identify the source of error such as carrying out consistency checks for CPI weights and price data.
• Measures are undertaken to incorporate the findings from revision and/or update studies in data compilation.
Documentation on revisions including methods and data sources during
rebasing are documented. However, Studies to examine bias of CPI is not
conducted.
Findings from revision studies are not carried out to define the optimal
revision cycle. However, studies to refine data collection programs for
the subsequent periods is normally carried out and outcome of
assessment of other sources of error informs CPI compilation methods
such as improving the CPI processor. There is no adequate
documentations on revision processes.
- Serviceability
25
4.1 Periodicity and timeliness
4.1.1 Periodicity follows dissemination standards
• The periodicity of the statistics follows the IMF data dissemination standards (Special Data Dissemination Standard (SDDS) or General Data Dissemination System (GDDS))
The CPI is disseminated on monthly basis that follows the IMF data dissemination standards (Special Data Dissemination Standard (SDDS)). The NBS website http://www.nbs.go.tz is used to disseminate published monthly CPI and inflation statistics while disseminated time series monthly and annual data are published through the Statistical Abstract.
4.1.2 Timeliness follows dissemination standards
• The timeliness of statistical series follows the IMF data dissemination standards (SDDS or GDDS.
Monthly CPI and Inflation reports are disseminated to the public (ie. policy makers, BOT, EAC, SADC, IMF, researchers and other interested individuals) on the 8th of the month following the reference month. This is aligned with the IMF data dissemination standards (SDDS). Moreover dissemination of CPI data is also largely been disseminated within two months according to GDDS.
4.2 Consistency
4.2.1 Statistics are consistent within the dataset
• Statistical series is internally consistent
Compilation of CPI aggregation is being done hierarchically from elementary aggregates to sub classes, classes, groups, divisions and all items. Each of these levels of aggregations produces all item index regardless of typology of aggregation.
4.2.2 Statistics are consistent or reconcilable over a reasonable period of time
26
• The statistical series is consistent over time.
During the CPI rebasing exercises, a significant proportion of items in the old basket resurface in the new basket, price data for these items in the old basket are used to recompute indices for previous 12 months. For the foreseen new emerging items, data collection for such items begins 12 months before rebasing period to enable a 12-month index comparison. This facilitates compilation of 12-month price changes of the rebased CPI without altering previous published series.
There is clear documentation regarding a list of new items that have emerged in the rebased series, items that have disappeared and those that have resurfaced and necessary adjustment to link the series is well elaborated during the rebasing exercise.
CPI is normally affected by changes in economic trends such as decreased production of commodities or changes in demand and supply of commodities in the market. Effect on prices as a result of these changes are well elaborated and made public to users.
4.2.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical framework
• The statistics are consistent or reconcilable with other statistical frameworks.
Compilation of CPI uses the same internationally accepted classification to ensure consistency of the CPI series over time is ensured by classifying goods and services into the respective divisions, groups, classes, sub- classes and elementary aggregates according to COICOP 2018. The same classifications is used in aggregation of other economic statistics, making it possible to map CPI with older series and other economic statistics.
4.3 Revision policy and practice
4.3.1 Revisions and/or updates follow a regular and transparent schedule
• The practice of revisions (e.g., from provisional estimates, for weight updates, for changes in methodology) follows a predictable pattern of which users of statistics are informed.
CPI Revision is recommended to be done after every five years as per the requirement of the EAC Guideline where the rebasing cycles are well known to CPI users, e.g. 2010, 2015, 2020 and 2025. Incidences of
27
revisions are minimized by effective validation and ensuring completeness of data submitted from the field.
Availability of new consumption and expenditure data reflecting current consumer behavior from HBS forms a basis for CPI revision. Such reasons are made public to CPI users through stakeholders’ engagement and clear documentation is made in the rebasing report. However, revisions or updates outside the regular cycle are not conducted and planned to be implemented in the future.
4.3.2 Preliminary and/or revised/ updated data are clearly identified
• Data series subject to revision are clearly identified
Data series subject to revision are not clearly identified, planned to be implemented in the future
• Users are informed about the preliminary nature of the data.
The CPI data released to the press and the publications for the most recent month are not indicated as preliminary and could be revised where applicable. Planned to be implemented in future.
• Users are informed about the revised and/or updated nature of the data Since CPI does not include data revision, users are not informed about the nature of revision if any over disseminated preliminary data.
4.3.3 Studies and analyses of revisions are made public
• Users are informed of the results and studies of the revisions to the statistics.
Tanzania has not adopted the CPI revision policy, and therefore studies are not conducted to assess the differences between the preliminary data and revised data to determine the reliability of preliminary data.
- Accessibility 5.1 Data accessibility