| 9
Harmonized index of consumer prices at constant tax rates
The Harmonized Index of Consumer Prices at constant tax rates ( HICP-CT)8 is calculated as established
by the Regulation (EC) no 119/2013 of the 11 th February 2013. It measures the change of prices at
constant tax rates. It follows the same co mputation principles as the HICP, but it is based on prices ‘at
constant tax rates’.
Prices at constant tax rates are estimated cancelling out the effects due to changes in taxes in the current
month compared to the tax rates system in force in December of previous year (calculation period base).
The taxes considered in the HICP -CT are those directly linked to final consumption. They are mainly VAT,
excise duties and other taxes on some specific items (such as cars and insurance). Subsidies and taxes
paid on intermediate stages (e.g. production, transportation) are not taken into account. In principle, for the
compilation of HICP -CT, all taxes should be included and kept constant; however, due to practical
consideration, taxes which generate very small tax revenues may not be taken into account. In detail,
according to recommendations reported in the Eurostat HICP -CT Manual, taxes which cover less than 2%
of the total tax revenue can be excluded. On the whole, included taxes must cover a minimum of 90% total
tax revenue. Therefore in the compilation of the Italian HICP -CT, taxes kept constant are the following:
VAT, excise duties on tobacco and energy items (fuels, heating oil, gas, electricity, etc.), the main local
surcharge on electricity and gas, tax for the public liability insurance and contribution to the National Health
Service for transport means insurance. On the basis of National Accounts data taxes which cover less than
1% of the total tax revenue are excluded and, on the whole, taxes included cove r almost 98% of total
revenues carried out with taxes on final consumption.
The HICP -CT covers the same goods and services as those covered by the HICP. The same weight
structure is applied as for the HICP (Table 1). As HICP, it has expressed in 2015=100 as a reference base
year.
The HICP -CT provides a measure of the theoretical impact of changes of indirect taxes on the overall
HICP inflation. It has to be emphasised that it does not provide an exact measure of this impact, rather an
indication for its upper limit. In effect, the difference between HICP and HICP -CT growth rates points to the
theoretical impact of tax changes on overall HICP inflation, assuming an instantaneous and full pass -
through of tax rate changes on the price paid by the consumer.
It has to be pointed out that, during the year, the Italian HICP -CT may be revised following introduction of
methodological changes required by indirect taxation system changes. Data become final in the next year
to the reference one.
Indices rates of change calculation
Hereafter formulae for the calculation of monthly, annual and annual average rates of change for consumer
price indices are described 9. The HICP formulae apply also to HICP -CT. The first expression concerns
calculation of rates of change between indices in the same reference base period:
Monthly rate of change (NIC, HICP)
The monthly rate of change is the current month’s index in respect to the previous month’s index (with one
decimal place), for example:
1100100
2012
2012
20122012 .;I
IRoundI;IMOR
,Jan
,Feb
,Feb,Jan
Annual rate of change (NIC, HICP)
The annual rate of change is the current month’s index in respect to the same month’s index a year
previously (with one decimal place), for example:
8 The HICP-CT has been released starting from data referred to March 2012. Back series starting from January 2002 are published on
I.Stat, inside the theme Prices http://dati.istat.it.
9 The expressions and the rounding rules described for NIC are also carried out for FOI.
| 10
1 100 100 2011 2012 2012 2011 . ; I I Round I; I ANR , Feb , Feb , Feb , Feb
Annual average rate of change (NIC) The annual average rate of change is the current annual average index in respect to a previous annual average index (with one decimal place), for example: 1 100 100 2011 2012 2012 2011 . ; I I Round I; I AVR
Annual average rate of change (HICP) For the HICP, in a different way compared to NIC, the annual average rate of change is obtained directly from the monthly indices and therefore it is based on the unrounded annual average indices. This method, applied in compliance with Eurostat, guarantees international comparability of data. For example:
1 100 100 2011 2011 2011 2012 2012 2012 2012 2011 . ; I ... I I I ... I I Round I; I AVR , Dec , Feb , Jan , Dec , Feb , Jan
The following expression describes the calculation of monthly rate of change between indices expressed in different reference base year; it can be also used for the calculation of the annual rate of change and the annual average rate of change:
Monthly rate of change - Indices expressed in different reference base year
t X h n X j m I I MOR , , ; 1 1. ; 100 100 ; ... ; ; 1 2 2 1 1 1 , , X X C X X C X X C I I Round t t t t X j m t X h n
where 1 , X j m I is the index, with one decimal place, of the month m year j, expressed in the more remote reference base 1 X , t X h nI , is the index, with one decimal place, of the month n year h, expressed in the more recent reference base t X , and ) ; ( 1 i i X X C with i=2.….t are the splicing coefficients between contiguous reference bases. These coefficients are equal to the annual average index of the year corresponding to the new reference base expressed in the previous base, divided by 100. They are as many as base changes have been carried out during the considered period.
Flash estimates of HICP: accuracy and computation methodology
Flash estimate of Italian HICP (and NIC) are usually published on the last working day of the reference
month according to the Eurostat release calendar of HICP flash estimate for euro area. Final data are
generally published around 13 days later.
The aim of the inflation flash estimates is to provide a timely information on inflation, predicting as
accurately as possible the final HICP (and NIC) annual rate of change released about two weeks later. The
analysis of their revisions represents an important tool to evaluate the correct balancing between the two
quality dimensions, timeliness and accuracy.
Totally in line with the Eurostat Statistics Explained on Inflation – methodology of the euro area flash
estimate, this section analyses the accuracy of the Italian HICP flash estimates and describes the
methodology used in their computation.
| 11
Accuracy of flash estimates Table 2 compares the flash estimates and the final HICP annual rates of change for the same reference month. Over the last thirteen months, the maximum difference between the flash estimate – all items and the HICP – all items annual rates of change was +0.3 in January 2017. Over the same period, with reference to the main special aggregates, the maximum differences between the flash estimate and the final HICP annual rates of change concerned Food, including alcohol and tobacco (+1.1 recorded in January 2017), Processed food (including alcohol, tobacco) (+1.2 in January 2017), Unprocessed food (+0.9 in January 2017), Energy (+0.6 in February 2017) and Non energy industrial goods (+0.5 in January 2017). The highest frequency of revisions for Non energy industrial goods ( ten months out of 13) are mainly due to th e seasonal sales dynamics of Clothing and footwear, for which the partial information available has a higher impact on the flash estimate and therefore it turns out to be less accurate.
TABLE 2. FLASH ESTIMATES AND HICP ANNUAL RATES FOR THE ALL-ITEMS AND MAIN SPECIAL AGGREGATES November 2016 - November 2017, percentage values (base 2015=100) Special aggregates Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 May-17 Jun-17 Jul-17 Aug-17 Sep-17 Oct-17 Nov-17 Flash 0.5 1.1 1.2 3.5 2.7 1.9 1.8 0.8 0.8 0.8 1.3 1.9 1.6 HICP 0.5 1,0 2.3 3.6 2.7 1.9 1.7 0.8 0.8 0.7 1.3 1.9 1.7 Flash 0.5 0.5 -0.7 0.6 0.7 0.5 0.5 0.4 0.5 0.6 0.6 0.7 0.8 HICP 0.5 0.5 0.5 0.6 0.7 0.5 0.4 0.4 0.5 0.6 0.6 0.7 0.9 Flash 0.5 1.8 3.6 7.4 5.2 3.8 3.4 1.2 1.3 0.9 2.1 3.3 2.8 HICP 0.5 1.8 4.5 7.4 5.4 3.8 3.4 1.3 1.3 0.9 2.1 3.3 2.8 Flash -2.9 -2.0 2.6 4.2 4.5 7.5 6.4 4.6 3.5 4.5 3.4 3.7 4.4 HICP -2.9 -2.0 2.7 4.8 4.6 7.4 6.4 4.6 3.4 4.5 3.4 4.0 4.4 Flash 0.2 0.3 0.0 0.4 -0.4 0.3 0.2 0.3 0.3 0.7 0.7 0.3 0.3 HICP 0.3 0.4 0.5 0.1 0,0 0.2 0.3 0.3 0.3 0.7 0.8 0.2 0.4 Flash 0.6 0.9 0.6 1.0 1.1 1.8 1.3 1.4 1.3 1.6 1.3 0.7 0.5 HICP 0.5 0.9 0.6 1.0 1.1 1.8 1.3 1.4 1.3 1.6 1.3 0.6 0.5 Flash 0.1 0.5 0.7 1.6 1.3 2.0 1.5 1.2 1.2 1.4 1.3 1.1 1.1 HICP 0.1 0.5 1.0 1.6 1.4 2.0 1.6 1.2 1.2 1.4 1.3 1.1 1.1 Flash 0.5 0.7 0.1 0.7 0.6 1.2 0.8 0.9 0.8 1.1 0.9 0.6 0.4 HICP 0.5 0.7 0.5 0.6 0.7 1.2 0.8 1.0 0.8 1.1 1.0 0.5 0.5 Flash 0.4 0.7 0.3 0.7 0.5 1.3 0.9 1.0 0.9 1.2 1.1 0.5 0.4 HICP 0.4 0.7 0.5 0.7 0.6 1.3 0.9 1.0 0.9 1.2 1.1 0.5 0.4 Flash 0.5 0.8 0.5 1.4 1,0 1.4 1.1 0.9 0.9 1.1 1.1 0.8 0.7 HICP 0.5 0.9 0.9 1.3 1.1 1.4 1.1 1.0 0.9 1.1 1.1 0.8 0.7 All-items All items excluding energy and unprocessed food (Core inflation) All items excluding energy, food, alcohol and tobacco All items excluding energy Food including alcohol and tobacco: Processed food (including alcohol, tobacco) Unprocessed food Energy Non energy industrial goods Services
The Mean Absolute Deviation (MAD) provides another way to measure accuracy. It is calculated as the average of the absolute differences between the flash estimate a nd the final HICP annual rates of change over the last thirteen months. Figure 1 shows the MAD for the all -item index and the main special aggregates. Over the last thirteen months Non energy industrial goods (0.1 46 percentage points) and Food including alcohol and tobacco (0.123 percentage points) have recorded the highest MADs.
| 12
FIGURE 1. MEAN ABSOL UTE DEVIATION BETWEE N FLASH ESTIMATES AND HICP A NNUAL RATES . November 2016 - November 2017, percentage points
0.062 0.023 0.077 0.038 0.015 0.146 0.100 0.092 0.108 0.123 All items excluding energy All items excluding energy, food, alcohol and tobacco All items excluding energy and unprocessed food (Core inflation) All-items Services Non energy industrial goods Energy Unprocessed food Processed food (including alcohol, tobacco) Food including alcohol and tobacco 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
The direction of inflation is correctly predicted if both the flash estimate and the final one show increasing (declining or no changing) annual rates of change with respect to those ones calculated in the previous month. There are three possible outcomes for the comparison of the direction of inflation:
-
the flash estimate correctly predicts the direction of inflation, s o the predicted rise, decline or no change in inflation is confirmed by final data (denoted by );
-
the flash estimate wrongly predicts the direction of inflation, namely it predicts an increase when there is a decrease or vice versa (denoted by );
-
the flash estimate points to an increase or a decrease but the final annual rate of change remains unchanged; or the flash estimate predicts no change in inflation but the final figure points to an increase or a decrease (denoted by ). Over the last thirteen months, the flash estimate accurately predicted the inflation direction in 1 19 out of 130 estimates. TABLE 3. FLASH ESTIMATE PREDICTION CAPACITY OF THE DIRECTION OF INFLATION MEASURED BY HICP November 2016 - November 2017 Special Aggregates Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 May-17 Jun-17 Jul-17 Aug-17 Sep-17 Oct-17 Nov-17 Food including alcohol and tobacco: Processed food (including alcohol, tobacco) Unprocessed food Energy Non energy industrial goods Services All-items All items excluding energy and unprocessed food (Core inflation) All items excluding energy, food, alcohol and tobacco All items excluding energy
| 13
Computation methodology of flash estimates For the Italian HICP (and NIC) flash estimate compilation, each month.
prices collected at local level by 61 municipalities (out of 80) are used. Out of these municipalities, there are the 38 municipalities which calculate the preliminary local consumer price indices and publish them independently, at the same time of Istat national CPI and HICP release. Data collected by the other 16 municipalities participating in the survey for a subset of products (local tariffs and some local services) are not used; these data are used for the compilation of final indices;
all prices collected directly by ISTAT (via internet and other sources) are used. As soon as indices are
calculated for aggregate products for which prices are collected directly by ISTAT, product aggregate
indices for the municipalities, which participate in the flash estimate of inflation rate, are compiled. For
the other municipalities, which do not participate in the flash estimation, product aggregate indices are
generally10 calculated applying to the indices of the previous month, the monthly rate of change of the
regional product aggregate indices. The latter are calculated using data of municipalities which
participate in the flash estimate, as follows:
R
i
a
m
h
i
R
i
i
i
a
m
h
R
I
I
,
,
where
a,
m
h
i I
is the elementary index of product aggregate h at municipality level i of the reference month
m of year a and
R
i
i
i
is equal to the share of resident population in the municipality i of region R on the
total resident population of the region.
As soon as product aggregate indices of all municipalities are compiled, regional and, then, national indices
are calculated (by product aggregates, by upper aggregates and for all items).
If all municipalities of a certain region are not included in the flash estimate, the product aggregate indices
of this region are calculated applying to the indices of the previous month, the monthly rate of change of
national product aggregate indices. The latter are calculated using data of regions which participate in the
flash estimate, as follows:
20
1
,
20
1
,
R
a
m
h
R
R
h
R
h
R
a
m
h
I
I
where
a,
m
h
R I
is elementary index of product aggregate h at regional level of the reference month (m) of
year (a) and
20
1
R
h
R
h
R
is equal to the share of household consumption expenditure for the product
aggregate h in the region R on the national household consumption expenditure for the same product
aggregate.
Once product aggregate indices of all regions are compiled, national indices are calculated (by product
aggregates, by upper aggregates and for all items).
10 For some product aggregates – among others, rents and local tariffs such as water supply, solid waste, sewerage collection, urban transport services by road – for the municipalities that do not participate in the flash estimation, indices are estimated by carrying forward the price of the previous month. The adoption of this different estimation technique is due to the fact that the evolution of prices in the other municipalities of the same region is not considered a satisfactory proxy.