Trend differences between CPI and HICP

In the present analysis we intend to turn out that the methodological differences between the computation of the Consumer Price Index ( CPI ) and that of the Consonant Index of Consumer Prices ( HICP ) are non important, and that the basket used for the consonant index ( demuring some minor differences ) is composed of the same nutrient and non-food trade goods and services. However, it should be mentioned that in the instance of the consonant index, goods and services are divided into 12 groups, unlike the three chief groups of CPI. Consumer monetary value index ( CPI ) is a man-made index of economic activity that best expresses and characterizes the overall development of monetary values for the goods bought and of the duties of services used by the population in a given period ( the current period ) , as compared with a old period ( reference period ) . Harmonized index of consumer monetary values measures the alterations in monetary values and duties for goods and services subject to retail in Romania, the weighting coefficients being derived from the construction of the disbursals made for this intent by Rumanian and foreign consumers. Therefore, harmonised index is based on the internal construct of ingestion, taking into history the ingestion of all families in the economic district of the state, irrespective of income, nationality, societal or residential position except for the foreign embassies in Romania. More precisely, the index definition requires non merely the taking into history of occupants ‘ ingestion, but besides of the disbursals made by foreign visitants in Romania. Calculations made in Romania by the National Institute of Statistics are besides communicated to Eurostat. The Regulations of the European Council acknowledge the being and rightness of utilizing both of the indices. The two indices complement each other though they have different maps. Therefore, HICP is used for comparing the rates of rising prices in Member States and for set uping the Central European Bank pecuniary policy, whereas CPI is used for lucubrating domestic economic policies.

As a regulation, HICP covers non merely the pecuniary ingestion of the population, but besides the national CPI, excepting the undermentioned: auto-consumption, expenditures for investing and accretion, the involvement paid on loans, insurance rates, mulcts, revenue enhancements, etc. , and disbursals incurred by agricultural production of single families. In both instances, the weighting strategies used in twelvemonth T are based chiefly on the consequences of the Household Budget Survey conducted in twelvemonth t-2 and are updated each twelvemonth. For the computation of the HICP, the strategies are filled in with information on the ingestion of foreign tourers in Romania provided by The National Institute of Tourism Research and Development. Additional information on the costs incurred by the Rumanian citizens outside the state which is needed to cipher the CPI is obtained from balance of payments ( Balance of services ) .

Sing the method, both the HICP and CPI are Laspeyres indices with fixed base, although there are differences in their weighting strategies. In general, the differences between the CPI -s calculated in conformance with the national definition and those between the CPI and HICP specific to a certain state are generated by the undermentioned factors:

The update of the composing and construction of the ingestion basket used to cipher the CPI in conformance with the national definition and the basket of HICP can be done at different periods of clip. An advantage of this periodic update ( e.g. yearly ) of HICP basket and weights is the fact that in this manner the index reflects present inclinations of ingestion. HICP is a Laspeyres index with fixed base that allows but does non enforce an one-year update of the weights of goods and services in ingestion basket. Annually, it is necessary to look into HICP weights representatively to let for alterations if the weights in usage have a negative influence on the relevancy or comparison of the monetary value indices.

Some national Institutes of Statistics usage for the most elaborate degree of collection, different expressions when ciphering the CPI ( harmonizing to the national definition ) /HICP.

In the instance of CPI harmonizing to the national definition, the scope of wellness, societal protection and instruction services differs significantly from one state to another. For illustration, the existent disbursals of consumers with medical and instruction services are non ever taken into history in CPI rising prices measurement.The harmonisation of the attacks to these costs used in the computation of HICP was a major accomplishment. There are important differences in the statistical attack to the alteration in the quality of merchandises and besides in the attack to the debut of new merchandises in ingestion basket, seasonal merchandises etc.

HICP s comparison has increased by Eurostat and the National Institutes of Statistics activity, but attempts are still necessary in this regard.

It should besides be mentioned the fact that, since the national institutes of Statisticss have often adopted consonant policies for the computation of HICP and that of CPI ( harmonizing to the national definition ) , even the CPI comparison has improsed to a certain extent as a consequence to the concerns for CPI harmonisation.

Comparison between Indicators based on Registered Values

In order to turn out the absence of important differences between indexs ‘ values ( HICP and CPI ) we use the values registered in Romania at the terminal of each twelvemonth ( last month of the twelvemonth ) . Reference period is the same month of the old twelvemonth. There are registered values from 2002 to 2009 ( table 1 ) .

Table 1 – Annual rising prices rate ( % )


Annual rising prices rate ( % ) calculated by:


Consumer price index

December 2002



December 2003



December 2004



December 2005



December 2006



December 2007



December 2008



December 2009



Beginning: Eurostat, National Institute of Statisticss

Empirical ( Intuitive ) Comparative Analysis

This sort of analysis is based instead on an intuitive perceptual experience of the economic world, but it is sustained through rational statements. The analysis is grounded on logical logical thinking that might be simplistic, but non deprived of public-service corporation. The analysis does n’t necessitate much clip and the rational attempt is low.

The alleged empirical or intuitive analysis applied to this peculiar instance involves a bi-dimensional analysis for the same minute of clip. On the one manus, we have a transversal or perpendicular analysis ( based on the rising prices rates [ calculated utilizing HICP and CPI ] in the passage from one period to another, from one twelvemonth to another, in this instance set uping the being of the same tendency ( go uping /descending ) for both types of rates ) . On the other manus, there is the longitudinal or horizontal analysis ( for each clip period we compare the two types of rising prices rates, which have similar values ) .

Graphic Representation

The graphical representation offers an easy manner to set up the determinations. the chart below, the curves that represent the parametric quantities ‘ values about overlap, so there are non major differences between HICP and CPI.

Fig 1 Annual rising prices rate

Economic Analysis

This clip the analysis is based on computation methods and processs that are recognized in the economic theory therefore restricting the grade of subjective reading, which may happen in old analysis.

Between the CPI calculated in conformance with the national definition and HICP there are some conceptual differences, but their influence on the concluding consequence is marginal-less than 0.1 per centum for one-year rates in the period 2002-2009. The ground for this fact is the undistinguished weight of the disbursals of foreign tourers in Romania and of the Rumanian citizens abroad. An account for this might be the deficient statistical information related to the two constituents.

Statistical Analysis

The highest grade of objectiveness is provided by using strict scientific methods from statistical theory. In this instance we applied statistical trials and Chen-Sargent methodological analysis used to compare the tendencies of clip series.

a. Classical Statistical Trials

T-TEST ( Student trial ) Processing the information in SPSS, by using the trial of difference between the agencies calculated on the same sample, we found no statistically important difference. In SPSS, this is achieved by Paired-Samples T-test option from Analyze bill of fare. Null hypothesis provinces that there is no important difference between the values of the two monetary value indices, degree of significance being 0.05.

Table 2 Student Trial

Paired Samples Correlations




Pair 1





Paired Samples Statisticss



Std. Deviation

Std. Error Mean

Pair 1






Consumer price index





Paired Samples Correlations




Pair 1





Since the degree of significance exceeds the bound of 0.05, void hypothesis is accepted. So, the values of the two monetary value indices do non differ statistically in the analyzed period.

CHI-SQUARE TEST Sig is greater than 0.05, and chi-square statistics value is less than the tabulated value matching to 8 grades of freedom. These observations lead to the same decision: there are no important differences between the values of the two monetary value indices for the period 2002-2009.

Table 3 Chi Square Test

Trial Statisticss









Asymp. Sig.



WILCOXON TEST ( a non-parametric trial ) Non-parametric trials are applied to the quantitative or qualitative informations when we do non cognize the form of the distribution. Wilcoxon trial for two samples is an alternate for independent samples t test. Differences between the consequences are determined by utilizing Wilcoxon trial, and they are ordered by rank. Sig is 0.362, which is higher than 0.05, so, void hypothesis is accepted: there are no important differences between HICP and CPI.

ROMANOWSKI TEST Romanowski trial ( t trial ) presupposes an riddance method that consists in comparing the selected value with the mean corresponding to the other n-1 values in the scope of consequences. This trial is applied for a figure of values of the scope less than 100. It is besides necessary to cipher the standard divergence, neglecting, and the selected value ( n-1 values ) . Then, we determine the ratio:

( 1 )

The consequence of this ratio is compared with the critical values established for a proposed hazard. If t exceeds the critical T value: T a‰? T critical, can be safely removed with a grade of safety that peers at least the one proposed. Otherwise, there are non sufficient grounds to extinguish the valueAfter using the trial, with a chance of 95 % , we have concluded that the first two values in each row of Numberss should be eliminated, because the deliberate value exceeds the critical value, which is 1.24. Therefore, the CPI ‘s and the HICP ‘s values for 2002 and 2003 are considered outliers for analyzed series. The statistical trials antecedently used for the original series were so applied once more to the series from which we eliminate the outliers. The decision was the same: there are non important differences between the values of the two series.

Table 4 Wilconix Test

Test Statisticsb




Asymp. Sig. ( 2-tailed )


a. Based on positive ranks.

b. Wilcoxon Signed Ranks Test



Mean Rank

Sum of Ranks


Negative Ranks




Positive Ranks








a. IPC & lt ; IAPC

b. IPC & gt ; IAPC


Test Statisticsb




Asymp. Sig. ( 2-tailed )


a. Based on positive ranks.

b. Wilcoxon Signed Ranks Test

b. Chen-Sargent Methodology used to Compare the Trends of Time Series

Methodological alterations encountered in economic pattern led to alterations in the absolute dimension of the indexs, but besides to alterations in their tendency. The illustrations are legion and we can retrieve the passage from the ’77 version of SEC to those from ’95 and ’99, which imposed alterations in the computation of macroeconomic sums replacings occur in order to update the weights used in ciphering the consumer monetary value index and the industrial merchandise monetary values index. The alteration in the categorization of economic activities leads to alterations in the volume and construction of activity in assorted economic subdivisions, but besides in the population employed in these activities. If the legal relationship between employer and employee alterations the volume of employees in some economic subdivisions alterations, excessively.

In these instances, the re-composition of the clip series by linking the sections based on different methodological analysiss may alter the tendency and the significance of this alteration of tendency should be determined. tendency is an indispensable constituent, enabler in the development in clip of an index, as a series. In the linear hypothesis of constituents, the tendency is written:

( 2 )

– The tendency constituent

– The cyclical constituent

The seasonal constituent

The random constituent

We consider two clip series:

and ( 3 )

where, , = , = are appraisals for theoretical values and

In fact, to compare the tendency of the two lines depicting the two clip series means to compare the angular coefficients sing that, , are appraisals of the same angular coefficient I? = , so the coefficients differ merely by opportunity. Assuming that changeless footings of the two lines are equal = & gt ; which is the angular coefficient of consecutive line go throughing through the points ( ) and ( ,

In orderto trial the equality of the two coefficients we can utilize a Student trial like the one below:

( 3 ) ,

which follows the Student jurisprudence with grades of freedom. This value is compared with the tabulated value for a pre-chosen degree of significance ( I± ) . If the deliberate value is lower than the theoretical value we can state, with a chance of P = 1 – I± that the two coefficients are non significantly different and, hence, the two chronological heterosexuals can be considered appraisals of the same theoretical line. In pattern, we build assurance intervals for clip series parametric quantities, which suppose the finding of the asymptotic distributions for the selected parametric quantity. This procedure is based on the belongingss of the Brownian gesture.

In clip series analysis we use the construct called “ – commixture stochastic procedure ” ( phi-mixing ) , for which the far hereafter is independent from the present or from the yesteryear. If ( is a discrete-time stochastic procedure ; this is a- blending stochastic procedure if:

an event whose realisation depends merely on

an event whose realisation depends merely on the undermentioned status is accomplished: is a sequence of existent Numberss so that

Schruben shows that even a stronger status may be imposed for:

( 4 )

In general, – blending stochastic procedure category includes most of the types of stationary procedures, peculiarly ARMA type processes. Some techniques used for constructing assurance intervals for the parametric quantities of a clip series are based on the simulation of the discrete-time stochastic procedure. show that this attack becomes more efficient if we use a standardised clip series.

Let be a sample of observations on a stationary – blending stochastic procedure so that the status is accomplished divide this sample into b separate sub-samples.

Each of the sub-sample size is m. Let and be theaverage of the first J observations in one sub-sample. Let be the choice norm for the full sample. If we achieve the transmutation


We define

( 4 )

where with.

If, so the series defined by s the standardised series of one sub-sample.

Schruben showed that converges in chance to a Brownian gesture on the interval [ 0,1 ] . Using this belongings Chen and Sargent built a assurance interval for the difference of two clip series means. This assurance interval is more efficient than the one built utilizing conventional methods.

We will use the Chen-Sargent methodological analysis to look into if there is a important difference between the tendency of consumer monetary value index series and the one of consonant index of monetary values series for the period 2002-2009 ( the series are and while difference series is where. In fact, we determine whether there is a important difference between the agencies of the two series or non.

Before using the specified methodological analysis we check whether the two samples are likely to hold come from populations with equal discrepancies. In this regard, we apply the t-test for equal discrepancies, based on distribution belongingss. This status is satisfied by both clip series. Hereinafter, we have:

B = 1 = & gt ;

The average difference of the two series: = 0,045061 ( the norm was calculated as a geometric mean )

Assurance interval for agencies difference harmonizing to the methodological analysis Chen-Sargent is, in our instance, m = 8

= 0,010215

The consequences obtained for the two degrees of significance are the undermentioned:

For I± = 0.05 = & gt ; 0,003573

For I± = 0.01 = & gt ; CI=0.0450610,004645

The assurance intervals for the series without outliers were:

For both of the degrees of significance, in the original series every bit good as the one without outliers ) , there is no important difference between the two time-series tendencies. Performing the classical trial for the equality of agencies we arrive at the same decision, but with a lower grade of truth.


If the analysis is made from a right position, the differences between the two indices are undistinguished ; every index, including the one of consumer monetary values development, and the rate associated to it should be analyzed harmonizing to the mention base. Therefore, because of the different computation bases, the two indices ( CPI and HICP ) do non bear comparing without careful interpretationWhen calculated utilizing the CPI footing, the consonant index reached an about similar value. The same thing happens if the CPI is calculated utilizing the HICP ‘s mention base.

Making a statistical analysis we really demonstrated that the values of the indices used to mensurate rising prices for the terminal of the twelvemonth, in the period 2002-2009 did non differ significantly. But, the intent for which these indices are determined is the 1 that differs: HICP is used for comparing the rates of rising prices in Member States and for set uping the Central European Bank pecuniary policy, whereas CPI is used for lucubrating different economic policies.