The results from a recent study on the proposed effects of factors like working experience and education suggest that the effects on the Icelandic Wage Price Index (IWPI) are small or 0.024% on average per month based on data from 2008 to 2018. The effects are not statistically significant and the average grows linear with the number of index chaining.

In the autumn of 2018 a review of the IWPI was conducted by an independent specialist following criticism of the IWPI. In the review it was concluded that IWPI is robust and its computations sound and based on well-known and well established methods as well as reliable data that sufficiently cover the population. It was recommended that Statistic Iceland should continue with its current survey and index calculation architecture, but explore the effects of experience and education on the IWPI. In the review it was also reasoned that the use of arithmetic average wage (unit value) cell indices are not recommended for the IWPI, as international recommendations favour quality adjusted index numbers over unit values.

In short, both the results from the review and from Statistic Iceland’s study show that the methods used to compute the IWPI do not need to be changed.

Criticism of the Icelandic Wage Price Index
The IWPI has been criticised, mainly because of the methods used, implying an overestimation of the IWPI due to matched (paired) comparisons. The method involves the measurement of change by comparing two observations of a fixed unit between two adjacent time points which is a common method in price indices to control for quality changes. The criticism was partly based on a comparison between the IPWI and average earnings although these measures have different purpose and use different methods rendering them incomparable. It was also pointed out that improvements were needed as regards descriptions of the IWPI and that research was lacking.

The publication of the IWPI started in 1989, based on the Act on the Wage Index No 89/1989 having its roots in the indexation of financial obligations. According to the law, the wage index is to show changes in wages for fixed working hours. In the explanatory notes to the Act, it is further stated that changes in – and composition of – working hours should not affect the wage index, except when included in collective agreements and subsequently equivalent to changes in wages. Consequently, Statistic Iceland’s interpretation has been that the wage index should be a price index. A harmonized price index for wages has not yet been established internationally and different approaches for measuring wage changes are used in the few countries where a wage price index exists. The main reasons are a lack of a clear legal basis as well as a lack of sufficiently detailed wage data. When computing the IWPI, Statistic Iceland adheres to the general methods of price indices. Therefore, it is reasonable to expect that the IWPI encompasses some of the general shortcomings and limitations known to exist in price indices.

The IWPI is one of many measures disseminated by Statistic Iceland that can be used to evaluate changes in wages. Which one to use is dependent on each user’s needs. Other measures are the Total Wage Index and changes in Average Earnings. Furthermore, changes in Total Income from Employment provide indication of changes in wages. Still another measure is the Labour Cost Index which shows changes in both wages and labour costs and is published on Eurostat’s web. The main differences between these measures are methodology, data coverage and different effects of changes in the composition of the labour force and worked hours.

The assessment confirms the quality of the Icelandic wage price index
At the beginning of the year 2018, Statistic Iceland began improving descriptions of the methods used to compute the IWPI and conducting research on the index. At the same time the emphasis on diversity of statistical information was increased. In August, 2018, Statistic Iceland published a report[1] in which the methods used were explained along with a description of the wage data IWPI is based on. In this same report, results from a study on errors that are known to exist in price indices methods (e.g. chaining drift and life-cycle errors) were published. The main results from this show that the methods used by Statistic Iceland are robust and in accordance with well-established methods used in the field of price indices. Furthermore, the reported study found that data are of high quality. Results provided an evidence suggesting that neither chaining drift nor life-cycle errors are problems for the IWPI.

In 2018, a committee appointed by the prime minister to suggest improvement in the use and computations of wage statistics employed an independent specialist in indices, Dr. Kim Zieschang, to thoroughly review the IWPI. The review was to encompass the whole computation process (i.e. methods, data, etc.) of the IWPI and suggest improvements if needed. In the review[2] it was concluded that the IWPI is robust, based on advanced and well established methods, using high quality data with sufficient coverage. It was recommended that Statistics Iceland should continue with its current survey and index calculation architecture but explore quality adjusting IWPI employee wage records for experience and education using the hedonic method. In this context quality refers to increment in time of an employee with the same employer and increment in education. Furthermore, in the report it was reasoned that the use of arithmetic average wage (unit value) cell indices is not recommended for the IWPI, as international recommendations favour quality adjusted index numbers over unit values, which is in alignment with the IWPI.

Based on the result of the review, the committee suggested that Statistic Iceland should investigate how the IWPI reflects the effects of increasing working experience and education. The committee also suggested that the proposed assessment should be done in accordance with Dr. Zieschang’s recommendations, results should be officially published and Statistic Iceland should take action if the assessment suggests significant labour quality effect inherent in the IWPI.

Statistic Iceland has henceforth responded with a study following the recommendations of the committee and including consultations with Dr. Zieschang. The main results from the study[2] are reported below.

Main results
Education and experience are examples of attributes of the labour force that have tendency to change over time. With respect to IWPI, it is important to note that the aforementioned changes are individual attributes but the IWPI is not based on the individual, per se, but on the contract (formal or informal) between the individual and his employer. The properties of the contract are the employee, the employer, economic activity and occupation. Changes of one, or more, of the contract’s attributes invalidates it. Analysis shows that less than a half of these contracts may still be found in the total sample after about three years. Changes in education and experience by time can reflect labour quality effect, however that is not always the case, e.g. when education is irrelevant to the occupation in question.

The most common methods used to adjust for the effects of quality changes in price indices involves using a matched sample or hedonic models (hedonic models are based on a regression). The computation of the IWPI is based on the matched sample method and changes in regular wages3 of each contract between two adjacent months measured. The matched sample method use in the IWPI does not measure the effect of changes in experience and education in time. Furthermore, reliable information about education and working experience have not been available until recently.

In the study, hedonic models known from the methodology of price indices were used to investigate the effects of changes in factors like working experience and education of employees on the IWPI. The models with and without the working experience and education can be compared to evaluate the influence of the factors. The main results are that comparison shows that the effects of the factors are not significant on monthly basis. The effects are not linear and the direction of them is not constant, i.e. in one month the factors can increase the IWPI while in another month they can decrease it. On monthly basis the quality factors induce average fluctuation of 0.024% (CI = [–0.2%, 0.3%])1. The average reaches a value of 2.9% after 10 years and its confidence interval still covers the value zero although they are much wider. The results are based on data of the IWPI from 2008 to 2018. For more detailed information about the study see: Effects of changes in working experience and education on the Icelandic Wage Index.[4]

An innovative study
Study on the effects of changes in working experience and education on the IWPI is an innovative study as far as Statistic Iceland knows. This means that the results of this study cannot be compared with results from other studies on this matter. It should be noted that the period the study covers is not typical for economic evolution in Iceland and therefore it is not possible to conclude that its results hold for other periods.

The results, both from the review and from Statistic Iceland’s studies shows that there is no need to change the methods used to compute the IWPI. Although no important changes are needed concerning the IWPI, Statistic Iceland is interested in investigating the possible benefits of more methods when computing a change in prices.

1The effect of the factors can be anywhere within the range of the CI.

1. Methodology of the Icelandic Wage Index
2. Review of the Icelandic Wage Price Index
3. Icelandic Wage Price Index, metadata
4. Effects of changes in working experience and education on the Icelandic Wage Index