This paper investigates the effectiveness of the methodology for calculating the Consumer Price Index (CPI) when the frequency of data collection differs. It is clear that when the prices of the various products (goods) which contribute to the shaping of the CPI on a monthly basis are used, we obtain a completely different picture of price developments, at least relative to the respective index when calculated on a daily basis. Given that the prices of most – if not all – goods are shaped on a daily basis, for the purpose of calculating the values of a Consumer Price Index, the choice of frequency of data collection would appear to play an important role in the compiling of accurate indices.
This is the main purpose of this paper. We shall look at the effects on the Consumer Price Index when the data used for its compilation is collected at different frequencies. Because such types of de-aggregated data are not available, we examined the impact of varying data collection frequency on the values of the Consumer Price Index using stochastic simulation techniques. We created prices for 60 goods, for 10 different urban areas and for five data collection stages in each urban area, and experimented by applying the CPI methodology 5,000 times at 30 different levels of data collection frequency.
The results of our experimentation confirm that the effects of the frequency of data collection are determinative for the numerical approach to the inflationary pressures on consumers, at least in the context of the methodology presently being used, in which the prices of most goods are collected on a monthly or, at best, fortnightly basis. Our results point to a decrease of 11% in the percentage change of the Consumer Price Index for a period of 365 days when there is only one data collection in a period of 30 days. Our conclusions clearly show the necessity for Consumer Price Indices, in the immediate future, to be calculated – if not on a weekly basis – then at least a fortnightly one.
Although numerous studies have been conducted on the effects of time aggregation and systematic sampling in many methodological approaches, we have not come across such applications for the Consumer Price Index. This means we are unable to make relevant references and comparisons which would augment the conclusions of this paper. It should also be noted that the effects of the frequency of data collection on the values of the CPI are not combined with other impacts on the Consumer Price Index such as Consumer Price Index Bias