Tracking changes in digital cameras

Introduction

The Consumers Price Index (CPI) basket of representative goods and services is reviewed approximately once every three years to ensure it continues to reflect household purchases. This involves reviewing the composition and relative importance of goods and services in the basket. The last CPI review was undertaken in 2006 and was implemented at the September 2006 quarter.

During a periodic review, the pricing specifications of the goods and services surveyed at retail outlets are also reviewed to ensure that varieties and sizes surveyed are representative of household purchases. Some pricing specifications may become out of date between periodic reviews. For these goods, it is important to ensure that the specifications remain representative between periodic reviews, where possible. The risk of an unrepresentative sample is that it may show different price changes to a representative sample that reflects the mix of products being commonly purchased. Digital cameras are a good example, as they have been subject to rapid technological changes. This article explains the CPI digital camera collection practices and examines whether the sample of products being tracked for the CPI has remained representative of household purchases.

top

Digital cameras in the CPI

Analogue 35mm film cameras were part of the CPI basket up until the June 2005 quarter. They were replaced in the September 2005 quarter by digital cameras. Digital cameras were brought in between the scheduled CPI basket reviews in 2002 and 2006 for two main reasons: the need to reflect a significant shift by households from purchasing film cameras to digital cameras, and difficulties in collecting prices for film cameras from some retail outlets in the sample.

At the time of introduction, a range of digital camera makes and models was selected for pricing from retail outlets in the 15 urban areas surveyed for the CPI. This selection was mainly based on makes and models representative of sales at surveyed retail outlets, with some specific requirements such as having a minimum level of optical zoom.

One of the most important pricing determining characteristics of digital cameras is the number of effective megapixels. This figure provides a measure of how much fine detail a camera can capture. The number of megapixels in an image can be determined by multiplying the width by height (in pixels). So a 3,000 by 2,000 pixel image has 6,000,000 pixels, or six megapixels. At the September 2005 quarter, the most common megapixel range in the CPI sample when first selected was '4 and over but under 5'. Since digital cameras were introduced into the CPI, camera technology has evolved considerably. For example, digital cameras of at least 10 megapixels are now widely available.

top

Keeping the sample representative

The digital camera sample is maintained through the periodic review process and ongoing maintenance of the sample.

Sample reselection at a periodic review

The outlet and digital camera product samples were reselected at the 2006 CPI Review. The pricing specifications for digital cameras were reviewed to ensure that surveyed models remained representative of household purchases. This work was informed by retail transaction data and feedback from CPI price collectors. Summary information collated from retail transaction data was used to determine representative specifications and to provide price collectors with information about the market shares of the main brands.

Prices for digital cameras are currently surveyed in approximately 80 retail outlets across the 15 CPI pricing centres. The sample covers three retail outlet types: appliance stores, department stores and camera/photo shops.

Ongoing sample maintenance

This involves ensuring that when a camera model being surveyed is replaced, it is replaced with an up-to-date model that is representative of the market. Decisions on replacements are informed by advice given to price collectors by retail staff, the observations and judgement of price collectors, and by retail transaction data available to office-based compilers of the CPI.

The retail transaction data used to inform sample maintenance decisions is acquired from GfK, a market research company. GfK collects details of sales of a range of small and large appliances (including digital cameras) sold through most of New Zealand's main appliance retailers and department stores. This data, broken down by quarters, is currently received annually by Statistics New Zealand. It should be noted that the GfK data will include business purchases (such as those by professional photographers) through retail outlets and spending by overseas tourists, which are both out of scope of the CPI (which covers purchases by private New Zealand households).

top

Comparisons of CPI sample and retail transaction data

The following analysis compares the sample of cameras surveyed in the CPI with the GfK data (hereafter referred to as the retail transaction data). The purpose is to examine the effect of the 2006 CPI Review and the ongoing sample maintenance on the representativeness of the CPI sample.

The analysis compares the distribution of digital camera megapixels in the CPI sample with that of the retail transaction data. Four periods are examined, as shown in table 1:

Table 1
Time line of digital camera events

 

Figure 1 shows the distribution of megapixel ranges at the September 2005 quarter, when digital cameras were introduced to the CPI basket. The three most common megapixel ranges surveyed in the CPI were '4 and over but under 5', '5 and over but under 6' and ‘under 4’, respectively. This is consistent with the distribution shown by the retail transaction data. The CPI sample aligns well with the retail transaction data, except for 'under 4' being over-represented and the higher pixel ranges appearing to being under-represented. As at the September 2005 quarter, no cameras of at least six megapixels were in the CPI sample, whereas they accounted for 12.3 percent in the retail transaction data.

Figure 1

Note: The distribution for the retail transaction data was calculated based on quantities sold, to be consistent with the quantity-weighted elementary aggregate formula used in the CPI at the time.

As previously outlined, the CPI sample was reselected as part of the periodic CPI review in 2006. The June 2006 quarter was the first quarter in which the new sample was surveyed. Comparisons of the CPI sample and the retail transaction data at this period are shown in Figure 2. The three most common megapixel ranges surveyed in the CPI were '6 and over but under 7', '5 and over but under 6' and ‘4 and over but under 5', respectively. However, the '4 and over but under 5' category is only the fourth largest in the retail transaction data. It appears that fewer cameras in the highest pixel range (that is, the '8 and over' category) were surveyed than called for. However, it should be noted that the retail transaction data will include business purchases (which are out of scope of the CPI) through retail outlets and these may tend to be in the higher pixel ranges.

Figure 2

Note: The distribution for the retail transaction data was calculated based on the value of goods sold, to be consistent with the expenditure-weighted elementary aggregate formula used in the CPI at the time.

Figure 3 shows the distribution of megapixel ranges at the June 2007 quarter. It is clear that the pixel ranges have moved up considerably compared with the June 2006 quarter, with '7 and over but under 8' being the largest category in both sets of data. The CPI sample would have been affected by decisions made as part of the ongoing maintenance of the sample since the June 2006 quarter. The CPI sample appeared to have surveyed more cameras in the '6 and over but under 7' category than required. The '8 and over' category is again an area that appears to be under-represented.

Figure 3

Note: The distribution for the retail transaction data was calculated based on the value of goods sold, to be consistent with the expenditure-weighted elementary aggregate formula used in the CPI at the time.

The final quarter for which Statistics NZ currently has retail transaction data is the June 2007 quarter. Therefore, comparisons of the latest CPI quarter, December 2007, could only be made with the retail transaction data for the June 2007 quarter. Figure 4 shows that the CPI sample now includes a much higher proportion of cameras in the '7 and over but under 8' and '8 and over' categories than was the case in the June 2007 quarter. In general, the CPI sample had moved ahead of the retail transaction data for two quarters earlier.

Figure 4

Note: The distribution for the retail transaction data was calculated based on the value of goods sold, to be consistent with the expenditure-weighted elementary aggregate formula currently being used in the CPI.

Table 2 shows the full distribution of pixel ranges in the four periods examined. It is apparent that digital camera sales moved up in pixel ranges considerably over time in both the CPI sample and the retail transaction data. For example, cameras with under six megapixels represented 100 percent of the CPI sample in the September 2005 quarter. By the June 2007 quarter, their share had fallen to 10.7 percent and by the December 2007 quarter, they accounted for only 2.4 percent of the CPI sample. By comparison, three in every five cameras (61.2 percent) being tracked for the CPI in the December 2007 quarter were in the '7 and over but under 8' megapixel range.

Table 2
Percentage distribution of megapixel ranges in the CPI sample and retail transaction data

top

Conclusion

The pixel distribution of the CPI sample has changed significantly over the two years since digital cameras were added to the basket. Over the period examined, the megapixel distribution of the CPI sample has aligned reasonably well with that shown by the retail transaction data, although there has been a tendency for the CPI sample to lag behind the market. This could be partly attributed to that fact that the retail transaction data will include business purchases through retail outlets and these may tend to be in the higher pixel ranges.

This article also illustrates the importance of keeping the sample current, especially given the pace at which digital cameras have evolved. For example, if Statistics NZ had continued to track only the price change of digital cameras in the initial sample from September 2005, the sample would have quickly become out of date and therefore unrepresentative.

Back to Price Index News: April 2008