THE AVERAGE SIZE AND PROPORTION OF PRICE CHANGES IN THE CPI

By Maxwell Sutton, Research Economist, Reserve Bank of Australia (on secondment to the ABS)


INTRODUCTION

This article seeks to understand how consumer inflation has evolved since 2000 and to get a better understanding of the inflation process and price-setting behaviour in Australia. The drivers of periods of higher or lower inflation may vary for many reasons including changes in the price sensitivity of consumers, variations in competitive pressures and movements in the exchange rate.

Using Consumer Price Index (CPI) micro data collected by the Australian Bureau of Statistics (ABS), inflation can be decomposed into the proportion and average size of price changes (see Appendix 1 for details). The ‘proportion’ is the weight of the CPI basket whose prices change in a given quarter, while ‘average size’ is the weighted-average size of price movements for those prices that did change. This decomposition can show if periods of higher (lower) inflation are due to: a higher (lower) proportion of prices increasing; larger (smaller) changes in prices; or a combination of both.

Price changes were calculated by comparing the last observed price in the current quarter to the last observed price in the previous quarter. A limited number of price series were excluded from the analysis(footnote 1) . The graphs shown include data up to June 2017.

This work is the result of collaboration between the ABS and the Reserve Bank of Australia (RBA).


ANALYSIS

Headline
The average size of price changes has gradually declined since 2005 (Graph 1). This has resulted in a gradual disinflation in consumer prices and may reflect weaker demand or cost pressures and, in some instances, more intense competition. On average, under half of the CPI basket experiences a price change every quarter. This implies that the average length of time between price changes is around two quarters.

Variations in inflation are generally dominated by changes in the average size of price movements; average size accounts for around 85 per cent of the variation in inflation. This is consistent with evidence found for the United States by Klenow and Kryvstov (2008). The disinflation in prices during 2014 and 2015 and the increase in headline inflation in 2016 predominantly reflect movements in the average size of price changes.

The increase in the proportion of prices changing from mid-2004 until the end of 2009 amplified the movements in inflation during the pre-crisis period. This indicates that this period of higher inflation was partially due to a broader increase in prices as a higher share of firms responded to growing demand. The proportion of prices changing remained high during 2009 as firms adjusted their prices in reaction to the global financial crisis. The higher proportion but lower average size of price changes since 2014 may suggest that more firms have been responding to evolving economic conditions.

The high proportion and average size of price changes in 2000¬–2001 is due to the implementation of A New Tax System (the policy included the introduction of the Goods and Services Tax), which caused a one-time change in the prices of many goods and services.

The average size and proportion of price changes can be further decomposed by separating price increases and decreases (Graph 2). Each quarter, one third of the CPI basket experiences a price increase while around one sixth of the CPI basket experiences a price reduction. When prices increase, they increase by around 7–8 per cent while prices that are reduced tend to fall by around 11 per cent.

The average size of price increases and decreases have both declined since 2012, which has resulted in a smaller variance of price changes. This is often the case during periods of low inflation.



Examining tradable and non-tradable goods and services can provide some insight into the effect of the exchange rate on inflation. This is because movements in the exchange rate reflect a change in the marginal cost of imported goods and services, which should also affect the price of those domestically-produced goods and services that compete with these imports.


Non-tradable goods and services

The weakness in inflation since 2013 has been largely due to a reduction in the average size of changes in the prices of non-tradable goods and services (Graph 3). Low wage inflation measured by the wage price index and average earnings per hour from the National Accounts is both a cause and a symptom of these lower price pressures (see the feature article The Size and Frequency of Wage Changes (Bishop 2016) for analysis of wages). The prices of non-tradable goods and services are currently changing more frequently than average, but prices are being adjusted by smaller amounts.



Tradable goods and services

The average size of price changes in tradable goods and services has generally been lower than in non-tradable goods and services while the proportion of prices changing has generally been higher (Graph 4). This may be because the prices for tradable goods and services are more exposed to fluctuations in the exchange rate, so firms expect their marginal costs to change more frequently and may therefore be more responsive to these variations. Import prices have also exerted some downward pressure on tradeable goods and services inflation, which might reflect greater global competitive influences. Exposure to exchange rate variations is likely to be a strong contributing factor to the higher volatility in average size of price changes and higher proportion of prices changing compared with non-tradable goods and services.


The proportion of prices of tradable goods and services changing has broadly moved with the exchange rate (Graph 5). A smaller proportion of tradable goods and services changed price as the Australian dollar appreciated from 2001. The sustained depreciation since 2014 has coincided with an increase in the proportion of prices changing. This suggests that the prices of tradable goods and services are more sensitive during periods of depreciations of the exchange rate, when costs are rising, than during periods of appreciations.




Rents

The pickup in rent inflation from 2003 to 2008 was caused by an increasing proportion of rents changing, while those rents that were changing were being changed by ever higher rates (Graph 6). This period of rent inflation has been attributed to the low vacancy rates recorded during that period. The low proportion of rents changing in the early 2000s is in need of further exploration.

The disinflation in rents since late 2011 has been caused solely by a gradual decline in the average size of price changes. The proportion of rents changing each quarter has gradually increased to around 24 per cent (so rents have been changing every 13 months on average). However, this is due to a greater proportion of rents falling in price (Graph 7). These movements may be due to landlords facing weaker demand in some capital cities and, more recently, an increase in supply in other capital cities.




Food

Inflation in food prices has been lower on average since 2013 (Graph 8). This is because the average size of price increases has fallen (Graph 9). The effect on inflation has been muted by a simultaneous decline in the average size of price reductions. Firms are reducing their prices more frequently but at lower rates. This suggests that firms may be facing a combination of competitive pressures, leading them to keeping their prices lower than they would have in the absence of these pressures, and higher costs. Currently, increases in the prices of some foods are being largely offset by reductions in the prices of other foods, so that the total cost of food is increasing at a slower rate compared with the period prior to 2013.

Separating the average size of price increases and decreases also shows the effect of Cyclones Larry and Yasi on the prices of food, and bananas in particular. Prices of bananas increased significantly in 2006 and 2011 causing the average size of price increases in food to grow. As the supply of bananas improved over the following years, banana prices fell, which is evident in the peaks of the average size of price decreases in 2007 and 2012.




Clothing

Clothing prices have decreased by 0.7 per cent each quarter, on average (Graph 10). However, this is because the average size of price increases and decreases have increased simultaneously, from around 10 per cent in 2002 to around 25 per cent in 2017 (Graph 11). During this period, the proportion of prices changing has averaged around 35 per cent, which is split evenly between price increases and decreases. This pattern of offsetting price changes may reflect an increase in the seasonal variation of clothing and clothing prices, and the increasing presence of international retailers expanding the range of products available.





CONCLUSION

Variations in inflation are dominated by changes in the average size of price movements rather than fluctuations in the proportion of prices changing. The recent period of low inflation is associated with lower variation in price changes across goods (in absolute terms, prices are changing by less, whether they are increasing or decreasing). Low inflation in non-tradable goods and services may be a cause and a symptom of low wage pressures. Prices of tradable goods and services seem to be more sensitive to exchange rate fluctuations during periods of currency depreciations compared with periods of appreciations.


BIBLIOGRAPHY

Bishop J (2016), ‘The Size and Frequency of Wage Changes’, Cat. no. 6345.0.

Klenow PJ and O Kryvstov (2008), ‘State-dependent or time-dependent pricing: Does it matter for recent U.S. inflation?’, The Quarterly Journal of Economics, pp 863-904.


FOOTNOTE
1 The deposit and loan facilities expenditure class was excluded due to the extreme price movements. Large changes caused by goods and services which recorded a price of zero for one more periods were excluded. <back

 
APPENDIX 1

Headline inflation at a point in time can be expressed as the weighted-average inflation of each good (or service) in the CPI basket,

where is the percentage price change for good i in quarter t and is the proportion of the expenditure base that is spent on good i in quarter t . Using a simple identity, it is then possible to decompose headline inflation into two terms measuring the proportion and average size, respectively,



where is an indicator function that takes the value of one when its argument is true i.e. when good i experiences a price change in quarter t. The ‘proportion’ term measures the fraction of goods that experience a change in price in quarter t, adjusted for each good’s share of the expenditure basket , while the ‘average size’ term is the weighted-average size of price changes for those prices that did change.
The proportion is the sum of the proportions of prices rising and falling,



The average size of prices can be decomposed to separate the effects of price increases and decreases,