Distributional Effects of Latest CPI Figures

A lot of the recent analysis of the CPI figures on this blog has examined  which households (in terms of poor versus rich) have benefitted most from deflation.  The latest CPI figures released today  (http://www.cso.ie/releasespublications/documents/services/current/rsi.pdf) has prompted me to publish some preliminary results from some work I have been doing which looks at this from a slightly different (though complementary) angle to that taken by Jennings, Lyons and Tol in their recent ESRI working paper.  I have mentioned before on this blog the idea of what is known as the distributional characteristic of a good (or aggregate of goods), which essentially summarises the extent to which consumption of the good is concentrated amongst lower income households.  By calculating this measure we can then see which price changes will have the most impact upon poor (or rich) households.  Calculation of the measure requires detailed knowledge of expenditure patterns across households and this data is available in the Household Budget Survey (and thus unfortunately only be calculated for the years the HBS is carried out).  Some analysis I have done looking at the 2004/2005 HBS suggests the following ranking of goods in terms of their distributional characteristic (a high ranking indicates a good whose consumption is more concentrated amongst poorer households):

1.  Tobacco and Fuel/Light (their values are practically identical)

3.  Food

4.  Non-durable Household Goods

5. Miscellaneous Household Goods

6.  Housing (including mortgage interest)

7.  Durable Household Goods

8. Alcohol

9.  Clothing and Footwear

10. Transport

11.  Services

Unfortunately the classification of goods into aggregates in the HBS tables (http://www.cso.ie/releasespublications/documents/housing/hbs.pdf) does not correspond exactly with that in the published CPI but for many goods it is very close, if not exact.

So, taking this approach, what have been the relative distributional effects of recent changes in the CPI (bearing in mind that all households will benefit from price falls)?  The July fall in clothing and footwear will give greater benefit to richer households relative to poorer ones, given that this consumption of this good is relatively more concentrated amongst richer households (this is true for the broad aggregate though of course may not be so for some individual clothing items).  The fall in fuel prices is definitely very good news for poorer households as this category consistently has the highest distributional characteristic (along with tobacco).  Housing is pretty much bang in the middle in the ranking so the effect of falls in mortgage interest payments is fairly neutral.

It should also be borne in mind that these figures are based upon the HBS from about five years ago but having looked at previous HBS the rankings don’t seem to change much.  I will publish the detailed results in a UCD working paper with more information about the methodology etc in the next couple of weeks.

4 replies on “Distributional Effects of Latest CPI Figures”

@David,

Many thanks for highlighting this useful research. As you indicate, the latest CSO data indicates that prices for Electricity, Gas and Other Fuels (which presumably corresponds to your Fuel/Light category) have fallen by over 10% in the last 12 months – which should be good news for poorer families. But disaggregating the fuels indicates the following:
Electricity +4.7%
Natural Gas +6.5%
Bottled Gas + 8.2%
Liquid Fuels -46.1%
Solid Fuels +6.3%

Gven the likely limited use of liquid fuels in poorer households, I’m not sure your provisional conclusion stands up.

While the impliciations for each income decile are interesting, I suggest that it would also be useful, if possible, to identify implications for specific groups to whom social welfare payments are substantial and under review, e.g. children, pensioners, unemployed.

Paul, you are quite right that the figures I have calculated are for broad aggregate groups and it is possible that individual categories within these aggregates will have different distributional characteristics – hence you could calculate the separate characteristic for liquid fuels and it might weel differ from that for the broad aggregate of fuel and light. But given the hundreds (thousands?) of goods in the HBS it is usually the practice to only make calculations for broad aggregates.

Antoin, the approach you suggest is pretty much the one adopted in the recent ESRI working paper. The approach I take is to look at specific goods (or aggregates of goods) and calculate the ratio of the weighted average to the unweighted average (where poorer households get higher weights). It is straightforward to come up with different weights for households based upon their income but much trickier to do so on the basis of a demographic characteristic since it involves coming up with an answer to questions like: how much more should society value the welfare of children versus, say, pensioners (or the unemployed, or single parents etc).

David,

Many thanks for your response. I agree that the HBS has to deal in broad aggregates, but, perhaps, in the design of future HBSs – and in the context of the share of household budgets (and, particularly, that of lower income hosuehold budgets) allocated to energy – it might make sense to move on from the archaic Fuel/Light category and to examine the expenditure on the major fuels.

I get uneasy when analysis of this nature may be employed to inform Government policy decisions and, particularly, when the share of household budgets allocated to energy is higher that it might be due to the levying of unjustified implicit taxes on major fuels.

Comments are closed.