Public Sector Economics

374
Views



51
Downloads

A cost-of-living squeeze? Distributional implications of rising inflation



Orsetta Causa*
   
Emilia Soldani*
   
Nhung Luu*
   
Chiara Soriolo*
Article   |   Year:  2023   |   Pages:  431 - 460   |   Volume:  47   |   Issue:  4
Received:  March 14, 2023   |   Accepted:  September 11, 2023   |   Published online:  December 11, 2023
Download citation        https://doi.org/10.3326/pse.47.4.3       


 

Abstract


In most OECD countries inflation sharply increased since the end of 2021, mostly driven by energy and food prices. Certain categories of households are particularly vulnerable, as they spend large portions of their consumption on energy and food. Drawing on national micro-based household budget survey data, this paper quantifies the impact of rising prices on households’ welfare. Declines in household purchasing power between August 2021 and August 2022 are estimated to range from 3% in Japan to 18% in Czechia. This decline is driven by energy prices in most countries, especially Denmark, Italy, and the United Kingdom. In general, inflation weighs relatively more on lower income households. The effects are stronger for rural households, due to energy price inflation. These findings call for a careful targeting of income and price support measures, notwithstanding their administrative and logistical complexity, taking into account their effects on economic activity, inflation, and environmental goals.

Keywords:  inflation; purchasing power; distribution; inequality; energy; policy analysis

JEL:  H12, H23, I3, Q41, Q48


1 Introduction


Inflation in the OECD has been on the rise since the start of 2021, intensifying sharply following Russia’s war of aggression against Ukraine. Soaring energy and, to a lesser extent, food prices are a global phenomenon, yet the inflationary picture differs across countries (figure 1). 

Figure 1
Consumer price inflation across OECD countries, change between August 2021 and August 2022 (%)
DISPLAY Figure

This paper looks at the distributional impact of rising inflation1 over the past year, with a focus on energy and, to a lesser extent, food price inflation, for ten OECD countries selected on the basis of adequate and timely data availability, i.e., the Czech Republic, Denmark, Germany, France, Italy, Japan, Mexico, Spain, the United Kingdom and the United States. The exercise draws on national microbased household budget surveys (HBS) providing information on the structure of household spending across the distribution of households based on income, age, and area of residence, depending on data availability. This paper associates expenditure shares from national HBS with price changes from national CPIs.

This work informs the policy debate by identifying households more exposed and vulnerable to the recent rise in inflation and in particular to changes in energy prices, which in the future might also fluctuate as we move along the path to a low-carbon economy (IEA,  2021). The main findings can be summarised as follows:
  • Rising prices, especially for energy, have been squeezing households’ purchasing power, but with large differences across countries, partly reflecting differences in the rate of inflation, its breadth across consumer items and the spending structure of the average household. 
  • Low-income, rural and senior households are more exposed to rising energy prices than the average household, but the variation in purchasing power losses across these three vulnerable groups is highly heterogeneous across countries. 
  • The difference in estimated energy prices effects between rural and urban areas are even larger than between lower and higher income households in most countries, and especially in Czechia, Spain, and France.
  • Non-energy non-food-price inflation tends to be progressive, thus somewhat mitigating, but not offsetting the effect of rising energy and food prices. This reflects the fact that “other” rising consumer prices correspond to items representing a higher share of spending for more affluent households, i.e., nonenergy transport, recreation, restaurants and hotels.
  • Improving the timeliness and granularity of the data would help in the designing of well-targeted policy support: for instance, to identify households most exposed to shifts in energy prices, e.g. those with limited financial resources and possibilities to substitute in the short-run. The digital transformation is the opportunity to build agile targeting instruments based on data collection and management.
The rest of this paper is structured as follows. Section 1 provides an overview of the data and empirical approach – additional methodological aspects are provided in the annex. Section 2 delivers the core results of the analysis, i.e., it sheds light on the distributional effects of the recent rise in inflation, with a focus on energy and food driven price inflation across a selection of OECD countries. The emphasis is on differences across income groups, but alternative relevant dimensions are also explored. This evidence is followed by a short policy discussion on key policy challenges to achieve effective targeted support for households most vulnerable to energy price swings while pursuing decarbonisation objectives.



2 The approach


2.1 Data and country coverage


The analysis draws on household budget surveys, which are national surveys on households’ expenditure on goods and services. Countries covered are those for which data are available and are as up to date as possible. Harmonised Eurostat HBS data for European countries would have been well-suited for this exercise but are not up to date, only being updated every five years. The following ten countries can be covered, with HBS data available for the year 2020 (with the exception of France): the Czech Republic, Denmark, France, Germany, Italy, Japan, Mexico, Spain, UK, US.2

2.2 Methodology


Distributional effects of inflation are assessed based on household exposure to the change in the prices of the items that make up households’ consumption baskets. This is expressed in terms of change in purchasing power following the conceptual framework of the compensating variation approach (Deaton, 1989). The compensating variation (CV) measures how much expenditure can be decreased (increased) when consumer prices fall (rise) so that the utility level remains the same as before the price decrease (increase). For household i, the CV is measured relative to total household expenditure (Ci). This is a measure of the change in household purchasing power resulting from changes in consumer prices underlying inflation; that is, the price change in item k (indicated as ) weighted by the share of expenditure that is spent on item k (scik,), defined as , where pk and qik refer, respectively, to consumption item k’s price and to the quantity purchased by household i, and ci refers to the household’s total expenditure

(1)

This approach has been used to assess the distributional effects of consumption taxes (OECD/KIPF, 2014) and of trade-driven price changes (Luu et al., 2020; Porto, 2006). The change in purchasing power can be computed for the average household and by income or other socioeconomic groups defined by, e.g. age, education and urbanisation of the area of residence, depending on data availability. For the purpose of the current exercise, the analysis relies on semi-aggregated data provided by national sources, that is, expenditure shares by socioeconomic groups.

The CV for household i corresponds to the average of the percentage changes in prices across categories of expenditures, weighted by the household’s expenditure share on each category. This is close, but not equal, to the inflation rate, i.e., percentage change in CPI, for household i, which would be the percent change in the average of prices across categories of expenditures, weighted by the household’s expenditure shares. The CV approach is adopted here because it has a conceptually-grounded economic interpretation for the purpose of the current exercise, similar to previous papers on distributional aspects of consumer price changes. In addition, the CPI calculations rely on weights that are adjusted according to consumption estimates from national accounts, with country-specific frequencies and methodologies (OECD, 2023a). Given the focus on distributional effects and for internal consistency, the current exercise uses the original expenditures shares from domestic HBS surveys, including those for the average household.3

The analysis uses 2020 household expenditure shares to assess the impact of inflation over the past year,4 consistent with the standard CPI methodology: national CPIs are constructed as weighted averages of sub-indices covering different products in the consumption basket, using the total household expenditure shares of a base year, as weights. These weights are regularly updated5 and, in normal times, are very stable. 6

To introduce the exercise, figure 2 reports expenditure shares on energy and food across the household income distribution for the Czech Republic, Germany, Japan, Mexico, the United Kingdom, and the United States.7 The main insights are:
  • Households devote a significant share of their spending to food and energy, and this share declines with increasing income level. Their combined spending share ranges from more than 30 per cent at the bottom to less than 15 per cent at the top of the income distribution, across the advanced economies covered here, yet with large cross-country differences within that broad spectrum. 
  • Food is a major component of the consumption basket. Low-income households spend around 20 per cent on food in the advanced economies covered by this paper with the notable exception of the United States, where that share is around 10 per cent. In line with Engels’ law, the weight of food is larger in Mexico, where low-income households devote approximately half of their spending on food and high-income ones 30 per cent. 
  • Energy represents around 10 per cent of household spending in most of the countries covered and this share declines across the income distribution. Mexico stands out as energy represents a much higher share of spending and this share increases across the income distribution. The share of consumption spent on energy is highest in the Czech Republic and lowest in Japan and the United States. Such cross-country differences likely reflect differences both in relative prices and in consumption patterns.

Figure 2
Shares of expenditure on energy and food, by income group (%)
DISPLAY Figure

Differences in energy spending are more pronounced across place of residence than across households’ incomes. This is illustrated in panel A of figure 3 with France and the United States. For example in France, people living in rural areas devote around 12 per cent of their budget to energy, more than twice as much as people living in the Paris agglomeration. It is often the case that people living in metropolitan areas have diverse commuting and mobility options, including public transportation, walking, biking, and while people living in rural or non-central areas may have no other option than driving a personal car (for example to go to work or the doctor). Differences in energy spending by age are less systematic across countries than they are by income and place of residence: taking again France and the United States as examples, panel B of figure 3 shows that the share of spending devoted to energy increases almost monotonically with age in France, but not in the United States.8



3 The results


3.1 The big picture: the cost-of-living pressure for the average household


Rising prices, especially for energy, have been squeezing households’ purchasing power, yet with large differences reflecting differences in the rate of inflation, its breadth across consumer items, and the spending structure of the average household.9 The main findings on the basis of year-on-year August 2021 – August 2022 inflation are (figure 4):
  • Declines in households’ purchasing power range from around 3% in Japan to 18% in the Czech Republic.10
  • The effect of rising energy prices is large but differs across countries, being particularly important in Italy, Denmark and the United Kingdom. 
  • Rising food prices weigh less than rising energy prices on the purchasing power of the average household. Mexico is an exception given the high share of food in the consumption basket and the relatively mild increase in energy prices over the period. 
  • The effect of “non-food-non-energy” prices on the average household’s purchasing power is relatively higher in countries like the Czech Republic and the United States, reflecting the fact that inflation in these countries was more broad-based. 
  • These differences across countries are in part due to differences in the relative price changes and in part to differences in the consumption share allocated to food, energy and the residual category.
Figure 3
Expenditure shares on energy (%)
DISPLAY Figure

These results are subject to two main caveats: (1) the computations include the purchasing power effects of price-based policy interventions, e.g., reduced prices or taxes on electricity, gas and gasoline; they do not include the purchasing power effects of non-price-based policy interventions, e.g. income support via cash transfers or reduced income taxes. This might affect the cross-country comparison exercise, in light of differences in the adopted mitigation measures. In general, at least during 2022, price support measures introduced to contrast the cost-of-living crisis following Russia’s attack on Ukraine outnumbered and involved higher costs than income support measures (OECD, 2022). Evidence for France and the United States suggests that the combined effect of these measures has been effective in supporting the purchasing power of vulnerable households (Madec, Plane and Sampognaro, 2022; CBO, 2022). The current calculations can therefore be interpreted as measures of potential exposure to purchasing power losses, abstracting from the mitigation achieved through ad hoc income support measures. (2) energy effects are to be taken as a lower-bound, given that energy items are key intermediate inputs for the production of non-energy items, and therefore for the pass-through to non-energy consumer items (see Ari et al. (2022), and OECD (2023b) for a quantification exercise of direct and indirect effects from rising energy prices).11

Figure 4
Purchasing power changes for the average household (%)
DISPLAY Figure

3.1.1 Beyond the average household: distributional effects of the cost-of-living pressure


The effects of inflation are highly heterogeneous across households and distributional patterns differ across countries (figure 5). A comparison of low- and high-income households (defined as first and last decile, quintile, or country-specific threshold of the household income distribution, see annex) yields the following insights (figure 5, panel A).

Inflation has a greater impact on low than high-income households, but with marked differences across countries. Nevertheless, such differences do not appear to be strongly correlated with the level of inflation in the country. The gap between low and high-income households is the largest in the United Kingdom while it is almost absent in the Czech Republic and Denmark despite similar or even higher headline inflation over the period covered.

Energy price inflation is strongly regressive in all countries except Mexico, where it has a relatively higher effect on high-income households; this is consistent with the fact that in Mexico the share of spending on energy is positively associated with household income (figure 2).12

Food price inflation is also regressive but less so than energy price inflation in most countries covered. Mexico stands out, again, since food price inflation is the single major driver of regressivity. The regressivity of the effects of food price inflation is also more marked than that of energy price inflation in the Czech Republic and Spain.

The effect of non-food non-energy price inflation is progressive, i.e., affecting high- more than low-income households, which is why the total gap is lower than the sum of the energy and food gaps.

Living on limited income resources is not the only and often not the first factor of vulnerability to the current inflationary picture. Living in a small, isolated village is a major vulnerability factor. Inflation tends to disproportionately affect rural households and thus to amplify spatial inequalities. In most countries, the purchasing power gap between rural and metropolitan households tends to be larger than that between low and high-income households and this gap is driven by energy (figure 5, panel B). Age is another factor of vulnerability to energy and food price inflation, as indicated by the finding of larger purchasing power losses for senior relative to prime-aged households in all countries except Denmark and Spain (figure 5, panel C). But age-related gaps are generally lower than place of living- and income-related gaps.

Figure 5
Differences in purchasing power effects between various types of households (pp)
DISPLAY Figure

In summary, figure 6 reports the purchasing power loss associated with rising energy prices for the three vulnerable groups covered in the analysis, i.e., low-income, rural and senior households. Main insights are:
  • Low-income, rural and senior households are more exposed to rising energy prices than the average household but the variation in purchasing power losses across these three vulnerable groups is highly heterogeneous across countries. 
  • In most countries, rural households are more vulnerable than low-income households to energy price inflation, in particular in the Czech Republic, Spain, and France. This could in part be due to the relatively limited access to public transport and the need to drive higher mileages, but also to differences in energy efficiency of the primary dwellings. Disentangling such drivers is however beyond the scope of the current exercise, and would require access to data at a more granular level. 
  • Senior households tend to experience milder purchasing power losses than low-income and rural households, in particular in Italy and in the United States.
Figure 6
Purchasing power losses from energy price increases: low-income, rural, senior households (%)
DISPLAY Figure



4 Granular analysis of non-energy-non-food price inflation


The evidence in the baseline analysis is that inflation is regressive because energy and food price inflation is strongly regressive, while non-energy non-food price inflation tends to be progressive, thus somewhat mitigating (but not offsetting) the effect of rising energy and food prices. Such evidence is obtained by computing the average change in prices on food, energy, and “non-food non-energy” items, each weighted by the respective share on household expenditures. Given the importance of non-energy non-food purchasing power effects, an important question is what drives them.

A granular analysis, giving illustrative insights, is possible for France, Germany, Italy and the United Kingdom, because of the collection of more granular data.13 Detailed HBS and CPI data are combined to compute the compensating variation (CV) for each of the COICOP-99 one-digit expenditure categories, net of food (COICOP-99 category 1) and energy (COICOP-9 categories 4.5 and 7.2.2) expenditures. The sum of the resulting CVs is equal to the average of the change in prices of each one-digit category, weighted by the respective expenditure shares (exact formulas in the annex). The sum corresponds to the CV of the category “Other” in figure 4, up to a small approximation margin. Performing this exercise for each income bracket sheds light on the distributional effects of changes in major non-food non-energy price items. The results are presented in figure 7 and summarised below.

Average household effects (figure 7, panel A) 
  • Rising prices of recreation and restaurants and of non-energy categories of transport goods and services (e.g., buying a car or an airline ticket) are major drivers of purchasing power losses for the average household across the sample of countries considered. 
  • Rising prices of non-energy housing (i.e., mostly actual and, in some countries, imputed rentals) have a significant effect in Germany and, to a lesser extent, the United Kingdom. Rising prices of housing furniture add to purchasing power losses for the average household.

Distributional effects (figure 7, panel B) 
  • The effect of rising prices of recreation, restaurants, furnishing and nonenergy transport is progressive, i.e., the price increase affects higher income households more than lower income ones. 
  • The progressivity of non-energy transport price inflation is particularly marked for Germany. This reflects the significantly large share of expenditures on non-energy transport, especially on the purchase of vehicles, among German high-income households (over three times that of low-income households).
  • Non-energy housing price inflation is regressive i.e., affecting low more than high-income households, but the purchasing power difference between high-and low-income households is minor relative to all other spending categories; and furnishing also has a progressive effect.14
Figure 7
Non-energy non-food price inflation tends to affect high- more than low-income households
DISPLAY Figure

One key implication of this analysis is the potential importance of improving the consistency, granularity, and timeliness of the data, as a basis for research and all the more for policymaking. Reliable timely information on consumption patterns would allow to quantify the reactions of demand to price shifts and expectations. At the moment, detailed data by consumption category is published with a lag. For example, at the time of the analysis, in 2022, the latest harmonized HBS data released by Eurostat was for 2015.



5 Policy discussion


As outlined above, a quantification of the impact on household purchasing power of the price and income based support measures adopted by governments is beyond the scope of this paper. Nonetheless, the results showcase the stark heterogeneity in the exposure of different categories of households to food and energy price shocks. Such heterogeneity in turn suggests the use of targeted support measures. Well-designed income support can limit the burden on government budgets as they preserve price signals for energy savings while providing a financial lifeline to those who need it the most and, in the longer run, improve resilience to price swings while also facilitating a just transition towards a greener economy (OECD, 2022). With respect to effective targeting, our results underscore the need to consider further factors of vulnerability beyond income, notwithstanding the likely correlation among the different factors. One important dimension in this respect may be the area of residence and in particular whether it is rural, as shown in this paper. Yet targeting households living in rural areas is also likely to be crude, as e.g., affluent retirees in the countryside are less vulnerable to rising energy prices than younger people looking for a job in rural areas. Furthermore, other dimensions of vulnerability that could not be covered in the current analysis would also need to be considered, such as housing quality (e.g., energy efficiency) and access to infrastructure (e.g., public transport). The implication is that effective targeting may be logistically