Regional and income disparities in cost of living changes: evidence from Egypt

ABSTRACT

Inflation has been rising in Egypt since 2007, and reached record levels in 2017. It was more pronounced in rural Egypt and likely hurt the poor proportionately more, since rising food prices were a major factor behind higher prices over this period. Moreover, rising prices, as measured by the Consumer Price Index (CPI), do not accurately measure changes in the cost of living. When inflation is high, people resort to substitution to hedge themselves against a declining standard of living. Regardless of price changes, habit formation and taste changes can also render the fixed basket of the CPI less representative over time. To accurately monitor changes in the cost of attaining a given utility level, I constructed True Cost of Living Indices (TCLI) and used them to examine the regional and income disparities in cost of living changes, and the extent of the bias in the CPI. Results confirmed that cost of living increases were higher in rural regions, and that there were far larger regional disparities in cost of living increases over time using the TCLI. The bias in the CPI was quite substantial, ranging from underestimating the change in cost of living by 1.86 percentage points, to overestimating it by 1.05 percentage points, depending on region. Finally, I found strong evidence that households at the bottom of the expenditure distribution fared much worse. Depending on region, cost of living increases were 2.8 to 4.1 percentage points higher per year for the poorest urban households than for the richest, during the period under study.

KEYWORDS:

• Cost of living inequality
• Consumer Price Index
• substitution bias
• poor
• Egypt

JEL CLASSIFICATION:

• C43
• E31
• I31
• O18

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Throughout the paper, I use the expenditure levels of the household to determine their position along the distribution. To ensure this is clear I use the term ‘expenditure distribution’ and ‘expenditure inequality’ to avoid confusion where appropriate, and these should be understood to be analogous to ‘income distribution’ and ‘income inequality’.

2 These group specific prices were collected by the US Embassy in Cairo for a brief period during the 1990s to gauge differences in prices between ‘poor’, ‘middle class’ and ‘rich’ neighborhoods in Cairo (these classifications were arbitrarily made by Embassy staff and there was no clear documentation on the basis of the classification system). These were not nationally representative however, and the series has since been discontinued.

3 Diewert (1976) defined a price index as ‘superlative’ if it is exact for an expenditure function which is capable of providing a second-order differential approximation to an arbitrary twice continuously differentiable linearly homogeneous expenditure function. (The Laspeyres price index for example is not superlative because it is only a first-order Taylor approximation to the TCLI).

4 The following exposition follows Braithwait (1980).

5 The notion that the Konus-TCLI must lie between the Laspeyres and the Paasche price index, and therefore that the TCLI calculated in this paper should always show lower changes in the cost of living that the official CPI, only applies in the case of homothetic preferences of consumers such that all indifference curves are the same shape and expenditure shares are independent of expenditure levels. Such assumptions are unlikely to hold in reality since consumers do have variable preferences (see Deaton & Muellbauer, 1980, pp. 170–173). For example, Phlips and San-Ferrer (1975) calculate a Taste dependent TCLI and find that the Laspeyres index was not an upper bound to their constructed TCLI: cost of living changes for the US were 1 percentage point higher by their TCLI than the Laspeyres CPI. Similarly, Heien and Dunn (1985) estimate a true cost-of-living index based on a quadratic expenditure system estimated with quarterly data from 1960–1981 for the US. They found that the true cost-of-living index grew faster than both the Paasche or Laspeyres indices and also displayed more variability. More recently Lieu et al. (2008) estimate TCLIs for Taiwan allowing for variable preferences. They also computed non-parametric GFT based TCLIs and found that cost of living changes are underestimated when variable preferences are ignored: i.e. when the Laspeyres index was used it would appear that changes in the cost of living were lower than when the GFT-TCLIs were used (they performed this analysis for 4 different time periods between 1951 and 2001 and this finding holds for 3 of the 4 time periods). Intuitively, one could think of factors such as habit formation and changing preferences that may actually imply that consumers stick to a more expensive bundle even if its price increases, by lowering quantities consumed, or giving up on other items that they deem less important. For example, studies have found that as the poor’s incomes rise they tend to buy better tasting food even if the nutritional value is lower than some other less tasty more nutritious food item-see for example Banerjee and Duflo (2012, Chapter 2) for a detailed discussion.

6 The same analysis was repeated using income quintiles rather than expenditure and the results were almost identical and therefore not shown to avoid duplication.

Additional information

Funding

This work was sponsored by the Economic Research Forum (ERF) and has benefited from both financial and intellectual support. The contents and recommendations do not necessarily reflect ERF’s views.