Reassessing the econometric measurement of inequality and poverty: toward a cost-of-living approach

The paper examines how inequality and poverty are commonly measured in advanced capitalist economies and argues that dominant relative measures—the Gini Index for inequality and poverty lines defined as a proportion of median household income—are inadequate. Building on debates since the 1970s about absolute versus relative measures, the author contends that these relative metrics can mask material deprivation and social mobility, particularly during economic shocks. Using Hong Kong as a case, the study aims to demonstrate that a cost-of-living approach, which anchors the poverty line to the prices of essential goods and services, better captures actual needs and economic pressures than measures based on relative income distances from the median.

The study situates itself within longstanding debates on measuring inequality and poverty. Absolute poverty measures (e.g., US$1–$2/day thresholds) are criticized for ignoring social and economic context, as evidenced by issues in China’s dibao program (Kakwani et al., 2019). Relative measures, increasingly adopted for developed and middle-income contexts, are justified by arguments that needs are socially contextual and reference-group dependent (Brady, 2003, 2019). However, the Gini Index—though widely used—relies on relative deprivation and can be insensitive to changes in the distribution (Alvaredo & Piketty, 2014; Osberg, 2017), stagnating around asymptotes in high-inequality contexts. Debates also emphasize the inelasticity of key needs (e.g., housing) and rising housing costs (Albouy et al., 2016; Piketty, 2014), which relative measures may miss. The paper aligns with calls (Ravallion, 2020) for approaches that lie between absolute and purely relative methods, and with Sen’s capabilities perspective that emphasizes material needs.

Design and scope: A methodological reassessment using Hong Kong as a case study to compare relative, income-based measures of inequality and poverty with a cost-of-living approach. The analysis covers 1996–2020. Data sources: Hong Kong Census and Statistics Department (multiple series including household income distribution, CPI components); Legislative Council of Hong Kong (household income and Gini information). Comparative reference statistics from World Bank and other official sources. Measures and equations: - Inequality: Gini Index at aggregate level and decomposed by subgroups (e.g., household size). Aggregate Gini formulated as a scaled sum of pairwise income differences (relative deprivation). Subgroup decomposition compares within- and between-group differences, following standard Gini decompositions. - Relative poverty (benchmark): Poverty line defined as 50% of the median household income (OECD convention). Poverty rate P(X) is the proportion of households below 0.5 × median income. The analysis tracks poverty line and rate over time and during crises (Asian Financial Crisis 1998–1999; 2001–2003 recession; Global Financial Crisis 2008–2009; COVID-19 pandemic 2020), comparing changes in the poverty line, inflation (CPI), and unemployment. - Cost-of-living approach: Construct a cost-of-living index and poverty line based on essential goods and services prices. Base period t=0 in 1996; one-year intervals. Household equivalence size Q=2.5 (to match government’s conventional unit). Use CPI categories: food (including eating out), housing (rent, property management fees), clothing and footwear, durable goods, miscellaneous goods, transportation, and miscellaneous services (healthcare, education, banking, etc.). Compute index level F1=144.1 for 2020 with base F0=100 (1996). Derive the cost-of-living L(X) and the cost-of-living-based poverty line for a Q=2.5 household, yielding HK$20,476.20 in 1996 and HK$28,815.08 in 2020. Compute poverty rates by counting households with incomes below the cost-of-living poverty line (Eq. (6)). Analytical steps: 1) Evaluate sensitivity of the Gini Index (aggregate and decomposed by household size) to shifts in the income distribution and household composition over 1996–2020. 2) Compare relative-poverty-based poverty line and rates to economic conditions across crises, assessing lags vs CPI inflation and unemployment. 3) Re-estimate poverty rates annually using the cost-of-living poverty line, and compare levels and trajectories to the relative measure, including the number of households below each line.

- The Gini Index, even when decomposed by household size, is insensitive to symmetric shifts at the distribution’s tails and can mask structural changes in incomes and household composition. It tends to stagnate around asymptotic levels (~0.5–0.6 in high-inequality societies), suggesting misleading stability in inequality. - Relative-poverty measures pegged to 50% of median income understate material deprivation, especially during crises, because the median and thus the poverty line can fall even as unemployment rises and costs increase or fail to fall commensurately. - In Hong Kong, during four crises (1998–1999 Asian Financial Crisis; 2001–2003 recession; 2008–2009 Global Financial Crisis; 2020 COVID-19), the poverty line based on the median declined while unemployment rose, and inflation outpaced changes in median income, generating lags: 7.40%, 5.00%, 3.58%, and 5.91%, respectively (Inflation minus change in median household income). - Relative measure (50% median) yields for 2020: poverty line HK$13,450; poverty rate 23.6%; poor households ≈623,540. - Cost-of-living approach yields for 2020: poverty line HK$28,815.08; poverty rate 44.47%; poor households ≈1,174,930—implying a shortfall of ≈551,400 households not captured by the relative measure. - Long-run trajectory: cost-of-living poverty rate declines from 52.07% (1996) to 44.47% (2020), but remains substantially higher than the relative measure (which rises from 16.43% to 23.6%). - Household distribution dynamics show growth in both lowest (HK$0–$9,999) and highest (≥HK$80,000) income brackets over time, coupled with population shifts from larger to smaller household sizes, which the Gini may misinterpret as stagnation or improvement.

The findings show that relative measures of inequality and poverty can misrepresent underlying economic wellbeing. The Gini Index’s insensitivity to symmetric changes at distribution tails, alongside its tendency toward asymptotes, obscures social mobility and structural shifts (e.g., changes in household size composition). Pegging the poverty line to a proportion of the median income yields lagging and countercyclical behavior during economic crises—lowering the bar for poverty precisely when unemployment rises and purchasing power is pressured, thus understating poverty. By contrast, the cost-of-living approach ties the poverty line to actual prices of essential goods and services, capturing inelastic needs (notably housing) and better reflecting real material deprivation. This approach identifies a substantially higher poverty rate and many more poor households than the relative measure, addressing the research objective of more accurately depicting poverty and informing policy thresholds for support. The results underscore the importance of distinguishing changes in income distribution from summary indices, and of aligning poverty measurement with real costs to improve welfare policy targeting and adequacy.

This study reassesses conventional econometric measurements of inequality and poverty and demonstrates that relative measures—the Gini Index and poverty lines set as a fraction of median income—mask changes in economic wellbeing and understate poverty in Hong Kong. A cost-of-living approach, grounded in essential goods and services prices, yields a higher and more realistic poverty line and rate, revealing a large shortfall in households counted as poor by relative measures. The paper contributes an intermediate path between absolute and purely relative metrics, aligning with capabilities-based perspectives and recent calls for methods that reflect both needs and context. Future research should refine cost-of-living baskets for varying household sizes and compositions, model substitution within consumption modules, extend analysis beyond urban service economies, and empirically explore the linkage between inequality dynamics and poverty.

The analysis fixes the household equivalence size at 2.5 persons (matching government practice), which may not reflect shifts in household-size distribution over time and can affect the poverty line and rate. The cost-of-living approach, while addressing inelastic needs, does not fully capture within-module substitution behavior when incomes fall (e.g., substituting bread for rice), leaving uncertainty over how households adjust consumption. The study focuses on an urban, service-oriented economy (Hong Kong), limiting generalizability to contexts with larger agricultural or rural labor shares. Finally, the relationship between inequality and poverty is not empirically modeled here; thus, their causal linkages remain to be investigated.