Kazakhstan’s Regional Divide Shows Why Poverty and Inequality Need Different Fixes

Kazakhstan's progress on poverty looks markedly different once national averages are broken down by region. A new study "Regional Disparities in Poverty in Kazakhstan: An Empirical Analysis Across 20 Administrative Regions, 2010–2023," published in Economies by Ainura Alibekova, Magbat Spanov and Gulnar Bugubaeva, shows that the country's poverty landscape is not defined by a single national trajectory, but by sharply different regional development patterns.

Using official data from Kazakhstan's Bureau of National Statistics, the authors examine poverty rates across all 20 administrative regions alongside income distribution, purchasing power, poverty depth and severity, and the income share of the poorest 40%. Their key finding is counterintuitive: some of Kazakhstan's poorest regions also record relatively low inequality because low incomes are broadly shared, while more industrialised regions combine lower poverty with greater income concentration.

The result complicates conventional development analysis. Poverty and inequality do not necessarily rise and fall together, and policies designed around national averages risk missing the structure of deprivation within individual regions. Kazakhstan's experience suggests that reducing poverty, narrowing inequality and raising purchasing power can require very different interventions depending on local economic conditions.

Poverty fell fast, then stopped falling

Kazakhstan recorded a substantial decline in poverty during the first half of the study period. The national poverty rate fell from 6.5% in 2010 to 2.5% in 2016, reflecting a period of stronger economic growth, rising hydrocarbon revenues, expanding formal employment and wage gains.

Progress did not continue at the same pace. Poverty stood at 2.7% in 2017 before rising to 4.3% in 2018, although the authors stress that much of this jump reflected a revision to the subsistence-minimum methodology rather than a sudden deterioration in household welfare. The poverty threshold itself changed, making direct comparisons across that break more difficult.

The COVID-19 shock produced a more substantive setback. National poverty rose to 5.3% in 2020, while the share of household spending devoted to food increased to 53.9%, signalling a squeeze on living standards. Poverty then stabilised at 5.2% between 2021 and 2023, leaving the rate roughly twice its 2016 low.

Regional convergence has also been weak. The study's coefficient of variation declined only marginally over the full period and moved unevenly from year to year. Rather than showing a steady narrowing of regional differences, the evidence points to slow and fragile convergence shaped by economic shocks, measurement changes and structural differences between regions.

One country, several kinds of poverty

The scale of regional divergence becomes clearer when individual regions are compared. In 2023, poverty stood at 9% in Turkestan, compared with just 2.4% in Astana. Over the full observation period, Turkestan recorded the highest mean poverty rate at 8.84%, while Astana had the lowest at 1.36%.

Turkestan is especially important because its poverty trend was moving in the wrong direction. The study estimates a positive long-run trend in poverty there, indicating that deprivation was not simply high but persistent. The region also combines low household incomes with large family sizes, reinforcing pressure on per capita resources.

Mangistau illustrates a different form of vulnerability. Its poverty rate was high, but the more revealing indicator was purchasing power. The region recorded the lowest purchasing-power ratio in the study, reflecting the burden of particularly high living costs. The authors describe this pattern as "expensive poverty," where households near the poverty line face greater risk because local prices erode already limited incomes.

Ulytau presents another variation. Its poverty headcount was not among the highest, yet it recorded the greatest poverty depth and severity in the data examined. In practical terms, fewer people may fall below the poverty line than in some other regions, but those who do can face much larger income shortfalls.

These differences show why headcount poverty alone gives an incomplete picture. A region can have a moderate poverty rate but very severe deprivation among the poor, while another can have widespread poverty concentrated close to the threshold. Effective policy requires knowing not only how many people are poor, but how far below the poverty line they are and what structural conditions keep them there.

Low inequality does not always mean shared prosperity

The study reveals a negative relationship between poverty and the Gini coefficient. Across regions, higher poverty is associated with lower measured inequality, while regions with lower poverty frequently show greater income concentration.

The authors describe this as a "uniform poverty paradox." In poorer agrarian regions, incomes may be compressed because a large share of the population earns similarly low amounts. A relatively low Gini coefficient in such circumstances does not indicate prosperity distributed fairly; it can indicate that deprivation itself is broadly distributed.

Turkestan illustrates this pattern. Its poverty rate was the country's highest in 2023, yet its Gini coefficient was comparatively low. Large parts of the population depend on agricultural self-employment and lower-productivity activity, creating a narrow income distribution at a low level of welfare.

Industrialised regions show almost the reverse. Pavlodar and Karaganda record lower poverty but higher Gini coefficients, reflecting the coexistence of relatively well-paid industrial employment with lower-income groups in services and other activities. Their income distributions are more unequal, but a larger share of households remain above the poverty line.

Mangistau again sits outside a simple binary classification. Oil-sector employment produces high wages for some workers, yet households excluded from that part of the economy face high living costs and weaker purchasing power. The resulting pattern combines low measured inequality with substantial poverty risk.

For policymakers, this is more than a statistical curiosity. Progress on SDG 1 and SDG 10 can move in different directions within the same region. A locality may perform better on inequality indicators while still struggling with poverty, or reduce poverty while income concentration increases. Tracking one indicator without the other can therefore produce an incomplete assessment of development outcomes.

Regional policy must match the structure of deprivation

The authors argue that Kazakhstan's social policy should move away from uniform national approaches and respond more directly to regional conditions. High-poverty regions such as Turkestan, Abai and Zhetysu require interventions aimed at employment, household vulnerability and access to essential services rather than relying only on cash redistribution.

Household structure is particularly relevant. The study reports that 84.5% of poor people nationally live in households of five or more members, with the share reaching 100% in Turkestan. The authors do not treat household size as a standalone cause of poverty, but the concentration is strong enough to raise questions about whether transfer systems adequately reflect family composition.

They propose examining household-size adjustments within targeted social assistance, especially in regions where large families dominate the poverty profile. Youth employment measures and stronger rural access to education and healthcare are also identified as priorities for areas where poverty is persistent and linked to weaker economic opportunity.

Mangistau requires a different response. Because high living costs weaken household purchasing power, social transfers and poverty thresholds may need stronger regional calibration. A uniform national benchmark can underestimate vulnerability where essential expenses are substantially higher.

Lower-poverty but higher-inequality regions face another set of challenges. In Pavlodar, Karaganda, Almaty and Astana, the policy problem is less about widespread poverty and more about how income gains are distributed across sectors and groups. Labour standards, contract enforcement and access to public services become more important where overall prosperity is stronger but unevenly shared.

The pandemic experience also points to the need for urban-specific resilience mechanisms. Almaty and Shymkent recorded the largest poverty increases during the COVID-19 shock, highlighting the exposure of service workers, informal workers and renters in large cities. Social protection systems that can be activated quickly during future shocks would need to reflect these urban vulnerabilities rather than assuming rural and urban poverty respond in the same way.

The authors also caution that the analysis is descriptive and correlational, not causal. Regional panel data on unemployment, education, urbanisation, social spending, sectoral employment and gross regional product were unavailable, preventing a fuller explanation of what drives poverty differences between regions.

Other constraints are equally important. The Gini coefficient is available only as a 2025 cross-section, while poverty is analysed over 2010–2023, meaning correlations do not compare indicators measured over exactly the same period. Abai, Zhetysu and Ulytau also have much shorter time series because they were created through the 2022 administrative reform.

Spatial spillovers across Kazakhstan's southern regions remain an important unanswered question, particularly whether persistent poverty in one area reinforces disadvantage in neighbouring regions. Testing household-size-adjusted social transfers could also show whether policy redesign reduces not only the number of people in poverty but also the depth and severity of deprivation. Comparative analysis with other resource-rich Central Asian economies would help establish whether Kazakhstan's poverty–inequality pattern is country-specific or part of a wider regional structure.