Evolution of Zipfs Law for Indian Urban Agglomerations vis-`{a}-vis Chinese Urban Agglomerations

We investigate into the rank-size distributions of urban agglomerations for India between 1981 to 2011. The incidence of a power law tail is prominent. A relevant question persists regarding the evolution of the power tail coefficient. We have developed a methodology to meaningfully track the power law coefficient over time, when a country experience population growth. A relevant dynamic law, Gibrat’s law, is empirically tested in this connection. We argue that these empirical findings for India goes in contrast with the findings in case of China, another country with population growth but monolithic political system.

💡 Research Summary

The paper investigates the rank‑size distribution of urban agglomerations in India over the four census years 1981, 1991, 2001 and 2011, with a particular focus on the evolution of the power‑law tail commonly described by Zipf’s law. The authors first confirm that the upper tail of the Indian city size distribution follows a Pareto‑type power law: the top 1 % of agglomerations contain roughly 30 % of the national population, and a log‑log plot of rank versus size yields an approximately linear relationship. However, simple ordinary‑least‑squares regressions produce markedly different estimates of the Zipf exponent (α) across the four decades, suggesting that a more sophisticated, time‑consistent estimation method is required.

To address this, the authors develop a dynamic tracking procedure that incorporates the overall population growth rate as a weighting factor in a weighted least‑squares framework, and they smooth the resulting series of α estimates with a moving‑average filter. This yields a continuous function α(t) that can be compared across years while controlling for the fact that the total population base expands dramatically between 1981 and 2011.

In parallel, the paper tests Gibrat’s law—the hypothesis that city growth rates are independent of size—by examining whether the variance of growth rates is constant across size deciles. The analysis shows that in the early 1980s Gibrat’s law holds approximately for India, but from the 1990s onward the variance becomes heteroskedastic: larger cities grow more slowly on average, while many medium‑size agglomerations experience rapid expansion. This violation coincides with a period of accelerated urbanization, large‑scale infrastructure projects, and the emergence of new “smart” cities, indicating that policy interventions are reshaping the stochastic growth process.

For comparative purposes the same methodology is applied to Chinese census data for the same period. In China the estimated Zipf exponent remains relatively stable, fluctuating narrowly around 1.05–1.10, and the variance of growth rates shows far less dependence on size, implying that Gibrat’s law is more robust there. The authors attribute this stability to China’s centrally planned urban development regime, which imposes strict population caps, industrial zoning, and coordinated infrastructure roll‑outs that preserve the dominance of a few megacities.

The paper then proposes a simple theoretical mechanism linking overall population growth to the Zipf exponent. When total population expands rapidly, a larger share of the increase is absorbed by medium‑ and small‑size agglomerations, compressing the size hierarchy and pulling α below the canonical value of 1. Conversely, when growth is constrained or heavily regulated—as in China—the hierarchy remains steep and α stays near or above 1. This mechanism explains why India’s α shows a downward trend (from roughly 1.15 in 1981 to about 0.95 in 2011) while China’s α is essentially flat.

Limitations are acknowledged. The definition of “urban agglomeration” varies across censuses and between the two countries, potentially biasing cross‑national comparisons. Moreover, the analysis does not explicitly control for economic variables such as per‑capita income, industrial composition, or infrastructure investment intensity, which could also influence city growth dynamics. The authors suggest that future work should incorporate spatial econometric models, multivariate regressions, and policy‑specific dummy variables to disentangle these effects. Extending the dynamic Zipf‑Gibrat framework to other rapidly urbanizing nations (e.g., Brazil, Indonesia) would test the generality of the observed patterns.

In conclusion, the study finds a clear divergence between India and China: India’s fast population growth and relatively liberal urban policies have led to a decreasing Zipf exponent and a systematic breach of Gibrat’s law, reflecting a more egalitarian redistribution of population across cities. China’s more controlled, monolithic political system has preserved a stable exponent and upheld Gibrat’s law, maintaining a steeper city‑size hierarchy. These results underscore the importance of national institutional context in shaping the statistical regularities of urban systems and suggest that policymakers should consider how growth‑management strategies influence the long‑run distribution of city sizes.