<![CDATA[The elasticity of a poverty line with respect to mean living standards is the parameter that distinguishes an absolute poverty line from a weakly relative one in the sense of Ravallion and Chen (2011). This paper shows that three deflation and measurement choices materially distort estimates of that parameter, quantifies each using Indonesian data, and then estimates the parameter itself on a provincial panel. Trap one is the nominal-on-nominal ratio, which recovers a weighted average of the true elasticity and unity and is biased upward whenever inflation is positive. Trap two is deflating a national-accounts aggregate with a consumer price index, which over 2004-2013 implies Indonesian real growth of 88 percent against a national-accounts figure of 48 percent. Trap three is the denominator concept: GDP per capita includes investment, government spending, and net trade, whereas a Cost of Basic Needs line is a consumption bundle. Estimating equation (3) on 390 province-by-area observations for 2013-2019, with province fixed effects and standard errors clustered by province, gives an elasticity of 0.64 (95 percent CI 0.57 to 0.71) in real terms, against 0.85 on the same data in nominal terms, representing trap one reproduced in panel form. The line is therefore weakly relative in Ravallion and Chen's sense, significantly bounded away from both zero and unity. Two further results follow. First, on disjoint windows built from a single BPS series, the elasticity rises from 0.05 (urban) and 0.51 (rural) over 2004-2010 to 1.02 and 1.25 over 2013-2022: Indonesia's poverty line was close to absolute before 2010 and at or beyond the strongly relative boundary afterwards. Second, adding year fixed effects drives the elasticity to zero and statistical insignificance, indicating that the mechanism operates through national annual re-basing rather than through province-specific responsiveness, a localization of the mechanism that the existing literature has not reported]]>
Copyrights © 2026
