<![CDATA[Energy poverty is usually studied as a static condition, yet whether it persists at the household level, and why, determines how policy should respond. Using four waves of the Indonesia Family Life Survey covering 1997 to 2014 and a dynamic random-effects probit with the Wooldridge correction for the initial-conditions problem, a complete Mundlak specification and standard errors clustered on the origin household, this paper provides the first estimates of genuine state dependence in multidimensional energy poverty for Indonesia. Averaged across the panel, a household energy poor at the previous interview is about nineteen percentage points more likely to be energy poor now, and roughly half of the raw persistence a cross-sectional study would observe is spurious, reflecting durable household characteristics rather than a causal effect of past deprivation. Allowing the effect to vary by inter-wave interval reveals a strongly non-stationary regime: the observed effect falls from 26.5 to 7.1 percentage points and the implied spell shortens from about twenty-seven years to about five. Placed on a common annual footing, however, scarring per unit of time barely weakens and remains roughly an order of magnitude above European benchmarks, while turnover through deprivation accelerates sharply. A single-parameter model of the kind standard in this literature would conceal this, which carries an implication beyond the Indonesian case.]]>
Copyrights © 2026
