154
Impact of Fuel-Switching on Early-Life Mortality: A Natural Experiment in Indonesia

Tuesday, June 14, 2016
Lobby (Annenberg Center)

Author(s): Imelda Wang

Discussant:

Does reducing exposure to indoor air pollution could reduce early life mortality through the adoption of cleaner cooking fuels? The selection biases and non-random assignment of fuel choice are often a concern in earlier literatures. Addressing the endogeneity issue, this paper uses Indonesia’s Kerosene to LPG (Liquefied Petroleum Gas) Conversion Program as a natural experiment. The program was introduced in 2007 to encourage kerosene users to switch to LPG. Fuel stacking (using both LPG and Kerosene) should not be possible as by the time the LPG was fully penetrated, the subsidy on kerosene would be removed. By offering subsidized price of gas cylinders and stoves, the government effectively encourage LPG switching in both urban and rural areas.  The fuel-switching did happen and it was reflected by a significant rise of LPG users to almost five fold within three years.

Previous literature find that solid fuel has a negative association with health, but few showed similar result for kerosene and LPG. In some contexts, compared to LPG, solid fuel has 19 times higher of carbon monoxide (CO) and 26 times higher of particulate matter (PM) on average per meal, while kerosene has only about three times higher of CO and 30% higher of PM. This study examines whether a slight improvement in daily and close proximity of air pollution could potentially improve early life health.

This paper uses Indonesian Demographic and Health Survey (DHS) for period 2002-2012 which covered 15 years birth and pregnancy data.  I assume that the program effects health outcomes only if household switch to LPG. Using the program as my instrument variable for cooking fuel choice  and controlling for other confounding factors (socio-demographic and maternal health characteristics), I find a significant decreasing marginal effect on early life mortality (i.e. stillbirth and infant death) after the program but no effect on adult's health outcomes. Moreover, the marginal effect is higher during prenatal period. For robustness check, I use OLS and matching methods to compare outcomes by who are observationally similar in whether they are actually shift to LPG. The result is consistent across different methods.

The result adds more evidence that providing energy subsidy that promotes a lower level of emission could be beneficial for early-life health. It suggests that targeting pregnant woman in public health efforts could be an important priority concerning air pollution.