The link between climate change, food security and fertility: The case of Bangladesh

impacts of climate change on crop production and the resulting food insecurity are visible and studied in different countries. In response to food insecurity, vulnerable populations see demographic adaptations, such as changes in fertility behaviour, as coping mechanisms. Research has already examined the effects of climate variables on crop production and fertility changes in different areas separately. However, to the best of our knowledge, there is no such attempt made to jointly study the direct and indirect, as well as short- and long-term, effects of climate variables on crop production and subsequently fertility. Data and methodology This study incorporates climate change, production of major crops (i.e., rice, wheat, and pulses), and fertility data of Bangladesh during 1966–2015. We collected climate data from the Bangladesh Meteorological Department (BMD) (http://www.bmd.gov.bd) [79]. Then we cal- culated the annual maximum, average and minimum values of temperature and precipitation for the period 1966–2015. Since the minimum temperature and precipitation for some years are zero, we did not use the two indicators for the convenience of our analysis. We also col- lected annual data for rice, wheat, and legume production from the United Nations Food and Agriculture Organization (FAO) (http://www.fao.org/home/en/) [80]. The data for infant mortality and total fertility rate (TFR) was from the United Nations Population Divisions (https://population.un.org/wpp/) [81]. Because crop production is not only sensitive to cli-

matic factors but also to fertilizer use, we include fertilizer consumption (i.e., the amount of plant nutrients used per unit of arable land) as a control variable, with data from the World Bank (https://data.worldbank.org/). The dependent variable is TFR. The independent variables are temperature, rainfall, infant mortality, as well as rice, wheat, and legume production. The GDP per capita, which reflects the average income of the population, is also included as a con- trol variable. However, because crop production is an important factor of agriculture income and a crucial contributor to the GDP in Bangladesh, GDP per capita becomes an endogenous factor. To avoid endogeneity of GDP per capita, we finally decided to take the GDP per capita from the industry and service sector, that is, GDP per capita�(1 - % of agriculture, forest and fishing value add in GDP). Both the GDP per capita and the percent of agriculture, forest and fishing value add in GDP are from the World Bank (https://data.worldbank.org/). For simplic- ity, in the remaining paper, we use the term “GDP per capita” to represent “the GDP per capita from the industry and service sector”. To examine the direct and indirect impact of climate change, we adopt the structural equa- tion model (SEM) to perform pathway analysis. Path analysis allows us to distinguish the direct impact of climate change on fertility and its indirect impact on fertility through food security. To assess the short- and long-term relationship between climate change and fertility, we apply

the error correction model (ECM). ECM is a very useful econometric method for exploring time series data, which allows us to study the long-term equilibrium relationship between cli- mate change and fertility, as well as how fertility responds to climate shock in the short term. Because agricultural production and infant mortality are endogenous, we do not include them in the ECM model. The error-correction equations are formulated as DTFR ¼ s þ Xp i¼1 gjDTFRt i þ Xp j¼1 ZjDclimatet j þ Xp m¼1 xmDGDPPCt m þ lECTt 1 þ ut ð1Þ ECTt 1 ¼ TFRt 1 aclimatet 1 bGDPPCt 1 ð2Þ PLOS ONE Climate change, food security and fertility PLOS ONE | https://doi.org/10.1371/journal.pone.0258196 October 21, 2021 5 / 18