A. Effects of In Utero Shock
Table 3.4 presents the effects of in utero price shock on Raven/IQ test an grade attainment estimated from equation (3.3). Column (1) provides estimates of in utero price shock on Raven/IQ test score for sample of children aged 9 to 17. Column (2) shows the effect of the in utero cocoa price shock on grade attainment for sample of children of aged 6 to 17. The in utero price shock has significant positive effect on both outcomes. Children conceived during higher cocoa prices and born in cocoa producing regions have grown to have higher Raven/IQ score and higher grade attainment. A standard deviation increase in real producer price of cocoa increases the correct Raven/IQ items answered by a child by 4.2 percentage points, which is equivalent to 11% of the average score. It also increases grade attained by 0.38 year, which is 13% of the average grade attained by the children in the sample. Similarly, Shah and Steinberg (2017) also find in utero and early life economic booms strongly and positively affect human capital development in India.
In Table 3.4, a standard deviation increase in prenatal year cocoa price is associated with 4.2% percentage points more in Raven/IQ test score. To gauge the economic impact of the shock, I estimate the impact of Raven/IQ test on different cognitive achievement test scores. Table B.6 in Appendix B.1 reports these results. Thus, the effect of the gain in Raven/IQ due to in utero price boom translates in to a gain of 2.73 percentage points in simple reading test score, 2 percentage points in simple maths test score, 1.12 percentage points in advanced reading and 0.63
Table 3.4: Estimated effect of in utero cocoa price shock on cognition Age 9 to 17 Age 6 to 17 (1) (2) Raven/IQ Grade Shock in utero 4.214*** 0.379*** (1.357) (0.047) [0.048] [0.000] Observations 2,826 49,621 R-squared 0.152 0.361
Region of birth FE Yes Yes
Year of birth FE Yes Yes
Region of birth trends Yes Yes
Controls Yes Yes
Robust standard errors (clustered at the region of birth level) in parentheses. Wild-bootstrap p-values in brackets. The asterisks next to the coefficients are for p-values associated with the main (non-wild bootstrap) regressions. *** p<0.01, ** p<0.05, * p<0.1. Controls include (X): Gender of the child and gender of household head. The Raven/IQ includes children of ages 9 to 17, while grade is reported for children of age 6 to 17. The source of data for Raven/IQ outcome is self-weighting samples. However, survey weights are used in the regression estimation of grade outcome.
percentage points in advanced maths test score.
Furthermore, in TableB.7 in AppendixB.1 I report the impacts of in utero price shock on other different cognitive achievement scores. Except for simple reading score, in utero price shock does not significantly affect achievement scores.
B. Early Life Shock Timing
The ‘fetal origins’ hypothesis predicts that the period in utero is the most critical in a terms of health and human capital development. Nutritional shocks during that period have persistent effects (Barker,1990,1992;Dobbing,1972;Almond and Currie,
2011). However, the literature has questioned that assertion by extending the critical period to early life after birth. Studies like Maccini and Yang (2009) concludes that the [rain fall] shock in the year of birth that results in nutritional shock in the following year is crucial for long-term health and socio economic development. Glewwe and King (2001) suggest that nutritional shocks at the second year of life have greater effect than shocks in the first year of life. Another set of studies find that both the period in utero and the first year of life are equally important (Dobbing,1972;Akresh et al., 2012), while some others find evidence that all early years are important and
there is no special unique period among the early life years (Adhvaryu et al., 2014;
Shah and Steinberg, 2017). These evidences show that not just price shock in utero but also shocks in other early life years matter for cognitive development. In addition, real cocoa price can be serially correlated over time. As a result, the results in the baseline (Table3.4) may be attributed not just to price shock in utero but also shocks in other early life years. In other words there might be omitted variable bias in the estimated effects in the baseline as a result of not controlling for price shocks in other years than in utero period. In this section, to control for possible omitted variables and to test if the early life shocks that happen from in utero to 3 years after that have variant effect on cognition, I re-estimate equation (3.3).12 I include not only in utero
price shock but also shocks that occur just before and after in utero.13 The results
are depicted in Table 3.5. Column (1) shows the effect of price shocks in the year before in utero, in utero, year of birth, year of birth plus 1 year, and year of birth plus 2 years for Raven/IQ test. Similarly, column (2) shows the same analysis for grade attainment outcome. For Raven/IQ test, though the coefficients on in utero and year of birth price shocks are positive they are insignificant.14 Column (2) shows that in
utero price shock is significantly related with grade attainment and the coefficient is comparable with the baseline effect in Table 3.4. This indicates that the result in the baseline is indeed due to the effect of shock in utero. There is also a marginally significant effect on grade attainment from price shock at year of birth plus 1 year.
12This discussion is in line withMaccini and Yang (2009).
13Note that the literature on early life shock considers all years before the 5th year birth day
as critical period of development. Therefore, to incorporate all these early life years and as an alternative to the specification thus far, I conducted another exercise as robustness. In this case, shock is constructed as interaction between the average prices in early life (in utero, year of birth, year of birth plus 1 year, year of birth plus 2 years, year of birth plus 3 years and year of birth 4 years) with the indicator if the region of birth is cocoa producer. Table B.8 in Appendix B.1
presents these results. It shows that average early life price boom is positively related with later childhood cognitive outcomes.
14The data for Raven test score consists of children born between 1971 to 1994. The short period of time considered in the analysis may affect the precision of estimates when serially correlated variables are included in the same regression.
Table 3.5: Cocoa price shock timing and cognition Age 9 to 17 Age 6 to 17 (1) (2) Raven/IQ Grade Shock year 0 - 2 -3.449 0.075 (2.458) (0.104)
Shock year 0 - 1 (in utero) 2.757 0.277***
(4.437) (0.044)
Shock year 0 (year of birth) 6.471 -0.043
(5.532) (0.040) Shock year 0 + 1 -3.267 0.141* (2.325) (0.078) Shock year 0 + 2 0.815 0.038 (1.394) (0.039) Observations 2,826 49,621
Region of birth FE Yes Yes
Year of birth FE Yes Yes
Region of birth trends Yes Yes
Controls Yes Yes
Robust standard errors (clustered at the region of birth level) in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Controls include (X): Gender of the child and gender of household head. The Raven/IQ includes children of ages 9 to 17, while grade is reported for children of age 6 to 17. The source of data for Raven/IQ outcome is self-weighting samples. However, survey weights are used in the regression estimation of grade outcome.