Is the development of obesogenic food environments a self-reinforcing process? Evidence from soft drink consumption

Table 5 Regression results (2SLS): Obesity, non-high-income countries

Dependent variable: Soft drink consumption, *QSD*				Dependent variable: Prevalence of obesity, *OBE*
Independent variables		Coefficient	Std. Error¹	Independent variables		Coefficient	Std. Error¹
Prevalence of obesity	OBE	1.776***	0.270	Soft drink consumption	QSD	0.058**	0.029
Soft drink price	PSD	−0.628*	0.324	Dietary energy supply	DES	2.572***	0.548
Income per capita	GNI	2.780***	0.481	Urban population	URB	0.389***	0.045
Population aged 65 and above	AGE	−1.652	1.185	Population aged 65 and above	AGE	0.825***	0.183
Economic globalisation	GLO	0.213*	0.086	Economic globalisation	GLO	−0.062**	0.019
N. of obs.	578			N. of obs.	578
F-statistic, F(5, 530)	85.95, Prob. 0.000			F-statistic, F(5, 530)	193.15, Prob. 0.000
Underidentification test	77.41, P-val. 0.000			Underidentification test	31.36, P-val. 0.000
Weak identification test	167.47			Weak identification test	45.41
Sargan-Hansen J statistic	7.303, P-val. 0.007			Sargan-Hansen J statistic	0.164, P-val. 0.685

Notes: Tables 4 and 5. 2SLS: Two-stage least square estimation, with fixed effects
¹ Heteroskedasticity-robust standard errors. *, ** and *** denote p < 0.05, p < 0.01 and p < 0.001, respectively
Underidentification test (Kleibergen-Paap rk LM statistic), Weak identification test (Cragg-Donald Wald F statistic),
Sargan-Hansen J statistic (overidentification test of all instruments)