I'm working on two GMM regressions to estimate the demand for sugar and demand for
g ur (a sugar substitute) in India. The regression for sugar worked very well and I'm at a loss because of the
gur regression which is specified to mirror the sugar regression in my models system of simultaneous equations and yet it's not working well. I've attached screenshots of the output log. Note:
the first screenshot is of the sugar regression where lnINSGRQDM=ln(sugardemand), lnINGDPPC=ln(GDP per capita), lnINGURPRA=ln($ of gur), lnINSGRPRA=ln($ of sugar), lnINSCAPRA=ln($ of sugarcane), INSGRPRT=sugarprofit($ of sugar-marginal cost of sugar), INGURPRT=gurprofit($ of gur-marginal cost of gur).The second screenshot is of the
gur regression where lnINGURQDM=ln(gur demand), INPLRCBI=interest rates, and every other included variable's specification is identical to those given in the sugar regression. Does anyone have advice on how I can improve the
gur regression? it's an integral part of my model and it wouldn't be optimal to treat it as exogenously given indications in my research that it is endogenous to the system. Why would the regression for sugar work so well while the one for gur is so poor? I'm fairly new to SEM modeling so any advice is helpful. Thank you!
