The Freedom of Constraint : A Multilevel Simulation Model of Politics , Fertility and Economic Development
income , and human capital from World Bank ( 2014 ), instability and political capacity from Fisunoglu ( 2014 ) and Kugler and Tammen ( 2012 ).
We then fuse POFED endogenous systems to agent attribute changes with a generalizable , noncooperative prisoner ’ s dilemma game following Axelrod ( 1997a , 1997b , 1997c ), Nowak and Sigmund ( 1993 , 1998 ) to simulate intra-societal , spatial economic transactions . Understanding the interactive political effects of macro-socio dynamics and individual agency in intra-societal transactions is a key element of a complex adaptive systems approach . Finally , we explore the parameter space via simulation methods to identify paths and pitfalls toward economic , demographic , and political development , as well as societal interaction across different stages of development .
4 . Empirical Estimation
This is the first time the full POFED model was tested in a system , following
the previous work ( Feng et al ., 2000 , 2008 ) that empirically tested different parts of the model separately . Previous research all focus on single equation and ignore simultaneous correlations between various equations ’ error terms , which results in inefficiency . More specifically , the classical linear regression model , general linear regression model , and seemingly unrelated regressions model make the assumption that the error term is uncorrelated with each explanatory variable , which doesn ’ t seem to be the case in the POFED model . Following the system dynamics approach by Abdollahian and Kang ( 2008 ) and Abdollahian et al . ( 2013 , 2014 , 2015 ), I first use three-stage least-squares estimation to generate POFED coefficient estimates of the system to gain more efficiency by incorporating unobservable correlation across equations . Three-stage least-squares is a combination of multivariate regression and two-stage least-squares that permits correlations of the unobserved disturbances across several equations , as well as restrictions among coefficients of different equations , and improves upon the efficiency of equation-by-equation estimation by taking into account such correlations across equations .
In our estimation , we use panel data that covers 171 countries in the period
7