Turning to our non-OECD estimation ,we observe in the correlation matrix (Table A5, Appendix 5) that this model too does not suffer from the collinearity of the first model ,and therefore GDP is used throughout. The results in Table A8 (Appendix 8) illustrates that Land is again statistically significant in all columns. But ,unlike our other estimations ,this variable is negative in each column ,has over all greater statistical significance, and has significantly higher magnitudes.The larger