‫اﻟﻤﺠﻠﺔ اﻟﻌﺮاﻗﻴﺔ ﻟﻠﻌﻠﻮم اﻹﺣﺼﺎﺋﻴﺔ )‪2012 (21‬‬

‫ص ص ]‪[285-266‬‬

‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﺜﺎﺒﺘﺔ ﻭﺍﻟﻌﺸﻭﺍﺌﻴﺔ‬

‫‪1‬‬
‫ﺯﻜﺭﻴﺎ ﻴﺤﻴﻰ ﺍﻟﺠ‪‬ﻤﺎل‬

‫ﺍﻟﻤﺴﺘﺨﻠﺹ‬

‫ﺘﻡ ﺍﺴﺘﺨﺩﺍﻡ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻭﻜﻴﻔﻴﺔ ﺍﻻﺨﺘﻴﺎﺭ ﻓﻲ ﺘﻭﻓﻴﻕ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ‬
‫ﺍﻟﺜﻼﺜﺔ‪ ،‬ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺄﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌـﺸﻭﺍﺌﻴﺔ‪.‬‬
‫ﺘﻡ ﺍﻻﻋﺘﻤﺎﺩ ﻋﻠﻰ ﻤﻌﻴﺎﺭ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﻤﺼﺤﺢ ﻭﻤﻌﻴﺎﺭ ﺍﻜﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤـﺎﺕ ﻓـﻲ ﺍﺨﺘﻴـﺎﺭ‬
‫ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ ﻤﻥ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻌﺎﻡ ﻤﻥ ﺨﻼل ﺩﺭﺍﺴﺔ ﺍﺭﺒﻌﺔ ﻤﺘﻐﻴﺭﺍﺕ ﺘﻭﻀﻴﺤﻴﺔ ﺘـﺅﺜﺭ‬
‫ﻓﻲ ﺍﻟﻤﻭﺍﺯﻨﺔ ﺍﻟﻌﺎﻤﺔ ﻷﻴﺔ ﺩﻭﻟﺔ‪ ،‬ﺤﻴﺙ ﺸﻤﻠﺕ ﺍﻟﺩﺭﺍﺴﺔ ﺜﻤﺎﻨﻲ ﻋﺸﺭﺓ ﺩﻭﻟـﺔ ﻋﺭﺒﻴـﺔ ﻭﻟﻠﻔﺘـﺭﺓ‬
‫ﺍﻟﺯﻤﻨﻴﺔ ﻤﻥ ‪ 1995‬ﺍﻟﻰ ‪.2009‬‬

‫ﺍﻟﻜﻠﻤﺎﺕ ﺍﻟﺩﺍﻟﺔ‪ :‬ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ‪ ،‬ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ‪ ،‬ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ‬
‫ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‪ ،‬ﻭﻤﻌﻴﺎﺭ ﺍﻜﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤﺎﺕ‪ ،‬ﻭﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﻤﺼﺤﺢ‬

‫‪Selecting Model in Fixed and Random Panel Data Models‬‬

‫‪Zakariya Y. Algamal‬‬

‫‪Abstract‬‬

‫‪In this paper we used panel data models and how to choose the‬‬
‫‪fitting three panel data models, the pooled regression model, fixed‬‬
‫‪effects model, and random effects model, depending on the adjusted‬‬
‫‪coefficient of determination and Akaike information criterion in‬‬

‫‪1‬‬
‫ﻤﺩﺭﺱ‪ /‬ﻗﺴﻡ ﺍﻹﺤﺼﺎﺀ ﻭﺍﻟﻤﻌﻠﻭﻤﺎﺘﻴﺔ ‪ /‬ﻜﻠﻴﺔ ﻋﻠﻭﻡ ﺍﻟﺤﺎﺴﻭﺏ ﻭﺍﻟﺭﻴﺎﻀﻴﺎﺕ‪ /‬ﺠﺎﻤﻌﺔ ﺍﻟﻤﻭﺼل‪.‬‬

‫ﺘﺎﺭﻴﺦ ﺍﻟﺘﺴﻠﻴﻡ ‪2012/3/20‬‬ ‫ﺘﺎﺭﻴﺦ ﺍﻟﻘﺒﻭل ‪2012/2/7‬‬

 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[267‬‬

‫‪selecting the best sub-model form the general model through‬‬

‫‪studying the four affected variables on the public budget. This study‬‬
‫‪included of eighteen Arab countries for the period from 1995 to 2009.‬‬

‫‪ -1‬ﺍﻟﻤﻘﺩﻤﺔ‬

‫ﺘﻌﺘﻤﺩ ﺍﻟﻌﺩﻴﺩ ﻤﻥ ﺍﻟﺩﺭﺍﺴﺎﺕ ﻭﺍﻟﺒﺤﻭﺙ ﻋﻠﻰ ﺃﺴﺎﻟﻴﺏ ﻤﺘﻁﻭﺭﺓ ﻤﻥ ﺃﺠل ﺍﻟﺤﺼﻭل ﻋﻠـﻰ‬
‫ﻨﺘﺎﺌﺞ ﺘﺘﺼﻑ ﺒﺎﻟﻔﻌﺎﻟﻴﺔ ﻭﺍﻟﺩﻗﺔ ﺍﻟﻌﺎﻟﻴﺘﻴﻥ‪ ،‬ﻭﻗﺩ ﻜﺎﻥ ﻟﻌﻠﻡ ﺍﻹﺤﺼﺎﺀ ﻭﻓﺭﻭﻋﻪ ﺍﻟﻤﺭﺘﺒﻁﺔ ﺒﻪ ﺍﻷﺜﺭ‬
‫ﻻ ﺍﻟﻰ ﻗﺭﺍﺭﺍﺕ ﺍﻟﺴﻠﻴﻤﺔ‪.‬‬
‫ﺍﻟﻜﺒﻴﺭ ﻓﻲ ﺒﻨﺎﺀ ﺍﻟﻨﻤﺎﺫﺝ ﺍﻟﺭﺼﻴﻨﺔ ﻭﺘﺤﻠﻴل ﺍﻟﺒﻴﺎﻨﺎﺕ ﻤﻥ ﺨﻼﻟﻬﺎ ﻭﺼﻭ ﹰ‬
‫ﻴﻬﺘﻡ ﺘﺤﻠﻴل ﺍﻻﻨﺤﺩﺍﺭ ﺒﺒﻨﺎﺀ ﺍﻟﻌﻼﻗﺔ ﺍﻟﺭﻴﺎﻀﻴﺔ ﺒﻴﻥ ﻤﺘﻐﻴﺭ ﺍﻻﺴﺘﺠﺎﺒﺔ ﻭﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﺘﻭﻀـﻴﺤﻴﺔ‬
‫ﻭﺘﺘﻤﺜل ﻫﺫﻩ ﺍﻟﻌﻼﻗﺔ ﻜﺘﺭﻜﻴﺒﺔ ﺨﻁﻴﺔ ﺘﺩﻋﻰ ﻤﻌﺎﺩﻟﺔ ﺍﻻﻨﺤﺩﺍﺭ‪ .‬ﻟﻘﺩ ﺍﻜﺘﺴﺒﺕ ﻨﻤـﺎﺫﺝ ﺍﻟﺒﻴﺎﻨـﺎﺕ‬
‫ﺍﻟﻁﻭﻟﻴﺔ ﻓﻲ ﺍﻟﻌﻘﺩ ﺍﻟﺤﺎﻟﻲ ﺍﻫﺘﻤﺎﻤﺎ ﺒﺎﻟﻐﺎ ﻭﺨﺼﻭﺼﺎ ﻓﻲ ﺍﻟﺩﺭﺍﺴـﺎﺕ ﺍﻻﻗﺘـﺼﺎﺩﻴﺔ ﻭﺍﻟﻁﺒﻴـﺔ‪.‬‬
‫ﺘﻌﺭﻑ ﻫﺫﻩ ﺍﻟﺒﻴﺎﻨﺎﺕ ﻋﻠﻰ ﺍﻨﻬﺎ ﺒﻴﺎﻨﺎﺕ ﻤﻘﻁﻌﻴﺔ ﻤﻘﺎﺴﺔ ﻓﻲ ﻓﺘﺭﺍﺕ ﺯﻤﻨﻴﺔ ﻤﻌﻴﻨﺔ‪.‬‬

‫ﺍﻥ ﺍﻟﻔﺎﺌﺩﺓ ﺍﻟﺭﺌﻴﺴﻴﺔ ﻤﻥ ﺍﺴﺘﺨﺩﺍﻡ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻫﻲ ﺯﻴﺎﺩﺓ ﺍﻟﺩﻗﺔ ﻓﻲ ﺍﻟﺘﻨﺒﺅ ﻤﻥ ﺨﻼل‬
‫ﺯﻴﺎﺩﺓ ﻋﺩﺩ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ ﻋﻥ ﻁﺭﻴﻕ ﺭﺒﻁ ﻋﺩﺩ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ ﺒﻌﺩﺩ ﺍﻟﻔﺘﺭﺍﺕ ﺍﻟﺯﻤﻨﻴﺔ‪.‬‬
‫ﺍﻟﺒﺎﺤﺜﻴﻥ ﻓﻲ ﺩﺭﺍﺴﺔ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻓﻤﻨﻬﻡ ﻤﻥ ﺍﻫﺘﻡ ﺒﺩﺭﺍﺴﺔ‬ ‫ﺫﻫﺏ ﺍﻟﻌﺩﻴﺩ ﻤﻥ‬
‫ﺨﺼﺎﺌﺹ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺭﻴﺎﻀﻴﺎ )‪، (Bramati & Croux, 2007‬‬
‫)‪، (Lee & Yu, 2010) ، (Sun, 2010) ، (Dustmann & Engarcia, 2007‬‬
‫‪(El-‬‬ ‫)‪ (Baltagi et al, 2010‬ﻭﻤﻨﻬﻡ ﻤﻥ ﺍﻫﺘﻡ ﺒﺘﻁﺒﻴﻕ ﻫﺫﻩ ﺍﻟﻨﻤﺎﺫﺝ ﻓﻲ ﺩﺭﺍﺴﺎﺘﻬﻡ‬
‫)‪(Mikhed & ، (Chuang & Wang,2099)،Gamal & Inanoglu, 2005‬‬
‫)‪.(Kai & Qin, 2011) ، (Lukas & Jan, 2011) ، Zemcik, 2009‬‬

‫ﻟﻘﺩ ﺴﻌﻰ ﻫﺫﺍ ﺍﻟﺒﺤﺙ ﺍﻟﻰ ﺘﻘﺩﻴﻡ ﻤﻔﻬﻭﻡ ﻨﻤﻭﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻭﻜﻴﻔﻴﺔ ﺍﻻﺨﺘﻴﺎﺭ ﺒﻴﻥ‬
‫ﻨﻤﺎﺫﺝ ﻫﺫﻩ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺜﻡ ﺍﺨﺘﻴﺎﺭ ﺃﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ‪.‬‬
 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[268‬‬

‫‪ -2‬ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ‬

‫ﻟﻘﺩ ﺍﻜﺘﺴﺒﺕ ﻫﺫﻩ ﺍﻟﻨﻤﺎﺫﺝ ﺍﻟﺘﻲ ﺘﺩﻋﻰ ﺍﻴﻀﺎ ﺒﻨﻤﺎﺫﺝ ﺍﻟﺒﺎﻨل )‪ (Panel Model‬ﻓﻲ ﺍﻟﻌﻘﺩ‬
‫ﺍﻟﺤﺎﻟﻲ ﺍﻫﺘﻤﺎﻤﺎ ﺒﺎﻟﻐﺎ ﻭﺨﺼﻭﺼﺎ ﻓﻲ ﺍﻟﺩﺭﺍﺴﺎﺕ ﺍﻻﻗﺘﺼﺎﺩﻴﺔ ﻭﺍﻟﻁﺒﻴﺔ ﻻﻨﻬﺎ ﺘﺎﺨﺫ ﻓﻲ ﺍﻻﻋﺘﺒـﺎﺭ‬
‫ﺍﺜﺭ ﺍﻟﺘﻐﻴﺭ ﻓﻲ ﺍﻟﺯﻤﻥ ﻭﻜﺫﻟﻙ ﺍﺜﺭ ﺍﻟﺘﻐﻴﺭ ﻓﻲ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ‪ .‬ﺘﻌﺭﻑ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴـﺔ‬
‫ﻋﻠﻰ ﺍﻨﻬﺎ ﻤﺸﺎﻫﺩﺍﺕ ﻤﻘﻁﻌﻴﺔ ﻤﻘﺎﺴﺔ ﻓﻲ ﻓﺘﺭﺍﺕ ﺯﻤﻨﻴﺔ ﻤﻌﻴﻨـﺔ‪ .‬ﺤﻴـﺙ ﺘـﺸﻤل ﺍﻟﻤـﺸﺎﻫﺩﺍﺕ‬
‫ﺍﻟﻤﻘﻁﻌﻴﺔ ﺍﻟﺩﻭل‪ ،‬ﺍﻟﻤﺤﺎﻓﻅـﺎﺕ‪ ،‬ﺍﻟﻤﺅﺴـﺴﺎﺕ‪ .....،‬ﺍﻟـﺦ )‪ .(Gujarati, 2003‬ﻭﻟﻐـﺭﺽ‬
‫ﺘﻭﻀﻴﺢ ﻤﻔﻬﻭﻡ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺴﻭﻑ ﻨﻔﺭﺽ ﺒﺎﻥ ﻟﺩﻴﻨﺎ ﺩﺭﺍﺴﺔ ﺤﻭل ﻤﻘـﺩﺍﺭ ﺍﻻﺴـﺘﻬﻼﻙ‬
‫ﻜﻤﺘﻐﻴﺭ ﺍﺴـﺘﺠﺎﺒﺔ ) ‪ ( y‬ﻭﻤﻘـﺩﺍﺭ ﺍﻟـﺩﺨل ﻜﻤﺘﻐﻴـﺭ ﺘﻭﻀـﻴﺤﻲ ) ‪ ( x‬ﻭﻟﺨﻤـﺱ ﻋﻭﺍﺌـل‬
‫)‪ (A, B, C, D, and E‬ﻭﻟﺜﻼﺙ ﺴﻨﻭﺍﺕ )‪ ،2010 ،2009‬ﻭ ‪ ،(2011‬ﻴﻭﻀﺢ ﺍﻟﺠـﺩﻭل‬
‫)‪ (1‬ﻜﻴﻔﻴﺔ ﻋﺭﺽ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻓﻲ ﻤﺜﺎﻟﻨﺎ ﻫﺫﺍ‪.‬‬

‫ﺍﻟﺠﺩﻭل)‪ :(1‬ﺍﺴﻠﻭﺏ ﻋﺭﺽ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ‬

‫ﻤﻘﺩﺍﺭ ﺍﻟﺩﺨل ) ‪( x‬‬ ‫ﻤﻘﺩﺍﺭ ﺍﻻﺴﺘﻬﻼﻙ )‪( y‬‬ ‫ﺍﻟﺴﻨﻭﺍﺕ‬ ‫ﺍﻟﻌﻭﺍﺌل‬

‫‪x A,2009‬‬ ‫‪y A,2009‬‬ ‫‪2009‬‬

‫‪x A,2010‬‬ ‫‪y A,2010‬‬ ‫‪2010‬‬ ‫‪A‬‬

‫‪x A,2011‬‬ ‫‪y A,2011‬‬ ‫‪2011‬‬

‫‪x B, 2009‬‬ ‫‪y B,2009‬‬ ‫‪2009‬‬

‫‪x B, 2010‬‬ ‫‪y B,2010‬‬ ‫‪2010‬‬ ‫‪B‬‬

‫‪x B,2011‬‬ ‫‪y B,2011‬‬ ‫‪2011‬‬

‫‪:‬‬ ‫‪:‬‬ ‫‪:‬‬ ‫‪:‬‬

 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[269‬‬

‫‪:‬‬ ‫‪:‬‬ ‫‪:‬‬ ‫‪:‬‬

‫‪x E,2009‬‬ ‫‪y E,2009‬‬ ‫‪2009‬‬

‫‪x E,2010‬‬ ‫‪y E,2010‬‬ ‫‪2010‬‬ ‫‪E‬‬

‫‪x E,2011‬‬ ‫‪y E,2011‬‬ ‫‪2011‬‬

‫ﻤﻥ ﺍﻟﺠﺩﻭل )‪ (1‬ﻨﻼﺤﻅ ﺒﺎﻨﻪ ﺘﻡ ﺩﻤﺞ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ ﻤﻊ ﺍﻟﻔﺘﺭﺍﺕ ﺍﻟﺯﻤﻨﻴﺔ‪ .‬ﺘﺘﻤﺘﻊ ﻨﻤـﺎﺫﺝ‬
‫ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻓﻲ ﺍﺴﺘﺨﺩﺍﻤﻬﺎ ﻤﻘﺎﺭﻨﺔ ﻋﻨﺩ ﺍﺴﺘﺨﺩﺍﻡ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ ﺒﻤﻔﺭﺩﻫـﺎ ﺍﻭ‬
‫ﻨﻤﺎﺫﺝ ﺒﻴﺎﻨﺎﺕ ﺍﻟﺴﻠﺴﻠﺔ ﺍﻟﺯﻤﻨﻴﺔ ﺒﻤﻔﺭﺩﻫﺎ ﺒﺎﻟﻌﺩﻴﺩ ﻤﻥ ﺍﻟﻤﺯﺍﻴﺎ ﺍﻟﺘﻲ ﻭﻀﺤﻬﺎ ﺍﻟﺒﺎﺤـﺙ ‪Blatagi‬‬
‫)‪ (Blatagi, 2005‬ﻤﻨﻬﺎ‪:‬‬

‫‪ -1‬ﺍﻟﺘﺤﻜﻡ ﻓﻲ ﻋﺩﻡ ﺘﺠﺎﻨﺱ ﺍﻟﺘﺒﺎﻴﻥ ﺍﻟﺨﺎﺹ ﺍﻟﺫﻱ ﻗﺩ ﻴﻅﻬﺭ ﻓﻲ ﺤﺎﻟﺔ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻤﻘﻁﻌﻴـﺔ‬
‫ﺍﻭ ﺤﺎﻟﺔ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﺯﻤﻨﻴﺔ‪.‬‬

‫‪ -2‬ﺘﻌﻁﻲ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻜﻔﺎﺀﺓ ﺍﻓﻀل ﻭﺯﻴﺎﺩﺓ ﻓﻲ ﺩﺭﺠﺎﺕ ﺍﻟﺤﺭﻴﺔ ﻭﻜﺫﻟﻙ ﺍﻗل ﺘﻌﺩﺩﻴﺔ‬
‫ﺨﻁﻴﺔ ﺒﻴﻥ ﺍﻟﻤﺘﻐﻴﺭﺍﺕ‪ ،‬ﻭﻤﺤﺘﻭﻯ ﻤﻌﻠﻭﻤﺎﺘﻲ ﺍﻜﺜﺭ ﺍﺫﺍ ﻤـﺎ ﺘـﻡ ﺍﺴـﺘﺨﺩﺍﻡ ﺍﻟﺒﻴﺎﻨـﺎﺕ‬
‫ﺍﻟﻤﻘﻁﻌﻴﺔ ﺃﻭ ﺍﻟﺯﻤﻨﻴﺔ‪.‬‬

‫ﻋﻨﺩﻤﺎ ﺘﻜﻭﻥ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ ﻤﻘﺎﺴﺔ ﻟﻨﻔﺱ ﺍﻟﻔﺘﺭﺍﺕ ﺍﻟﺯﻤﻨﻴﺔ ﻋﻨﺩﺌﺫ ﻴﻁﻠﻕ ﻋﻠﻰ‬
‫ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺒﺎﻨﻬﺎ ﺒﻴﺎﻨﺎﺕ ﻁﻭﻟﻴﺔ ﻤﺘﺯﻨﺔ )‪ (Balanced Panel Data‬ﻜﻤﺎ ﻓﻲ ﻤﺜﺎﻟﻨﺎ‬
‫ﺍﻟﺴﺎﺒﻕ‪ ،‬ﻭﻟﻭ ﻓﺭﻀﻨﺎ ﺒﺎﻥ ﻭﺍﺤﺩﺓ ﻤﻥ ﻫﺫﻩ ﺍﻟﻌﻭﺍﺌل ﻗﺩ ﺘﻡ ﺘﺴﺠﻴل ﻤﺸﺎﻫﺩﺍﺘﻬﺎ ﺍﻟﻤﻘﻁﻌﻴﺔ ﻟﺴﻨﺘﻴﻥ‬
‫ﻓﻘﻁ ﻭﺒﺎﻗﻲ ﺍﻟﻌﻭﺍﺌل ﻟﺜﻼﺙ ﺴﻨﻭﺍﺕ ﻋﻨﺩﺌﺫ ﻴﻁﻠﻕ ﻋﻠﻰ ﻫﺫﻩ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺒﺎﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻏﻴﺭ‬
‫ﺍﻟﻤﺘﺯﻨﺔ )‪ .(Unbalanced Panel Data‬ﻤﻥ ﺨﻼل ﻤﺜﺎﻟﻨﺎ ﺍﻟﺴﺎﺒﻕ ﻴﻤﻜﻨﻨﺎ ﺍﺴﺘﺨﺩﺍﻡ ﺍﺴﻠﻭﺏ‬
‫ﺍﻻﻨﺤﺩﺍﺭ ﻓﻲ ﺍﻟﺴﻼﺴل ﺍﻟﺯﻤﻨﻴﺔ ﻭﻟﺨﻤﺱ ﻤﺭﺍﺕ ) ﻟﻜل ﻋﺎﺌﻠﺔ ﻨﻤﻭﺫﺝ( ﺍﻭ ﻴﻤﻜﻨﻨﺎ ﺍﺴﺘﺨﺩﺍﻡ‬
‫ﺍﻻﻨﺤﺩﺍﺭ ﺍﻻﻋﺘﻴﺎﺩﻱ ﻋﻨﺩﻤﺎ ﺘﻜﻭﻥ ﺒﻴﺎﻨﺎﺘﻨﺎ ﻤﻘﻁﻌﻴﺔ ﻭﻟﺜﻼﺙ ﻤﺭﺍﺕ ) ﻟﻜل ﺴﻨﺔ ﻨﻤﻭﺫﺝ(‪.‬‬
 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[270‬‬

‫ﻫﻲ‪:‬‬ ‫ﺭﺌﻴﺴﻴﺔ‬ ‫ﺍﺸﻜﺎل‬ ‫ﺜﻼﺜﺔ‬ ‫ﻓﻲ‬ ‫ﺍﻟﻁﻭﻟﻴﺔ‬ ‫ﺍﻟﺒﻴﺎﻨﺎﺕ‬ ‫ﻨﻤﺎﺫﺝ‬ ‫ﺘﺎﺘﻲ‬ ‫ﻫﻨﺎ‬ ‫ﻤﻥ‬
‫ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ))‪ ، (Pooled Regression Model(PM‬ﻨﻤﻭﺫﺝ ﺍﻟﺘﺄﺜﻴﺭﺍﺕ‬
‫ﺍﻟﺜﺎﺒﺘﺔ‬
‫ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‬ ‫ﺍﻟﺜﺎﺜﻴﺭﺍﺕ‬ ‫ﻭﻨﻤﻭﺫﺝ‬ ‫‪(Fixed‬‬ ‫‪Effects‬‬ ‫))‪Model(FEM‬‬
‫))‪ . (Random Effects Model(REM‬ﻟﻴﻜﻥ ﻟﺩﻴﻨﺎ ‪ N‬ﻤﻥ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ ﻤﻘﺎﺴﺔ‬
‫ﻓﻲ ‪ T‬ﻤﻥ ﺍﻟﻔﺘﺭﺍﺕ ﺍﻟﺯﻤﻨﻴﺔ ﻓﺎﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻴﻌﺭﻑ ﺒﺎﻟﺼﻴﻐﺔ ﺍﻻﺘﻴﺔ‪:‬‬

‫‪k‬‬
‫‪y it = β o( i ) +‬‬ ‫‪∑ β j x j( it ) + ε it‬‬ ‫)‪, i = 1, 2 ,....., N t = 1, 2 ,...., T ……(1‬‬
‫‪j= 1‬‬

‫ﺤﻴﺙ ﺍﻥ ‪ y it‬ﺘﻤﺜل ﻗﻴﻤﺔ ﻤﺘﻐﻴﺭ ﺍﻻﺴﺘﺠﺎﺒﺔ ﻓﻲ ﺍﻟﻤﺸﺎﻫﺩﺓ ‪ i‬ﻋﻨﺩ ﺍﻟﻔﺘﺭﺓ ﺍﻟﺯﻤﻨﻴـﺔ ‪βo( i ) ، t‬‬

‫ﺘﻤﺜل ﻗﻴﻤﺔ ﻨﻘﻁﺔ ﺍﻟﺘﻘﺎﻁﻊ ﻓﻲ ﺍﻟﻤﺸﺎﻫﺩﺓ ‪ β j ، i‬ﺘﻤﺜل ﻗﻴﻤﺔ ﻤﻴل ﺨﻁ ﺍﻻﻨﺤﺩﺍﺭ‪ x j( it ) ،‬ﺜﻤﺜـل‬

‫ﻗﻴﻤﺔ ﺍﻟﻤﺘﻐﻴﺭ ﺍﻟﺘﻔﺴﻴﺭﻱ ‪ j‬ﻓﻲ ﺍﻟﻤﺸﺎﻫﺩﺓ ‪ i‬ﻋﻨﺩ ﺍﻟﻔﺘﺭﺓ ﺍﻟﺯﻤﻨﻴـﺔ ‪ ، t‬ﻭﺍﻥ ‪ ε it‬ﺘﻤﺜـل ﻗﻴﻤـﺔ‬
‫ﺍﻟﺨﻁﺄ ﻓﻲ ﺍﻟﻤﺸﺎﻫﺩﺓ ‪ i‬ﻋﻨﺩ ﺍﻟﻔﺘﺭﺓ ﺍﻟﺯﻤﻨﻴﺔ ‪ . t‬ﺘﺠﺩﺭ ﺍﻻﺸـﺎﺭﺓ ﻫﻨـﺎ ﺍﻟـﻰ ﺍﻥ ﺍﻟﻤﻘـﺼﻭﺩ‬
‫ﺒﺎﻟﻤﺸﺎﻫﺩﺓ ‪ i‬ﻫﻭ ﻋﺩﺩ ﺍﻟﻌﻭﺍﺌل ﻓﻲ ﻤﺜﺎﻟﻨﺎ ﺍﻟﺴﺎﺒﻕ‪ .‬ﺘﻌﺘﻤﺩ ﺘﻘﺩﻴﺭ ﺍﻟﻤﻌﻠﻤﺎﺕ ﻟﻠﻨﻤﻭﺫﺝ ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ‬
‫)‪ (1‬ﻋﻠﻰ ﻨﻭﻉ ﻨﻤﻭﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﻤﺴﺘﺨﺩﻡ‪.‬‬

‫‪ 2-1‬ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ‬

‫ﻴﻌﺘﺒﺭ ﻫﺫﺍ ﺍﻟﻨﻤﻭﺫﺝ ﻤﻥ ﺍﺒﺴﻁ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺤﻴﺙ ﺘﻜﻭﻥ ﻓﻴﻪ ﺠﻤﻴﻊ ﺍﻟﻤﻌـﺎﻤﻼﺕ‬
‫) ‪ βo( i‬ﻭ ‪ β j‬ﺜﺎﺒﺘﺔ ﻟﺠﻤﻴﻊ ﺍﻟﻔﺘﺭﺍﺕ ﺍﻟﺯﻤﻨﻴﺔ )ﻴﻬﻤل ﺍﻱ ﺘﺎﺜﻴﺭ ﻟﻠﺯﻤﻥ(‪ .‬ﺒﺎﻋﺎﺩﺓ ﻜﺘﺎﺒﺔ ﺍﻟﻨﻤـﻭﺫﺝ‬

‫ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (1‬ﻨﺤﺼل ﻋﻠﻰ ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ﻭﺒﺎﻟﺼﻴﻐﺔ ﺍﻻﺘﻴﺔ‪:‬‬

‫‪k‬‬
‫‪y it = β o + ∑ β j x j( it ) + ε it‬‬ ‫)‪, i = 1,2,....., N t = 1,2,...., T ……(2‬‬
‫‪j= 1‬‬
 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[271‬‬

‫ﺤﻴﺙ ﺍﻥ ‪ E(ε it ) = 0‬ﻭ ‪ . var(ε it ) = σ ε 2‬ﺘـﺴﺘﺨﺩﻡ ﻁﺭﻴﻘـﺔ ﺍﻟﻤﺭﺒﻌـﺎﺕ ﺍﻟـﺼﻐﺭﻯ‬

‫ﺍﻻﻋﺘﻴﺎﺩﻴﺔ ﻓﻲ ﺘﻘﺩﻴﺭ ﻤﻌﻠﻤﺎﺕ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (Greene, 2012) (2‬ﺒﻌﺩ ﺍﻥ ﺘﺭﺘـﺏ‬
‫ﺍﻟﻘﻴﻡ ﺍﻟﺨﺎﺼﺔ ﺒﻤﺘﻐﻴﺭ ﺍﻻﺴﺘﺠﺎﺒﺔ ﻭﺍﻟﻤﺘﻐﻴﺭ ﺍﻟﺘﻭﻀﻴﺤﻲ ﺒﺩﺀﺍ ﻤﻥ ﺍﻭل ﻤﺠﻤﻭﻋﺔ ﺒﻴﺎﻨﺎﺕ ﻤﻘﻁﻌﻴﺔ‬
‫ﻭﻫﻜﺫﺍ ﻭﺒﺤﺠﻡ ﻤﺸﺎﻫﺩﺍﺕ ﻤﻘﺩﺍﺭﻩ )‪. ( N ∗ T‬‬

‫‪ 2-2‬ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ‬

‫ﻓﻲ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻴﻜﻭﻥ ﺍﻟﻬﺩﻑ ﻫﻭ ﻤﻌﺭﻓﺔ ﺴﻠﻭﻙ ﻜـل ﻤﺠﻤﻭﻋـﺔ ﺒﻴﺎﻨـﺎﺕ‬
‫ﻤﻘﻁﻌﻴﺔ ﻋﻠﻰ ﺤﺩﺓ ﻤﻥ ﺨﻼل ﺠﻌل ﻤﻌﻠﻤﺔ ﺍﻟﻘﻁﻊ ‪ βo‬ﺘﺘﻔﺎﻭﺕ ﻤﻥ ﻤﺠﻤﻭﻋﺔ ﺍﻟﻰ ﺍﺨـﺭﻯ ﻤـﻊ‬
‫ﺒﻘﺎﺀ ﻤﻌﺎﻤﻼﺕ ﺍﻟﻤﻴل ‪ β j‬ﺜﺎﺒﺘﺔ ﻟﻜل ﻤﺠﻤﻭﻋﺔ ﺒﻴﺎﻨﺎﺕ ﻤﻘﻁﻌﻴﺔ ) ﺍﻱ ﺴﻭﻑ ﻨﺘﻌﺎﻤل ﻤﻊ ﺤﺎﻟـﺔ‬
‫ﻋﺩﻡ ﺍﻟﺘﺠﺎﻨﺱ ﻓﻲ ﺍﻟﺘﺒﺎﻴﻥ ﺒﻴﻥ ﺍﻟﻤﺠﺎﻤﻴﻊ(‪ ،‬ﻭﻋﻠﻴﻪ ﻓﺎﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻴﻜﻭﻥ ﺒﺎﻟﺼﻴﻐﺔ‬
‫ﺍﻻﺘﻴﺔ‪:‬‬

‫‪k‬‬
‫‪y it = β o ( i ) +‬‬ ‫‪∑ β j x j ( it ) + ε it‬‬ ‫‪, i = 1 , 2 ,....., N‬‬ ‫‪t = 1 , 2 ,...., T‬‬ ‫)‪…..(3‬‬
‫‪j= 1‬‬

‫ـﺼﻁﻠﺢ‬ ‫ـﺙ ﺍﻥ ‪ E(ε it ) = 0‬ﻭ ‪ .(Greene, 2012) var(ε it ) = σ ε 2‬ﻴﻘـ‬

‫ـﺼﺩ ﺒﻤـ‬ ‫ﺤﻴـ‬
‫ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺒﺎﻥ ﺍﻟﻤﻌﻠﻤﺔ ‪ βo‬ﻟﻜل ﻤﺠﻤﻭﻋﺔ ﺒﻴﺎﻨﺎﺕ ﻤﻘﻁﻌﻴﺔ ﻻﺘﺘﻐﻴـﺭ ﺨـﻼل ﺍﻟـﺯﻤﻥ‬
‫)‪ (time invariant‬ﻭﺍﻨﻤﺎ ﻴﻜﻭﻥ ﺍﻟﺘﻐﻴﺭ ﻓﻘﻁ ﻓﻲ ﻤﺠﺎﻤﻴﻊ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻤﻘﻁﻌﻴـﺔ ‪(Gujarati,‬‬
‫)‪ .2003‬ﻟﻐﺭﺽ ﺘﻘﺩﻴﺭ ﻤﻌﻠﻤﺎﺕ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (3‬ﻭﺍﻟـﺴﻤﺎﺡ ﻟﻤﻌﻠﻤـﺔ ﺍﻟﻘﻁـﻊ ‪βo‬‬
‫ﺒﺎﻟﺘﻐﻴﺭ ﺒﻴﻥ ﺍﻟﻤﺠﺎﻤﻴﻊ ﺍﻟﻤﻘﻁﻌﻴﺔ ﻋﺎﺩﺓ ﻤﺎ ﺘﺴﺘﺨﺩﻡ ﻤﺘﻐﻴﺭﺍﺕ ﻭﻫﻤﻴـﺔ ﺒﻘـﺩﺭ )‪ ( N − 1‬ﻟﻜـﻲ‬
‫ﻨﺘﺠﻨﺏ ﺤﺎﻟﺔ ﺍﻟﺘﻌﺩﺩﻴﺔ ﺍﻟﺨﻁﻴﺔ ﺍﻟﺘﺎﻤﺔ )‪ (Greene, 2012‬ﺜﻡ ﺘـﺴﺘﺨﺩﻡ ﻁﺭﻴﻘـﺔ ﺍﻟﻤﺭﺒﻌـﺎﺕ‬
‫ﺍﻟﺼﻐﺭﻯ ﺍﻻﻋﺘﻴﺎﺩﻴﺔ‪ .‬ﻴﻁﻠﻕ ﻋﻠﻰ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺍﺴﻡ ﻨﻤﻭﺫﺝ ﺍﻟﻤﺭﺒﻌﺎﺕ ﺍﻟـﺼﻐﺭﻯ‬
‫ﻟﻠﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﻭﻫﻤﻴـﺔ )‪ .(Least Squares Dummy Variable Model‬ﺒﻌـﺩ ﺍﻀـﺎﻓﺔ‬
‫ﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﻭﻫﻤﻴﺔ ‪ D‬ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (3‬ﻴﺼﺒﺢ ﺍﻟﻨﻤﻭﺫﺝ ﺒﺎﻟﺸﻜل ﺍﻻﺘﻲ‪:‬‬
 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[272‬‬

‫‪N‬‬ ‫‪k‬‬
‫‪y it = α 1 +‬‬ ‫‪∑ α dDd‬‬ ‫‪+‬‬ ‫‪∑ β j x j( it ) + ε it‬‬ ‫‪, i = 1, 2 ,....., N t = 1, 2 ,...., T‬‬ ‫)‪…(4‬‬
‫‪d=2‬‬ ‫‪j= 1‬‬

‫‪N‬‬
‫‪ α1 +‬ﺍﻟﺘﻐﻴﺭ ﻓﻲ ﺍﻟﻤﺠﺎﻤﻴﻊ ﺍﻟﻤﻘﻁﻌﻴﺔ ﻟﻤﻌﻠﻤـﺔ ﺍﻟﻘﻁـﻊ ‪. βo‬‬ ‫‪∑ αd Dd‬‬ ‫ﺤﻴﺙ ﻴﻤﺜل ﺍﻟﻤﻘﺩﺍﺭ‬
‫‪d= 2‬‬

‫ﻭﻴﻤﻜﻥ ﻜﺘﺎﺒﺔ ﺍﻟﻨﻤﻭﺫﺝ ﺒﺎﻟﻤﻌﺎﺩﻟﺔ )‪ (4‬ﺒﻌﺩ ﺤﺫﻑ ‪ α1‬ﺒﺎﻟﺸﻜل ﺍﻻﺘـﻲ )‪، (Gujarati, 2003‬‬
‫)‪:(Greene, 2012‬‬

‫‪N‬‬ ‫‪k‬‬
‫= ‪y it‬‬ ‫‪∑ α dDd‬‬ ‫‪+‬‬ ‫‪∑ β j x j( it ) + ε it‬‬ ‫‪, i = 1, 2 ,....., N t = 1, 2 ,...., T‬‬ ‫)‪…..(5‬‬
‫‪d =1‬‬ ‫‪j= 1‬‬

‫‪ 2-3‬ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‬

‫ﻓﻲ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻴﻜﻭﻥ ﺤﺩ ﺍﻟﺨﻁﺄ ‪ ε it‬ﺫﺍ ﺘﻭﺯﻴﻊ ﻁﺒﻴﻌﻲ ﺒﻭﺴـﻁ ﻤﻘـﺩﺭﺍﻩ‬
‫ﺼﻔﺭ ﻭﺘﺒﺎﻴﻥ ﻤﺴﺎﻭﻱ ﺍﻟﻰ ‪ ، σ ε 2‬ﻭﻟﻜﻲ ﺘﻜﻭﻥ ﻤﻌﻠﻤﺎﺕ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘـﺔ ﺼـﺤﻴﺤﺔ‬
‫ﻭﻏﻴﺭ ﻤﺘﺤﻴﺯﺓ ﻋﺎﺩﺓ ﻤﺎ ﻴﻔﺭﺽ ﺒﺎﻥ ﺘﺒﺎﻴﻥ ﺍﻟﺨﻁﺄ ﺜﺎﺒﺕ )ﻤﺘﺠﺎﻨﺱ( ﻟﺠﻤﻴﻊ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ ﺍﻟﻤﻘﻁﻌﻴﺔ‬
‫ﻭﻟﻴﺱ ﻫﻨﺎﻙ ﺍﻱ ﺍﺭﺘﺒﺎﻁ ﺫﺍﺘﻲ ﺨﻼل ﺍﻟﺯﻤﻥ ﺒﻴﻥ ﻜل ﻤﺠﻤﻭﻋـﺔ ﻤـﻥ ﻤﺠـﺎﻤﻴﻊ ﺍﻟﻤـﺸﺎﻫﺩﺍﺕ‬
‫ﺍﻟﻤﻘﻁﻌﻴﺔ ﻓﻲ ﻓﺘﺭﺓ ﺯﻤﻨﻴﺔ ﻤﺤﺩﺩﺓ‪ .‬ﻴﻌﺘﺒﺭ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻨﻤﻭﺫﺠﺎ ﻤﻼﺌﻤﺎ ﻓﻲ ﺤﺎﻟﺔ‬
‫ﻭﺠﻭﺩ ﺨﻠل ﻓـﻲ ﺍﺤـﺩ ﺍﻟﻔـﺭﻭﺽ ﺍﻟﻤـﺫﻜﻭﺭﺓ ﺍﻋـﻼﻩ ﻓـﻲ ﻨﻤـﻭﺫﺝ ﺍﻟﺘـﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘـﺔ‬
‫)‪ .(Gujarati, 2003‬ﻓﻲ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﺴﻭﻑ ﻴﻌﺎﻤل ﻤﻌﺎﻤـل ﺍﻟﻘﻁـﻊ ) ‪βo( i‬‬
‫ﻜﻤﺘﻐﻴﺭ ﻋﺸﻭﺍﺌﻲ ﻟﻪ ﻤﻌﺩل ﻤﻘﺩﺍﺭﻩ ‪ µ‬ﺍﻱ‪:‬‬

‫‪β o( i ) = µ + v i‬‬ ‫‪, i = 1,2,....., N‬‬ ‫)‪…..(6‬‬

‫ﻭﺒﺘﻌﻭﻴﺽ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (6‬ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (3‬ﻨﺤﺼل ﻋﻠﻰ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻭﺒﺎﻟﺸﻜل‬
‫ﺍﻻﺘﻲ‪:‬‬
 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[273‬‬

‫‪k‬‬
‫‪y it = µ + ∑ β j x j( it ) + v i + ε it‬‬ ‫)‪, i = 1,2,....., N t = 1,2,...., T …..(7‬‬
‫‪j= 1‬‬

‫ﺤﻴﺙ ﺍﻥ ‪ v i‬ﻴﻤﺜل ﺤﺩ ﺍﻟﺨﻁﺄ ﻓﻲ ﻤﺠﻤﻭﻋﺔ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻤﻘﻁﻌﻴـﺔ ‪ . i‬ﻴﻁﻠـﻕ ﻋﻠـﻰ ﻨﻤـﻭﺫﺝ‬

‫ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﺍﺤﻴﺎﻨﺎ ﻨﻤـﻭﺫﺝ ﻤﻜﻭﻨـﺎﺕ ﺍﻟﺨﻁـﺄ )‪(Error Components Model‬‬
‫ﺒﺴﺒﺏ ﺍﻥ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (7‬ﻴﺤﻭﻱ ﻤﺭﻜﺒﻴﻥ ﻟﻠﺨﻁﺄ ﻫﻤﺎ ‪ v i‬ﻭ ‪ . ε it‬ﻴﻤﺘﻠﻙ ﻨﻤـﻭﺫﺝ‬
‫ـﺎ ﺍﻥ ‪، var(ε it ) = σ ε 2 ، E(ε it ) = 0‬‬
‫ـﻴﺔ ﻤﻨﻬـ‬
‫ـﻭﺍﺹ ﺭﻴﺎﻀـ‬
‫ـﺸﻭﺍﺌﻴﺔ ﺨـ‬
‫ـﺎﺜﻴﺭﺍﺕ ﺍﻟﻌـ‬
‫ﺍﻟﺘـ‬
‫‪. var( v i ) = σ v 2 ، E( v i ) = 0‬‬

‫ﻟﻴﻜﻥ ﻟﺩﻴﻨﺎ ﺤﺩ ﺍﻟﺨﻁﺄ ﺍﻟﻤﺭﻜﺏ ﺍﻷﺘﻲ‪:‬‬

‫‪w it = v i + ε it‬‬ ‫)‪……(8‬‬

‫ﺤﻴﺙ ﺍﻥ‪:‬‬

‫)‪E( w it ) = 0 ……(9‬‬

‫‪var( w it ) = σ v 2 + σ ε 2‬‬ ‫)‪……(10‬‬

‫ﺘﻔﺸل ﻁﺭﻴﻘﺔ ﺍﻟﻤﺭﺒﻌﺎﺕ ﺍﻟﺼﻐﺭﻯ ﺍﻻﻋﺘﻴﺎﺩﻴﺔ ﻓﻲ ﺘﻘﺩﻴﺭ ﻤﻌﻠﻤـﺎﺕ ﻨﻤـﻭﺫﺝ ﺍﻟﺘـﺎﺜﻴﺭﺍﺕ‬

‫ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻜﻭﻨﻬﺎ ﺘﻌﻁﻲ ﻤﻘﺩﺭﺍﺕ ﻏﻴﺭ ﻜﻔﻭﺀﺓ ﻭﻟﻬﺎ ﺍﺨﻁﺎﺀ ﻗﻴﺎﺴﻴﺔ ﻏﻴﺭ ﺼﺤﻴﺤﺔ ﻤﻤﺎ ﻴﺅﺜﺭ ﻓﻲ‬
‫ﺍﺨﺘﺒﺎﺭ ﺍﻟﻤﻌﻠﻤﺎﺕ ﻜﻭﻥ ﺍﻥ ﺍﻟﺘﺒﺎﻴﻥ ﺍﻟﻤﺸﺘﺭﻙ ﺒﻴﻥ ‪ w it‬ﻭ ‪ w is‬ﻻﻴﺴﺎﻭﻱ ﺍﻟﺼﻔﺭ ﺍﻱ‪:‬‬

‫‪cov( w it , w is ) = σ 2 v ≠ 0‬‬ ‫‪, t≠s‬‬ ‫)‪……(11‬‬

 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[274‬‬

‫ﻟﻐﺭﺽ ﺘﻘﺩﻴﺭ ﻤﻌﻠﻤﺎﺕ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﺒﺸﻜل ﺼﺤﻴﺢ ﻋﺎﺩﺓ ﻤﺎ ﺘﺴﺘﺨﺩﻡ ﻁﺭﻴﻘـﺔ‬
‫ﺍﻟﻤﺭﺒﻌﺎﺕ ﺍﻟـﺼﻐﺭﻯ ﺍﻟﻤﻌﻤـﺔ ))‪(Greene, (Generalized Least Squares (GLS‬‬
‫)‪. 2012‬‬

‫‪ 3-1‬ﺃﺴﺎﻟﻴﺏ ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ﻟﻠﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ‬

‫ﻜﻤﺎ ﺫﻜﺭﻨﺎ ﺴﺎﺒﻘﺎ ﺒﻭﺠﻭﺩ ﺜﻼﺜﺔ ﻨﻤﺎﺫﺝ ﺭﺌﻴﺴﻴﺔ ﻤﻥ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻭﻋﻠـﻰ ﻫـﺫﺍ‬
‫ﺍﻻﺴﺎﺱ ﻴﻁﺭﺡ ﺍﻟﺴﺅﺍل ﺍﻷﺘﻲ‪ :‬ﻤﺎ ﻫﻭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻷﻜﺜﺭ ﻤﻼﺀﻤﺔ ﻟﺒﻴﺎﻨﺎﺕ ﺩﺭﺍﺴﺔ ﻤﺎ؟ ﻟﻐـﺭﺽ‬
‫ﺍﻻﺠﺎﺒﺔ ﻋﻥ ﻤﺜل ﻫﻜﺫﺍ ﺘﺴﺎﺅل ﺴﻭﻑ ﻨﻘﻭﻡ ﺒﻌﺭﺽ ﺍﺴﻠﻭﺒﻴﻥ‪ ،‬ﺍﻻﻭل‪ :‬ﺍﺴﻠﻭﺏ ﺍﻻﺨﺘﻴـﺎﺭ ﺒـﻴﻥ‬
‫ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭﺍﻟﺜﺎﻨﻲ‪ :‬ﻫﻭ ﺍﺴـﻠﻭﺏ ﺍﻻﺨﺘﻴـﺎﺭ ﺒـﻴﻥ‬
‫ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‪.‬‬

‫ﻟﻐﺭﺽ ﺍﻻﺨﺘﻴﺎﺭ ﺒﻴﻥ ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻨﺴﺘﺨﺩﻡ ﺍﺨﺘﺒـﺎﺭ‬
‫‪ F‬ﺍﻟﻤﻘﻴﺩ ﻭﺒﺎﻟﺼﻴﻐﺔ ﺍﻻﺘﻴﺔ‪:‬‬

‫)‪( R FEM 2 − R PM 2 ) / ( N − 1‬‬

‫= ) ‪F( N − 1, NT − N − k‬‬ ‫)‪…..(12‬‬
‫) ‪(1 − R FEM 2 ) / ( NT − N − k‬‬

‫ﺤﻴﺙ ﺍﻥ ‪ k‬ﻫﻲ ﻋﺩﺩ ﺍﻟﻤﻌﻠﻤﺎﺕ ﺍﻟﻤﻘﺩﺭﺓ ﻭﺍﻥ ‪ R FEM‬ﻴﻤﺜل ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﻋﻨـﺩ ﺍﺴـﺘﺨﺩﺍﻡ‬
‫ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭ ‪ R PM‬ﻴﻤﺜل ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﻋﻨـﺩ ﺍﺴـﺘﺨﺩﺍﻡ ﻨﻤـﻭﺫﺝ ﺍﻻﻨﺤـﺩﺍﺭ‬
‫ﺍﻟﺘﺠﻤﻴﻌﻲ‪ .‬ﺘﻘﺎﺭﻥ ﻨﺘﻴﺠﺔ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (12‬ﻤﻊ ) ‪ F(α, N − 1, Nt − N − k‬ﻓﺈﺫﺍ ﻜﺎﻨﺕ ﻗﻴﻤـﺔ‬
‫ﺍﻟﻤﻌﺎﺩﻟــــﺔ )‪ (12‬ﺍﻜﺒــــﺭ ﺍﻭ ﻤــــﺴﺎﻭﻴﺔ ﺍﻟــــﻰ ﺍﻟﻘﻴﻤــــﺔ ﺍﻟﺠﺩﻭﻟﻴــــﺔ‬
‫) ﺍﻭ ﺍﺫﺍ ﻜﺎﻨﺕ ﻗﻴﻤﺔ ‪ p − value‬ﺍﻗل ﻤﻥ ﺍﻭ ﺘﺴﺎﻭﻱ ‪ ( 0.05‬ﻋﻨﺩﺌﺫ ﻓﺎﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺘـﺎﺜﻴﺭﺍﺕ‬
‫ﺍﻟﺜﺎﺒﺘﺔ ﻫﻭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﺩﺭﺍﺴﺔ )‪ .(Greene, 2012‬ﺒﻌـﺩ ﺍﺨﺘﻴـﺎﺭ ﻨﻤـﻭﺫﺝ‬
 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[275‬‬

‫ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺒﻭﺼﻔﻪ ﻨﻤﻭﺫﺠﹰﺎ ﻤﻼﺌﻤﹰﺎ ﻨﻘﻭﻡ ﺒﺎﻻﺨﺘﻴﺎﺭ ﺒﻴﻨـﻪ ﻭﺒـﻴﻥ ﻨﻤـﻭﺫﺝ ﺍﻟﺘـﺎﺜﻴﺭﺍﺕ‬
‫ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻟﺘﺤﺩﻴﺩ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻨﻬـﺎﺌﻲ ﺍﻟﻤﻼﺌـﻡ ﻟﺒﻴﺎﻨـﺎﺕ ﺍﻟﺩﺭﺍﺴـﺔ ﻤـﻥ ﺨـﻼل ﺍﺴـﺘﺨﺩﺍﻡ‬
‫ﺍﺨﺘﺒﺎﺭ)‪ Hausman(H‬ﺤﻴﺙ ﺘﻜﻭﻥ ﻓﺭﻀﻴﺔ ﺍﻟﻌﺩﻡ ﺒﺎﻟﺸﻜل ﺍﻻﺘﻲ‪:‬‬

‫ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻫﻭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ‪H 0 :‬‬

‫ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻫﻭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ‪H1 :‬‬

‫ﻭﺘﻜﻭﻥ ﺼﻴﻐﺔ ﺍﻻﺨﺘﺒﺎﺭ ﺒﺎﻟﺼﻴﻐﺔ ﺍﻻﺘﻴﺔ‪:‬‬

‫) ‪H = (βˆ FEM − βˆ REM )′ [ var(βˆ FEM ) − var(βˆ REM )]−1 (βˆ FEM − βˆ REM‬‬
‫)‪…….(13‬‬

‫ﺤﻴﺙ ﺍﻥ ) ‪ var(βˆ FEM‬ﻫﻭ ﻤﺘﺠﻪ ﺍﻟﺘﺒﺎﻴﻥ ﻟﻤﻌﻠﻤﺎﺕ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭ ) ‪var(βˆ REM‬‬
‫ﻫﻭ ﻤﺘﺠﻪ ﺍﻟﺘﺒﺎﻴﻥ ﻟﻤﻌﻠﻤﺎﺕ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‪ .‬ﺤﻴﺙ ﺍﻥ ﻫﺫﻩ ﺍﻻﺤﺼﺎﺌﻴﺔ ﻟﻬﺎ ﺘﻭﺯﻴﻊ‬
‫ﻤﺭﺒﻊ ﻜﺎﻱ ﻭﺒﺩﺭﺠﺔ ﺤﺭﻴﺔ ﻤﻘﺩﺍﺭﻫﺎ ‪ . k‬ﻴﻜﻭﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻫﻭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌـﻡ‬
‫ﺍﺫﺍ ﻜﺎﻨﺕ ﻗﻴﻤﺔ ﺍﻻﺤﺼﺎﺌﻴﺔ ﺍﻜﺒﺭ ﻤﻥ ﻗﻴﻤﺔ ﻤﺭﺒﻊ ﻜﺎﻱ ﻭﻋﻠﻰ ﺍﻟﻌﻜﺱ ﺴﻭﻑ ﻴﻜـﻭﻥ ﺍﻟﻨﻤـﻭﺫﺝ‬
‫ﺍﻟﻤﻼﺌﻡ ﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﺩﺭﺍﺴﺔ ﻫﻭ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ )‪.(Hausman, 1978‬‬

‫‪-4‬ﻤﻌﺎﻴﻴﺭ ﺍﺨﺘﻴﺎﺭ ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ‬

‫ﻓﻲ ﻜﺜﻴﺭ ﻤﻥ ﺍﻟﺘﻁﺒﻴﻘﺎﺕ ﺍﻻﺤﺼﺎﺌﻴﺔ ﻴﺭﻏﺏ ﺼﺎﻨﻊ ﺍﻟﻘﺭﺍﺭ ﻓﻲ ﺍﻟﻭﺼﻭل ﺇﻟـﻰ ﺃﻓـﻀل‬
‫ﻨﻤﻭﺫﺝ ﻴﺤﻜﻡ ﺍﻟﺒﻴﺎﻨﺎﺕ ﻗﻴﺩ ﺍﻟﺩﺭﺍﺴﺔ ﻋﻥ ﻁﺭﻴﻕ ﺍﻴﺠﺎﺩ ﺍﻓﻀل ﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﺘﻲ ﺘﺤـﺩﺩ ﺍﻓـﻀل‬
‫ﻨﻤﻭﺫﺝ ﻴﻤﻜﻥ ﺒﻨﺎﺀ ﺍﻟﻘﺭﺍﺭ ﺍﻻﻤﺜل ﻋﻠﻴﻪ‪ .‬ﺃﻥ ﺍﺴﻠﻭﺏ ﺍﺨﺘﻴﺎﺭ ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ ﺒﻌﺩ ﺘﻭﻓﻴـﻕ‬
‫ﻨﻤﻭﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻴﺤﺘﺎﺝ ﺍﻟﻰ ﺍﻨﺘﺒﺎﻩ ﺸﺩﻴﺩ ﺤﻴﺙ ﺴﺒﻕ ﻭﺍﻥ ﺘﻁﺭﻗﻨﺎ ﻓﻲ ﺍﻟﻤﺒﺤﺙ ﺍﻟـﺴﺎﺒﻕ‬
‫ﺇﻟﻰ ﻜﻴﻔﻴﺔ ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ﻟﻠﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﻤﻥ ﺒﻴﻥ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﺜﻼﺜـﺔ‬
‫ﻤﻥ ﺤﻴﺙ ﺍﺠﺭﺍﺀ ﻫﺫﻩ ﺍﻻﺨﺘﺒﺎﺭﺍﺕ ﻟﻠﺘﺎﻜﺩ ﻤﻥ ﺍﻥ ﺍﺨﻴﺘﺎﺭﻨﺎ ﻟﻠﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ﻴﺒﻘﻰ ﻜﻤﺎ ﻫﻭ ﺍﺜﻨﺎﺀ‬
‫ﺍﺨﺘﻴﺎﺭ ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ‪ .‬ﻓﻲ ﺒﺤﺜﻨﺎ ﻫﺫﺍ ﺴﻭﻑ ﻨﺘﻁﺭﻕ ﺍﻟﻰ ﺍﺴـﺘﺨﺩﺍﻡ ﻤﻌﻴـﺎﺭﻴﻥ ﻫﻤـﺎ‪:‬‬
 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[276‬‬

‫ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴـﺩ ﺍﻟﻤـﺼﺤﺢ ) ) ‪ (Adjusted R 2 ( R 2adj‬ﻭﻤﻌﻴـﺎﺭ ﺃﻜـﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤـﺎﺕ‬

‫))‪ (Akaike Information Criteria (AIC‬ﺤﻴﺙ ﻴﻌﺘﻤﺩ ﺍﻟﻤﻌﻴﺎﺭ ﺍﻻﻭل ﻋﻠـﻰ ﺍﺨﺘﻴـﺎﺭ‬
‫ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ ﺫﻱ ﺍﻋﻠﻰ ﻗﻴﻤﺔ ﺒﻴﻨﻤﺎ ﻴﻌﺘﻤﺩ ﺍﻟﻤﻌﻴﺎﺭ ﺍﻟﺜﺎﻨﻲ ﻋﻠﻰ ﺍﺨﺘﻴﺎﺭ ﺍﻓﻀل ﻨﻤـﻭﺫﺝ‬
‫ﺠﺯﺌﻲ ﺫﻱ ﺍﻗل ﻗﻴﻤﺔ)‪(Shao ،(Lee & Russell, 2004) ،(Hughes & King, 2003‬‬
‫)‪.et al,2009‬‬

‫ﻴﻌﺭﻑ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﻤﺼﺤﺢ ﺒﺎﻟﺼﻴﻐﺔ ﺍﻻﺘﻴﺔ‪:‬‬

‫‪NT − 1‬‬
‫[ ‪R 2adj = 1 −‬‬ ‫)‪(1 − R 2panel data mod el )] …….(14‬‬
‫‪NT − k − 1‬‬

‫ﺃﻤﺎ ﻤﻌﻴﺎﺭ ﺃﻜﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤﺎﺕ ﻓﻴﻌﺭﻑ ﺒﺎﻟﺸﻜل ﺍﻻﺘﻲ‪:‬‬

‫)‪AIC = −2 log(max imized likehood) + 2k …….(15‬‬

‫‪ -5‬ﺍﻟﺠﺎﻨﺏ ﺍﻟﺘﻁﺒﻴﻘﻲ‬

‫ﻤﻤﺎ ﻻﺸﻙ ﻓﻴﻪ ﺍﻥ ﻨﻭﻋﻴﺔ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﺘﻲ ﺘﻼﺌﻡ ﺍﻟﺠﺎﻨﺏ ﺍﻟﺘﻁﺒﻴﻘﻲ ﻤﻥ ﺠﺎﻨﺏ ﻭﺩﻗﺘﻬـﺎ ﻤـﻥ‬
‫ﺠﺎﻨﺏ ﺍﺨﺭ ﻴﻌﺩﺍﻥ ﺭﻜﻴﺯﺓ ﺃﺴﺎﺴﻴﺔ ﻓﻲ ﺍﻟﻭﺼﻭل ﺍﻟﻰ ﻨﺘﺎﺌﺞ ﻴﻤﻜﻥ ﺍﻻﻋﺘﻤﺎﺩ ﻋﻠﻴﻬـﺎ ﻭﺍﻟﺭﻜـﻭﻥ‬
‫ﺍﻟﻴﻬﺎ ﻓﻲ ﺍﻱ ﻋﻤل ﺒﺤﺜﻲ‪ .‬ﻤﻥ ﻫﻨﺎ ﺠﺎﺀﺕ ﻓﻜﺭﺓ ﺍﻟﺨﺭﻭﺝ ﻋﻥ ﺍﻟﻤﺅﻟﻭﻑ ﻓﻲ ﺍﻏﻠﺏ ﺍﻟﺩﺭﺍﺴـﺎﺕ‬
‫ﻭﺍﻟﺘﻁﺭﻕ ﺍﻟﻰ ﺍﻟﺠﺎﻨﺏ ﺍﻟﻤﺎﻟﻲ ﻤﻥ ﺨﻼل ﺩﺭﺍﺴﺔ ﺍﻟﻤﻭﺍﺯﻨﺔ ﺍﻟﻌﺎﻤـﺔ ﻟﻠﺩﻭﻟـﺔ ﻤﻘﺎﺴـﺔ ﺒﻤﻼﻴـﻴﻥ‬
‫ﺍﻟﺩﻭﻻﺭﺍﺕ ﻜﻤﺘﻐﻴﺭ ﺍﺴﺘﺠﺎﺒﺔ ) ‪ ( y‬ﻭﺘﺄﺜﻴﺭﻫﺎ ﻓﻲ ﻋﺩﺩ ﻤﻥ ﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﺘﻔﺴﻴﺭﻴﺔ ﻭﺤﺴﺏ ﻭﺠﻬﺔ‬
‫ﻨﻅﺭ ﺍﻫل ﺍﻻﺨﺘﺼﺎﺹ ﻭﻫﻲ ﺍﻟﻤﻌﺩل ﺍﻟﺴﻨﻭﻱ ﻟﺴﻌﺭ ﺍﻟﻨﻔﻁ ﻟﺴﻠﺔ ﺨﺎﻤﺎﺕ ﺍﻭﺒﻙ ﻤﻘﺎﺴﺔ ﺒﺎﻟﺩﻭﻻﺭ‬
‫ﺍﻻﻤﺭﻴﻜﻲ ) ‪ ، ( x1‬ﻭﺍﻟﻌﺠﺯ ﻓﻲ ﻤﻴﺯﺍﻥ ﺍﻟﻤﺩﻓﻭﻋﺎﺕ ﻭﻴﻜﻭﻥ ﻤﻘﺎﺴﹰﺎ ﺒﻤﻼﻴﻴﻥ ﺍﻟـﺩﻭﻻﺭﺍﺕ ) ‪، ( x 2‬‬
‫ﺤﺠﻡ ﺍﻟﺴﻜﺎﻥ ﻤﻘﺎﺴﺎ ﺒﺎﻟﻤﻠﻴﻭﻥ ﻨﺴﻤﺔ ) ‪ ، ( x 3‬ﻭﺍﺨﻴﺭﺍ ﻤﻌﺩل ﺍﻟﺘﻀﺨﻡ ) ‪ . ( x4‬ﺸـﻤﻠﺕ ﺍﻟﺩﺭﺍﺴـﺔ‬
‫ﺍﻟﺩﻭل ﺍﻟﻌﺭﺒﻴﺔ ﺍﻟﺒﺎﻟﻎ ﻋﺩﺩﻫﺎ ‪ 18‬ﺩﻭﻟﺔ ﻜﺒﻴﺎﻨﺎﺕ ﻤﻘﻁﻌﻴﺔ )‪ ( N = 18‬ﺒﺎﺴﺘﺜﻨﺎﺀ ﺍﻟﻌﺭﺍﻕ ﺍﻟﺫﻱ ﻟـﻡ‬
‫ﺘﺘﻭﺍﻓﺭ ﺍﻹﺤﺼﺎﺌﻴﺎﺕ ﻜﺎﻤل ﻋﻨﻪ ﻭﺍﻟﺼﻭﻤﺎل ﻭﻓﻠﺴﻁﻴﻥ ﻻﻥ ﻟﻴﺱ ﻫﻨـﺎﻙ ﺩﻗـﺔ ﻓـﻲ ﺘـﺴﺠﻴل‬
‫ـﻭﻨﺱ)‪،(d‬‬
‫ـﺭﻴﻥ)‪ ،(c‬ﺘـ‬
‫ـﺎﺭﺍﺕ)‪ ،(b‬ﺍﻟﺒﺤـ‬
‫ـﻲ ) ﺍﻻﺭﺩﻥ)‪ ،(a‬ﺍﻻﻤـ‬
‫ـﺩﻭل ﻫـ‬
‫ـﺼﺎﺌﻴﺎﺕ ﻭﺍﻟـ‬
‫ﺍﻻﺤـ‬
 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[277‬‬

‫ﺍﻟﺠﺯﺍﺌﺭ)‪ ،(e‬ﺠﻴﺒﻭﺘﻲ)‪ ،(f‬ﺍﻟﺴﻌﻭﺩﻴﺔ)‪ ،(g‬ﺍﻟﺴﻭﺩﺍﻥ)‪ ،(h‬ﺴﻭﺭﻴﺔ)‪ ،(i‬ﻋﻤـﺎﻥ)‪ ،(j‬ﻗﻁـﺭ)‪،(k‬‬

‫ﺍﻟﻜﻭﻴﺕ)‪ ،(l‬ﻟﺒﻨﺎﻥ)‪ ،(m‬ﻟﻴﺒﻴﺎ)‪ ،(n‬ﻤﺼﺭ)‪ ،(o‬ﺍﻟﻤﻐﺭﺏ)‪ ،(p‬ﻤﻭﺭﻴﺘﺎﻨﻴﺎ)‪ ،(q‬ﺍﻟﻴﻤﻥ)‪ ((r‬ﻭﻟﻠﻔﺘﺭﺓ‬
‫ﺍﻟﺯﻤﻨﻴﺔ ﻤﻥ ‪ 1995‬ﺍﻟﻰ ‪ (T = 15) 2009‬ﻭﺒﺤﺠﻡ ﻋﻴﻨﺔ ‪ 270‬ﻤﺸﺎﻫﺩﺓ )‪ ( N ∗ T = 270‬ﺘﻡ‬
‫ﺍﻟﺤﺼﻭل ﻋﻠﻰ ﺒﻴﺎﻨﺎﺕ ﺍﻟﺩﺭﺍﺴﺔ ﺒﺎﻻﻋﺘﻤﺎﺩ ﻋﻠﻰ ﺍﻟﺘﻘﺭﻴﺭ ﺍﻻﻗﺘﺼﺎﺩﻱ ﺍﻟﻌﺭﺒﻲ ﺍﻟﻤﻭﺤـﺩ ﻟـﺴﻨﺔ‬
‫‪) 2010‬ﺼﻨﺩﻭﻕ ﺍﻟﻨﻘﺩ ﺍﻟﻌﺭﺒﻲ ﻭﺍﺨﺭﻭﻥ‪ .(2010 ،‬ﻭﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﺭﻴﺎﻀﻲ ﺍﻟﻤﻘـﺩﺭ ﻟﻠﺩﺭﺍﺴـﺔ‬
‫ﺤﺴﺏ ﺍﻟﻤﻌﺎﺩﻟﺔ )‪ (1‬ﻴﻜﻭﻥ ﺒﺎﻟﺼﻴﻐﺔ ﺍﻻﺘﻴﺔ‪:‬‬

‫‪4‬‬
‫) ‪ŷ it = βˆ o( i ) + ∑ βˆ j x j( it‬‬ ‫‪, i = 1,2,.....,18 t = 1,2,....,15‬‬
‫‪j=1‬‬

‫‪ 5-1‬ﺘﻘﺩﻴﺭ ﻤﻌﻠﻤﺎﺕ ﺍﻟﻨﻤﻭﺫﺝ‬

‫ﻓﻲ ﻫﺫﺍ ﺍﻟﻤﺒﺤﺙ ﺴﻭﻑ ﻨﻘﺩﺭ ﻤﻌﻠﻤﺎﺕ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﺩﺭﻭﺱ ﺒﺎﺴﺘﺨﺩﺍﻡ ﻨﻤـﺎﺫﺝ ﺍﻟﺒﻴﺎﻨـﺎﺕ‬
‫ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﺜﻼﺜﺔ ﻭﻫﻲ ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ‪ ،‬ﻭﻨﻤـﻭﺫﺝ ﺍﻟﺘـﺄﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘـﺔ‪ ،‬ﻭﻨﻤـﻭﺫﺝ‬
‫ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‪ .‬ﺍﻟﺠﺩﻭل)‪ (2‬ﻴﻭﻀﺢ ﻨﺘﺎﺌﺞ ﺍﻟﺘﻘﺩﻴﺭ ﻟﻠﻤﻌﻠﻤﺎﺕ ﻋﻨﺩ ﺍﺴﺘﺨﺩﺍﻡ ﺍﻟﻨﻤﺎﺫﺝ ﺍﻟﺜﻼﺜﺔ‬
‫ﺒﺎﻻﻋﺘﻤﺎﺩ ﻋﻠﻰ ﺒﺭﻨﺎﻤﺞ ‪ .EViews 7‬ﻓﻲ ﻀﻭﺀ ﻨﺘﺎﺌﺞ ﺘﻘﺩﻴﺭ ﺍﻟﻤﻌﻠﻤﺎﺕ ﺍﻟﻤﻭﻀﺤﺔ ﺒﺎﻟﺠـﺩﻭل‬
‫)‪ (2‬ﻨﻼﺤﻅ ﺒﺎﻥ ﻗﻴﻡ ﻤﻌﺎﻤل ﺍﻟﻘﻁﻊ )‪ (constant‬ﺘﺨﺘﻠﻑ ﻤﻥ ﺩﻭﻟﺔ ﺍﻟﻰ ﺍﺨﺭﻯ ﺤﻴـﺙ ﻴﻤﺜـل‬
‫ﺍﻟﻤﻘــﺩﺍﺭ )‪ (-639.609‬ﻗﻴﻤــﺔ ﻤﻌﺎﻤــل ﺍﻟﻘﻁــﻊ ﻟﺩﻭﻟــﺔ ﺍﻻﺭﺩﻥ ﻓــﻲ ﺤــﻴﻥ ﺘﺒﻠــﻎ‬
‫ﻗﻴﻤــــﺔ ﺍﻻﺨــــﺘﻼﻑ ﻓــــﻲ ﺩﻭﻟــــﺔ ﺍﻻﻤــــﺎﺭﺍﺕ )‪ (-10.341‬ﺍﻱ‬
‫)‪ (-639.609+629.268‬ﻭﻫﻜﺫﺍ ﻟﺒﺎﻗﻲ ﺍﻟﺩﻭل ﺍﻻﺨﺭﻯ ﻭﻴﻌﻭﺩ ﺴﺒﺏ ﻫﺫﺍ ﺍﻻﺨﺘﻼﻑ ﻓﻲ ﻗـﻴﻡ‬
‫ﻤﻌﺎﻤل ﺍﻟﻘﻁﻊ ﺍﻟﻰ ﺨﺼﻭﺼﻴﺔ ﻜل ﺩﻭﻟﺔ‪ .‬ﻤﻥ ﺍﻟﺠﺩﻭل )‪ (2‬ﻴﻤﻜـﻥ ﻜﺘﺎﺒـﺔ ﻤﻌﺎﺩﻟـﺔ ﻨﻤـﻭﺫﺝ‬
‫ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ﺍﻟﻤﻘﺩﺭﺓ‪ ،‬ﻤﻌﺎﺩﻟﺔ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺍﻟﻤﻘﺩﺭﺓ‪ ،‬ﻭﻨﻤـﻭﺫﺝ ﺍﻟﺘـﺎﺜﻴﺭﺍﺕ‬
‫ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﺍﻟﻤﻘﺩﺭﺓ ﻋﻠﻰ ﺍﻟﺘﺭﺘﻴﺏ ﺒﺎﻟﺸﻜل ﺍﻻﺘﻲ‪:‬‬

‫‪ŷ it = −1246.636 + 48.863 x1 + 0.8011 x 2 − 0.0857 x 3 + 37.45 x4‬‬

 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[278‬‬

‫‪ŷ it = −639.609 + 87334 x1 + 0.747 x 2 − 0.688 x 3 + 7.759 x4 + 629.268Db − 2269.77Dc‬‬

‫‪+ 2975.479Dd + 16760.03De − 2163.923Df + 19536.45Dg + 19248.92Dh‬‬
‫‪+ 7564.28Di − 1715.25D j − 2031.49Dk + 5443.78Dl − 3679.22Dm + 1331.36Dn‬‬
‫‪+ 36817.78Do + 15494.8Dp − 662.84Dq + 9883.53Dr‬‬

‫‪ŷ it = −1272.65 + 52.066 x1 + 0.779 x 2 − 0.088 x 3 + 39.93 x4‬‬

‫ﺍﻟﺠﺩﻭل)‪ :(2‬ﻤﻌﻠﻤﺎﺕ ﻨﻤﻭﺫﺝ ﺍﻟﺩﺭﺍﺴﺔ ﺍﻟﻤﻘﺩﺭﺓ ﺒﺎﺴﺘﺨﺩﺍﻡ ﺍﻟﻨﻤﺎﺫﺝ ﺍﻟﺜﻼﺜﺔ‬

‫ﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﺘﻭﻀﻴﺤﻴﺔ‬ ‫ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ‬ ‫ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ‬ ‫ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ‬

‫ﻭﺍﻟﻭﻫﻤﻴﺔ‬ ‫ﺍﻟﺘﺠﻤﻴﻌﻲ‬ ‫ﺍﻟﺜﺎﺒﺘﺔ‬ ‫ﺍﻟﻌﺸﻭﺍﺌﻴﺔ‬

‫‪constant‬‬ ‫‪-1246.636‬‬ ‫‪-639.609‬‬ ‫‪-1272.65‬‬

‫‪x1‬‬ ‫)*( ‪48.863‬‬ ‫)*( ‪87.334‬‬ ‫)*( ‪52.066‬‬

‫‪x2‬‬ ‫)*( ‪0.8011‬‬ ‫)*( ‪0.747‬‬ ‫)*( ‪0.779‬‬

‫‪x3‬‬ ‫)*( ‪-0.0857‬‬ ‫)*( ‪-0.688‬‬ ‫)*( ‪-0.0888‬‬

‫‪x4‬‬ ‫‪37.45‬‬ ‫‪7.759‬‬ ‫‪38.93‬‬

‫‪Db‬‬ ‫‪629.268‬‬

‫‪Dc‬‬ ‫‪-2269.77‬‬

‫‪Dd‬‬ ‫‪2975.479‬‬

‫‪De‬‬ ‫)*( ‪16760.03‬‬

‫‪Df‬‬ ‫‪-2163.923‬‬

‫‪Dg‬‬ ‫)*( ‪19536.45‬‬

 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[279‬‬

‫‪Dh‬‬ ‫)*( ‪19248.92‬‬

‫‪Di‬‬ ‫‪7564.28‬‬

‫‪Dj‬‬ ‫‪-1715.25‬‬

‫‪Dk‬‬ ‫‪-2031.49‬‬

‫‪Dl‬‬ ‫)*( ‪5443.78‬‬

‫‪Dm‬‬ ‫‪-3679.22‬‬

‫‪Dn‬‬ ‫‪1331.36‬‬

‫‪Do‬‬ ‫)*( ‪36817.78‬‬

‫‪Dp‬‬ ‫)*( ‪15494.8‬‬

‫‪Dq‬‬ ‫‪-662.84‬‬

‫‪Dr‬‬ ‫)*( ‪9883.53‬‬

‫‪R2‬‬ ‫‪0.7062‬‬ ‫‪0.688‬‬ ‫‪0.683‬‬

‫)*(ﻤﻌﻨﻭﻴﺔ ﺍﻟﻤﻌﻠﻤﺔ ﻋﻨﺩ ‪0.05‬‬

‫ﺒﻌﺩ ﺘﻘﺩﻴﺭ ﺍﻟﻨﻤﺎﺫﺝ ﺍﻟﺜﻼﺜﺔ ﻟﻠﻨﻤﻭﺫﺝ ﺍﻟﻤﺩﺭﻭﺱ ﺴﻭﻑ ﻨﻨﺘﻘل ﺍﻟﻰ ﺍﺴﺘﺨﺩﺍﻡ ﺍﺴﺎﻟﻴﺏ ﺍﻻﺨﺘﻴﺎﺭ ﺒﻴﻥ‬
‫ﻫﺫﻩ ﺍﻟﻨﻤﺎﺫﺝ ﺍﻟﺜﻼﺜﺔ ﻤﻥ ﺨﻼل ﺍﺨﺘﺒﺎﺭ ‪ F‬ﺍﻟﻤﻘﻴﺩ ﻭﺍﺨﺘﺒﺎﺭ ‪ Hausman‬ﻭﻜﻤﺎ ﻫﻭ ﻤﻭﻀﺢ ﻓـﻲ‬
‫ﺍﻟﺠﺩﻭل )‪.(3‬‬
 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[280‬‬

‫ﺍﻟﺠﺩﻭل)‪ :(3‬ﺍﺨﺘﺒﺎﺭ ‪ F‬ﺍﻟﻤﻘﻴﺩ ﻭﺍﺨﺘﺒﺎﺭ ‪Hausman‬‬

‫‪p-value‬‬ ‫ﻗﻴﻤﺔ ﺍﻻﺨﺘﺒﺎﺭ‬ ‫ﻨﻭﻉ ﺍﻻﺨﺘﺒﺎﺭ‬

‫)*( ‪0.0002‬‬ ‫‪2.779‬‬ ‫ﺍﺨﺘﺒﺎﺭ ‪ F‬ﺍﻟﻤﻘﻴﺩ‬

‫)*( ‪0.000‬‬ ‫‪29.63‬‬ ‫ﺍﺨﺘﺒﺎﺭ ‪Hausman‬‬

‫)*(ﻤﻌﻨﻭﻴﺔ ﺍﻻﺨﺘﺒﺎﺭ ﻋﻨﺩ ‪0.05‬‬

‫ﺒﺎﻻﻋﺘﻤﺎﺩ ﻋﻠﻰ ﻨﺘﺎﺌﺞ ﺍﻟﺠﺩﻭل )‪ (3‬ﻨﻼﺤﻅ ﺒﺎﻥ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ﻟﺒﻴﺎﻨﺎﺘﻨﺎ ﺍﻟﻤﺩﺭﻭﺴﺔ ﻫﻭ ﻨﻤﻭﺫﺝ‬
‫ﺍﻟﺘﺄﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺤﻴﺙ ﻻﺤﻅﻨﺎ ﻤﻥ ﺍﺨﺘﺒﺎﺭ ‪ F‬ﺍﻟﻤﻘﻴﺩ ﺒﺎﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺘﻐﻠﺏ ﻤـﻥ‬
‫ﻨﺎﺤﻴﺔ ﺍﻻﻫﻤﻴﺔ ﻓﻲ ﺍﻻﺴﺘﺨﺩﺍﻡ ﻤﻘﺎﺭﻨﺔ ﺒﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ‪ ،‬ﻭﻋﻨـﺩ ﺍﺠـﺭﺍﺀ ﺍﺨﺘﺒـﺎﺭ‬
‫‪ Hausman‬ﺍﻴﻀﺎ ﺘﻐﻠﺏ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻋﻠﻰ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻤﻥ ﻨﺎﺤﻴﺔ‬
‫ﺍﻫﻤﻴﺔ ﻫﺫﺍ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻤﻼﺀﻤﺔ ﻟﻠﺒﻴﺎﻨﺎﺕ ﺍﻟﻤﺩﺭﻭﺴﺔ‪ .‬ﺘﺠﺩﺭ ﺍﻻﺸﺎﺭﺓ ﻫﻨﺎ ﺍﻟﻰ ﺍﻥ ﻗﻴﻤﺔ ﺍﺨﺘﺒـﺎﺭ‬
‫‪ Hausman‬ﺘﺤﺴﺏ ﻟﻤﻌﻠﻤﺎﺕ ﺍﻟﻤﺘﻐﻴـﺭﺍﺕ ﺍﻟﻤﺩﺭﻭﺴـﺔ ﻭﻫـﻲ ) ‪ ، x 3 ، x 2 ، x1‬ﻭ ‪ ( x 4‬ﺍﻱ‬
‫) ‪ ، β 3 ، β 2 ، β 1‬ﻭ ‪ .( β 4‬ﻓﻲ ﻀﻭﺀ ﻨﺘﺎﺌﺞ ﺘﻘﺩﻴﺭ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻨﻼﺤﻅ ﺒﺎﻥ ﺴـﻌﺭ‬
‫ﺍﻟﻨﻔﻁ ﻭﻋﺠﺯ ﻤﻴﺯﺍﻥ ﺍﻟﻤﺩﻓﻭﻋﺎﺕ ﻭﺤﺠﻡ ﺍﻟﺴﻜﺎﻥ ﺘﻤﺜل ﻤﺘﻐﻴﺭﺍﺕ ﺭﺌﻴﺴﻴﺔ ﻓﻲ ﺍﻟﻤﻭﺍﺯﻨﺔ ﺍﻟﻌﺎﻤـﺔ‬
‫ﻟﻠﺩﻭﻟﺔ ﻓﻲ ﺤﻴﻥ ﺍﻅﻬﺭ ﺍﻟﺘﻀﺨﻡ ﻋﺩﻡ ﻤﻌﻨﻭﻴﺘﻪ ﻓﻲ ﺍﻟﻤﻭﺍﺯﻨﺔ ﺍﻟﻌﺎﻤﺔ‪ .‬ﻟﻐﺭﺽ ﺍﻟﻭﺼـﻭل ﺍﻟـﻰ‬
‫ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﻴﺤﺘﻭﻱ ﻋﻠﻰ ﺍﻫﻡ ﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﻤﺩﺭﻭﺴﺔ ﻟﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﺴﻭﻑ ﻨﻘـﻭﻡ‬
‫ﺒﺘﻭﻓﻴﻕ ﻋﺩﺩ ﻤﻥ ﺍﻟﻨﻤﺎﺫﺝ ﻤﻘﺩﺍﺭﻫﺎ ‪ 15‬ﻨﻤﻭﺫﺠﺎ ﻭﻓﻲ ﻜل ﻤﺭﺓ ﺴﻭﻑ ﻨﻘـﻭﻡ ﺒﺘﻭﻓﻴـﻕ ﻨﻤـﺎﺫﺝ‬
‫ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﺜﻼﺜﺔ ﻟﻠﺘﺎﻜﺩ ﻤﻥ ﺍﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻫﻭ ﺍﻟﻨﻤﻭﺫﺝ ﺍﻟﻤﻼﺌﻡ ﺜﻡ ﺒﻌـﺩ‬
‫ﺫﻟﻙ ﻨﻘﻭﻡ ﺒﺤﺴﺎﺏ ﻤﻌﻴﺎﺭ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﻤـﺼﺤﺢ ﻭﻤﻌﻴـﺎﺭ ﺍﻜـﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤـﺎﺕ ﻟﻜـل‬
‫ﻨﻤﻭﺫﺝ‪.‬ﺍﻟﺠﺩﻭل )‪ (4‬ﻴﻭﻀﺢ ﻗﻴﻡ ﻤﻌﻴﺎﺭﻱ ‪ R 2 adj‬ﻭ ‪. AIC‬‬
 ‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‪...........................‬‬ ‫]‪[281‬‬

‫ﺍﻟﺠﺩﻭل )‪ :(4‬ﻗﻴﻡ ﻤﻌﻴﺎﺭﻱ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﻤﺼﺤﺢ ﻭﺍﻜﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤﺎﺕ‬

‫ﺍﻟﻨﻤﻭﺫﺝ‬ ‫‪R 2 adj‬‬ ‫‪AIC‬‬

‫‪x1‬‬ ‫‪0.244‬‬ ‫‪21.52‬‬

‫‪x2‬‬ ‫‪0.717‬‬ ‫‪20.538‬‬

‫‪x3‬‬ ‫‪0.1648‬‬ ‫‪21.62‬‬

‫‪x4‬‬ ‫‪0.1562‬‬ ‫‪21.631‬‬

‫‪x1 x 2‬‬ ‫‪0.7256‬‬ ‫‪20.511‬‬

‫‪x1 x 3‬‬ ‫‪0.2472‬‬ ‫‪21.52‬‬

‫‪x1 x 4‬‬ ‫‪0.2447‬‬ ‫‪21.523‬‬

‫‪x2x3‬‬ ‫‪0.7164‬‬ ‫‪20.544‬‬

‫‪x2x4‬‬ ‫‪0.717‬‬ ‫‪20.539‬‬

‫‪x3x4‬‬ ‫‪0.1725‬‬ ‫‪21.615‬‬

‫‪x1 x 2 x 3‬‬ ‫‪0.733‬‬ ‫‪20.485‬‬

‫‪x1 x 2 x 4‬‬ ‫‪0.725‬‬ ‫‪20.513‬‬

‫‪x1 x 3 x 4‬‬ ‫‪0.2452‬‬ ‫‪21.526‬‬

‫‪x2x3x4‬‬ ‫‪0.716‬‬ ‫‪20.546‬‬

‫‪x1 x 2 x 3 x 4‬‬ ‫‪0.7323‬‬ ‫‪20.786‬‬

 ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ )‪2012 ( 21‬‬ ‫]‪[282‬‬

‫ﻓﻲ ﻀﻭﺀ ﻨﺘﺎﺌﺞ ﺍﻟﺠﺩﻭل )‪ (4‬ﻴﺘﻀﺢ ﺒﺎﻥ ﻤﻌﻴﺎﺭ ‪ R 2 adj‬ﻭﻤﻌﻴﺎﺭ ‪ AIC‬ﻗﺩ ﺍﺨﺘﺎﺭﺍ ﺍﻟﻨﻤـﻭﺫﺝ‬
‫) ‪ ، x 2 ، x1‬ﻭ ‪ ( x 3‬ﺤﻴﺙ ﺍﻋﻁﻰ ﺍﻋﻠﻰ ﻗﻴﻤﺔ ﻟﻤﻌﻴﺎﺭ ‪ R 2 adj‬ﻭﺍﻗـل ﻗﻴﻤـﺔ ﻟﻤﻌﻴـﺎﺭ ‪AIC‬‬
‫ﻭﻴﻌﺒﺭ ﻋﻥ ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺘﺎﺜﻴﺭﺍﺕ ﺜﺎﺒﺘﺔ ﺠﺯﺌﻲ ﺒﺎﻟﺸﻜل ﺍﻻﺘﻲ‪:‬‬

‫‪ŷ it = −569.822 + 88.252 x1 + 0.747 x 2 − 0.703 x 3 + 615.06Db − 2350.9Dc + 3043.6Dd‬‬

‫‪+ 17172De − 2232.6Df + 19768.94Dg + 19826.85Dh + 7749.03Di − 1769D j‬‬
‫‪− 2085.27 Dk + 5400.6Dl − 3697.4Dm + 1328.55Dn + 37759.9Do + 15845Dp‬‬
‫‪− 678.5Dq + 10188.9Dr‬‬
‫‪ -6‬ﺍﻻﺴﺘﻨﺘﺎﺠﺎﺕ‬

‫‪ -1‬ﺍﻥ ﺍﺴﺘﺨﺩﺍﻡ ﻤﻔﻬﻭﻡ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺃﻋﻁﻰ ﺩﺭﺍﺴﺔ ﺸﺎﻤﻠﺔ ﻋﻨﺩ ﺍﺴﺘﺨﺩﺍﻡ ﺍﻟﺒﻴﺎﻨـﺎﺕ‬
‫ﺍﻟﻤﻘﻁﻌﻴﺔ ﻭﺍﻟﺴﻼﺴل ﺍﻟﺯﻤﻨﻴﺔ ﻓﻲ ﺍﻥ ﻭﺍﺤﺩ ﺤﻴﺙ ﺍﻋﻁﻰ ﻨﻤﻭﺫﺝ ﺭﻴﺎﻀﻲ ﻭﺍﺤﺩ ﻴﺤﻜﻡ ﻁﺒﻴﻌـﺔ‬
‫ﺍﻟﺩﺭﺍﺴﺔ ﻤﻘﺎﺭﻨﺔ ﻓﻴﻤﺎ ﻟﻭ ﺘﻡ ﺍﺴﺘﺨﺩﺍﻡ ‪ 15‬ﻨﻤﻭﺫﺝ ﺍﻨﺤﺩﺍﺭ ﺍﻋﺘﻴﺎﺩﻱ ) ﻨﻤﻭﺫﺝ ﻭﺍﺤﺩ ﻟﻜل ﺴﻠـﺴﻠﺔ‬
‫ﺯﻤﻨﻴﺔ( ﺍﻭ ﻤﻘﺎﺭﻨﺔ ﻓﻴﻤﺎ ﻟﻭ ﺘﻡ ﺍﺴﺘﺨﺩﺍﻡ ‪ 18‬ﻨﻤﻭﺫﺝ ﺍﻨﺤﺩﺍﺭ ﺴﻠﺴﻠﺔ ﺯﻤﻨﻴﺔ ) ﻨﻤـﻭﺫﺝ ﺍﻨﺤـﺩﺍﺭ‬
‫ﺴﻠﺴﻠﺔ ﺯﻤﻨﻴﺔ ﻟﻜل ﺩﻭﻟﺔ(‪.‬‬

‫‪ -2‬ﺯﻴﺎﺩﺓ ﺩﺭﺠﺎﺕ ﺍﻟﺤﺭﻴﺔ ﻋﻨﺩ ﺍﺴﺘﺨﺩﺍﻡ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ ﺍﻟﻁﻭﻟﻴﺔ ﺤﻴﺙ ﺒﻠﻐﺕ ﻋﺩﺩ ﺍﻟﻤﺸﺎﻫﺩﺍﺕ‬
‫‪ 270‬ﻤﺸﺎﻫﺩﺓ ﻓﻲ ﺤﻴﻥ ﺒﻠﻐﺕ ﺩﺭﺠﺔ ﺤﺭﻴﺔ ﺍﻟﺨﻁﺄ ‪ 265‬ﻋﻨﺩ ﺍﺴـﺘﺨﺩﺍﻡ ﻨﻤـﻭﺫﺝ ﺍﻻﻨﺤـﺩﺍﺭ‬
‫ﺍﻟﺘﺠﻤﻴﻌﻲ ﻭﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻓﻲ ﺤﻴﻥ ﻜﺎﻨﺕ ﺩﺭﺠـﺔ ﺤﺭﻴـﺔ ﺍﻟﺨﻁـﺎ ‪ 249‬ﻋﻨـﺩ‬
‫ﺍﺴﺘﺨﺩﺍﻡ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭﻫﺫﺍ ﻤﺎ ﻴﻌﻁﻲ ﺩﻗﺔ ﻓﻲ ﻨﺘﺎﺌﺞ ﺍﻟﺘﻘﺩﻴﺭ ﻟﻠﻤﻌﻠﻤﺎﺕ ﻭﺍﻥ ﺩﺭﺠﺔ‬
‫ﺍﻟﺤﺭﻴﺔ ﻫﺫﻩ ﻜﺎﻨﺕ ﺍﻜﺒﺭ ﻤﻥ ﺩﺭﺠﺎﺕ ﺍﻟﺤﺭﻴﺔ ﺍﻟﺒﺎﻟﻐﺔ ‪ 13‬ﻋﻨﺩ ﺍﺴﺘﺨﺩﺍﻡ ﻨﻤـﻭﺫﺝ ﻭﺍﺤـﺩ ﻟﻜـل‬
‫ﺴﻠﺴﻠﺔ ﺯﻤﻨﻴﺔ ﻭ ‪ 10‬ﻋﻨﺩ ﺍﺴﺘﺨﺩﺍﻡ ﻨﻤﻭﺫﺝ ﺍﻨﺤﺩﺍﺭ ﺴﻠﺴﻠﺔ ﺯﻤﻨﻴﺔ‪.‬‬

‫‪ -3‬ﻤﻥ ﻨﺘﺎﺌﺞ ﺍﻟﺠﺩﻭل )‪ (2‬ﻨﻼﺤﻅ ﺒﺎﻥ ﻗﻴﻤﺔ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺘﻜﻭﻥ ﺍﻜﺒﺭ ﻓﻲ ﺤﺎﻟـﺔ ﺍﺴـﺘﺨﺩﺍﻡ‬
‫ﻨﻤﻭﺫﺝ ﺍﻻﻨﺤﺩﺍﺭ ﺍﻟﺘﺠﻤﻴﻌﻲ ﻤﻘﺎﺭﻨﺔ ﺒﺎﺴﺘﺨﺩﺍﻡ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻭﻜﺫﻟﻙ ﺒﺎﻥ‬
‫ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﺨﺎﺹ ﺒﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﻌﺸﻭﺍﺌﻴﺔ ﻴﻜﻭﻥ ﺍﻗل ﻤـﻥ ﻨﻅﻴـﺭﻩ ﻓـﻲ ﻨﻤـﻭﺫﺝ‬
 ...........................‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‬ [283]

‫ ﻟﺫﺍ ﻻ ﻴﻤﻜﻥ ﺍﻻﻋﺘﻤﺎﺩ ﻋﻠﻰ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﻓﻲ ﺍﺨﺘﻴـﺎﺭ ﻨﻤـﻭﺫﺝ ﺍﻟﺒﻴﺎﻨـﺎﺕ‬.‫ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ‬
‫ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﻤﻼﺌﻡ ﻷﻴﺔ ﺩﺭﺍﺴﺔ ﺒﺴﺒﺏ ﺍﻥ ﻤﻌﺎﻤل ﺍﻟﺘﺤﺩﻴﺩ ﻴﻌﺘﻤﺩ ﻋﻠﻰ ﻤﻘﺎﻴﻴﺱ ﻤﺨﺘﻠﻔﺔ ﻓﻲ ﺤﺴﺎﺒﻪ‬
.‫ﻤﻥ ﻨﻤﻭﺫﺝ ﻵﺨﺭ‬

‫ ﻻﺤﻅﻨﺎ ﺒﺎﻥ ﻨﻤﻭﺫﺝ ﺍﻟﺒﻴﺎﻨـﺎﺕ‬Hausman ‫ ﺍﻟﻤﻘﻴﺩ ﻭﺍﺨﺘﺒﺎﺭ‬F ‫ ﻤﻥ ﺨﻼل ﺍﺠﺭﺍﺀ ﺍﺨﺘﺒﺎﺭ‬-4

‫ﺍﻟﻁﻭﻟﻴﺔ ﺍﻟﻤﻼﺌﻡ ﻟﺒﻴﺎﻨﺎﺘﻨﺎ ﻫﻭ ﻨﻤﻭﺫﺝ ﺍﻟﺘﺎﺜﻴﺭﺍﺕ ﺍﻟﺜﺎﺒﺘﺔ ﻭﻤﻥ ﺨﻼل ﺍﺴـﺘﺨﺩﺍﻡ ﻤﻌﻴـﺎﺭ ﻤﻌﺎﻤـل‬
‫ﺍﻟﺘﺤﺩﻴﺩ ﺍﻟﻤﺼﺤﺢ ﻭﻤﻌﻴﺎﺭ ﺍﻜﺎﻜﻲ ﻟﻠﻤﻌﻠﻭﻤﺎﺕ ﺘﺒﻴﻥ ﺒﺎﻥ ﺍﻓﻀل ﻨﻤﻭﺫﺝ ﺠﺯﺌﻲ ﻫـﻭ ﺍﻟﻨﻤـﻭﺫﺝ‬
‫ ﻋﺠـﺯ ﻤﻴـﺯﺍﻥ‬،‫ﺍﻟﺫﻱ ﻴﺤﻭﻱ ﻋﻠﻰ ﺍﻟﻤﺘﻐﻴﺭﺍﺕ ﺍﻟﺘﻔﺴﻴﺭﻴﺔ ﻭﻫﻲ ﺍﻟﻤﻌﺩل ﺍﻟﺴﻨﻭﻱ ﻟﺴﻌﺭ ﺍﻟﻨﻔﻁ‬
.‫ ﻭ ﺤﺠﻡ ﺍﻟﺴﻜﺎﻥ‬،‫ﺍﻟﻤﺩﻓﻭﻋﺎﺕ‬

‫ ﺍﻟﻤﺼﺎﺩﺭ‬-7

‫ ﺸﺭﻜﺔ‬،"‫ " ﺍﻟﺘﻘﺭﻴﺭ ﺍﻟﻌﺭﺒﻲ ﺍﻻﻗﺘﺼﺎﺩﻱ ﺍﻟﻤﻭﺤﺩ‬،2010 ،‫ ﻭﺍﺨﺭﻭﻥ‬،‫ ﺼﻨﺩﻭﻕ ﺍﻟﻨﻘﺩ ﺍﻟﻌﺭﺒﻲ‬.1
.‫ﺍﺒﻭ ﻅﺒﻲ ﻟﻠﻁﺒﺎﻋﺔ ﻭﺍﻟﻨﺸﺭ‬

2. Baltagi, B., H., Jung, B., Ch. and Song, S., H., 2010, "Testing
for heteroskedasticity and serial correlation in a random effects
Panel Data Model", Journal of Econometrics, Vol.154, Iss.2,
pp. 122-124.
3. Blatagi, B., H., 2005, "Econometric Analysis of Panel Data", 3rd
ed., John Wiley & Sons, Ltd, West Sussex.

4. Bramati, M. and Croux, Ch., 2007, “Robust Estimators for the

Fixed Effects Panel Data Model", Econometrics Journal,
Vol.10, Iss.3, pp. 1-19.

5. Chuang, Ch. and Wang, Y., 2009, " Developed stock market
reaction to political change: a panel data analysis", Quality and
Quantity, Vol. 43, pp. 941-949.
6. Dustmann, Ch. And Engracia, M., 2007, " Selection correction
in Panel Data Models: An Application to the Estimation of
Female's Wage Equations", Econometrics Journal, Vol.10,
Iss.2, pp. 263-293.
 2012 ( 21) ‫ﺍﻟﻤﺠﻠﺔ ﺍﻟﻌﺭﺍﻗﻴﺔ ﻟﻠﻌﻭﻡ ﺍﻟﻌﺭﺍﻗﻴﺔ‬ [284]

7. El-Gamal, M., A. and Inanoglu, H., 2005, " Inefficiency and

Heterogeneity in Turkish Banking: 1990-2000", Journal of
Applied Econometrics, Vol. 20, No. 5 , pp. 641-664.
8. Greene, W., H., 2012, “Econometrics Analysis", 7th ed.,
Pearson Education, Inc., NJ.

9. Hausman, J., 1978, "Specification Test in Econometrics",

Econometrica, Vol.46, pp.1251-1271.

10. Hughes, A., W. and King, M., L., 2003, " Model Selecting
using AIC in the Presence of One-Sided Information", Journal
of Statistical Planning and Inference, Vol.115, pp. 397-411.

11. Kai,T. and Qin, L., 2011, " Analysis on consumption structure
of Shanghai resident by using Panel data model", Proceeding of
IEEE 2nd International Conference on Software Engineering and
Service Science, pp. 819-822.
12. Lee, D., G. and Russell, J., S., 2004, " Panel Data Analysis of
Factors Affecting As-Built Roughness of Asphaltic Concrete
Pavements", Journal of Transportation Engineering, Vol.4, pp.
479-485.

13. Lee, L. and Yu, J., 2010, "Estimation of spatial autoregressive

Panel Data Models with fixed effects", Journal of
Econometrics, Vol.154, pp.165-185.
14. Lukas, M. and Jan, N., 2011, "Application of Econometric
Panel Data Model for Regional Competitiveness Evaluation of
Selected EU 15 Countries", Journal of Competitiveness, Issue
4, pp. 23-38.
15. Mikhed, V. and Zemcik, P., 2009, "Testing for Bubbles in
Housing Markets: A Panel Data Approach", Journal of Real
Estate Financial Economics, No.38, pp. 366-386.

16. Shao, X., Chen, Y., and Wang, H., 2009, " Research on
Financial Warning for Chinese Listed Companies by Using
Panel Data Model", Proceeding of IEEE International
Conference on System, pp. 4663-4667.
 ...........................‫ﺍﺨﺘﻴﺎﺭ ﺍﻟﻨﻤﻭﺫﺝ ﻓﻲ ﻨﻤﺎﺫﺝ ﺍﻟﺒﻴﺎﻨﺎﺕ‬ [285]

17. Sun, Y., 2009, "Estimation of semiparametric regression model

with longitudinal data ", Lifetime Data Analysis, Vol.16, pp.
271-298.