Previous studies have shown a significant and positive relationship between agricultural output and long run economic development. Nigeria, with current extreme poverty rate of 48% is also characterized with high level of food insecurity. The country however, has about 99 million ha of arable land which supports numerous farming activities. Therefore, there exist an opportunity to increase agricultural output and achieve economic development. Further studies also showed that the percentage of agricultural loans to total loans in Nigeria have been declining over the last 15 years which suggests some form of neglect of the sector. This paper therefore examined the impact of agricultural loans and other relevant variables on agricultural GDP in Nigeria with a view to estimate a regression model that can explain variability in agricultural GDP. All the data used were secondary data collected from the annual statistical bulletin of the Central Bank of Nigeria (2018) and USDA, Economic Research Service (Nov. 2019). Ordinary Least Square regression model was used to estimate regression parameters and to show relationship between the variables. Analyzed results showed that agricultural GDP in Nigeria is statistically dependent on agricultural loan, government expenditure and available farm land. The result was statistically tested to be significant at >=95% level of confidence. Further results also showed that agricultural loan in Nigeria is statistically dependent on agricultural credit guarantee and Non-performing loan ratio. This was also tested to be significant at >=95% level of confidence.The study recommended that the results calls for more allocation of credit for agricultural purposes, more government spending to agriculture and more funding of the agricultural credit guarantee program by the government.

Universidade Federal do Cearà / A tese intitulada "Ensaios Econometricos sobre a dinÃmica do Pib AgrÃcola" à composta de trÃs artigos. O primeiro artifo à intitulado "Setor AgrÃcola Brasileiro: Uma anÃlise de ConvergÃncia Microregional" analise o processo de convergÃncia microregional agrÃcola utilizando o Modelo Threshold no perÃodo de 1970 a 1996. Os resultados mostraram a existÃncia de cinco clubes de convergÃncia; um com um grupo de microregiÃes mais ricas, um com um grupo de microregiÃes mais pobres e trÃs grupos intermediÃrios. Os resultados tambÃm mostraram que o capital fÃsico à mais importante que o capital humano na explicaÃÃo do processo de crescimento. O segundo artigo intitulado: "O Setor AgrÃcola Brasileiro: Uma aplicaÃÃo do Modelo de TendÃncias e Ciclos Comuns no perÃodo de 1990 a 2005", analisa o comportamento das variaveis produto agrÃcola, taxa de cÃmbio real e saldo da balanÃa comercial agrÃcola brasileira. Utilizou-se a tÃcnica de cointegraÃÃo para identificar um sistema de vetores auto-regressivos com tendÃncias estocÃsticas comuns e para investigar as respostas do sistema a choques transitÃrios e permanentes. Os testes comprovaram a existÃncia de uma tendÃncia estocÃstica comum e dois ciclos comuns entre as variÃveis. As decomposiÃÃes das variÃncias indicam que os choques transitÃrios explicam a maior parte das flutuaÃÃes de curto e longo prazo no produto agrÃcola. Constatou-se tambÃm que os choques permanentes sÃo mais importantes para explicar as variÃncias da taxa de cÃmbio e da balanÃa comercial agrÃcola, principalmente no longo prazo. O terceiro artigo intitulado: "PrevisÃo da Taxa de Crescimento do Produto AgrÃcola Brasileiro: Uma aplicaÃÃo de Modelos de Ãndice de DifusÃo Linear e NÃo Linear" aplica os modelos linear e nÃo linear de Ãndice de difusÃo com efeito threshold para prever, um perÃodo à frente, a taxa de crescimento trimestral do Pib agrÃcola brasileiro. Estes modelos sÃo compostos por fatores que sÃo observÃveis e representam uma caracterÃstica comum das variÃveis explicativas, permitindo uma reduÃÃo significativa do nÃmero destas variÃveis. Em seguida à feita a comparaÃÃo das previsÃes destes modelos entre si e em relaÃÃo ao modelo AR que à tomado como benchmark. Verifica-se que o modelo de Ãndice de difusÃo linear apresentou uma pequena superioridade, em termos de eficiÃncia preditiva, em relaÃÃo aos modelos nÃo linear e AR, que apresentaram resultados semelhantes. / The thesis entitled "Econometrical assays on the dynamics of the agricultural Gdp" is composed of three papers. The first article is entitled "Brazilian Agricultural Sector: An Analysis of Microregional Convergence" analyzes the process of agricultural micro regional convergence using the Threshold Model in the period 1970 to 1996. The results show the existence of five clubs of convergence: one with a group of richer micro regions; one with a group of poor micro regions and three intermediates groups. The results also show that physical capital is more important than human capital in the explanation of the growth process. The second paper with title "The Brazilian Agricultural Sector: A Application of the Model of Common Cycles in the period os 1990 the 2005" analyzes the behavior of Agricultural Gdp, real exchange rate and the current account balance agricultural in Brazil during the period of 1990 to 2005. Cointegration techniques were used to identify a VAR system with common stochastic trends, and to investigate the system responses to transitories and permanent shocks. The tests had proven the existence of a common trend and two cycles between the variables. Variance decompositions indicates that transiotories shocks accont form most of the short and long run fluctuations in agricultural Gdp. It was found that permanent shocks explain most of the variance of the exchance rate and of the current account balance mainly in the long run. The third paper entitled "Forecast of the Rate of Growth of the Brazilian Agricultural Gdp: An Application of Models of Diffusion Index Linear and Non-Linear". This article applies the linear and non-linear diffusion index model with a threshold effect to forecast, one step ahead, the qyarterly growt rate of Brazilian Agricultural Gdp. These models are composed by common factor which allow a significant reduction in the number of the original explaining variables. After comparing forecast of these two models between themselves and to an AR model, used as benchmark, one comes to the conclusion that the linear model presents a small superiority, in terms of predictive efficiency, in relation to the nonlinear and AR models.