Submitted by Sara Ribeiro (sara.ribeiro@ucb.br) on 2017-10-30T13:13:38Z No. of bitstreams: 1 MichelAngeloConstantinodeOliveiraTese2014.pdf: 517065 bytes, checksum: a528b5761c52e0544f3e4b39cddabf3f (MD5) / Approved for entry into archive by Sara Ribeiro (sara.ribeiro@ucb.br) on 2017-10-30T13:15:02Z (GMT) No. of bitstreams: 1 MichelAngeloConstantinodeOliveiraTese2014.pdf: 517065 bytes, checksum: a528b5761c52e0544f3e4b39cddabf3f (MD5) / Made available in DSpace on 2017-10-30T13:15:02Z (GMT). No. of bitstreams: 1 MichelAngeloConstantinodeOliveiraTese2014.pdf: 517065 bytes, checksum: a528b5761c52e0544f3e4b39cddabf3f (MD5) Previous issue date: 2014-11-28 / UCB / Economic theory uses the production function to develop microeconomic and macroeconomic analyses, respectively at the level of rm and wealth of nations. This thesis proposes a new model for application in the production function, introducing instrumental copulas for modelling the dependence structure of random variables for analysis of asymmetric behavior and extreme events. In the rst test the function of traditional production is modeled using vine-copula for rms from the USA, Germany and England, so it was possible to construct multivariate distributions and estimation of capital-output pairs (y ???? k) variables; work-product (y ???? l); working-capital as a production (k ???? ljy); and know the structures associated dependencies, evaluating the nonlinear dependence and heavy tails. For the second test, following the data and methodological procedures of the rst test, using vine-copula and the construction parameters and coe cient of dependence are the same. The di erence is in the modelling of stochastic frontier and approach insu ciency using this new type of instrument. The marginal distributions were estimated using the gamma function and normal function, several families of copulas were found, including Gumbel, Student-t, BB7, Normal and Clayton which implies changes in behavior and dependence of each pair of variables over the years and in each country, in addition to showing the heavy tails where rms of large and / or small focus. This change in families con rm the stochastic behavior of the data and the production function, in contrast to the inappropriate use of deterministic functions and multivariate normality assumption. / A teoria econ??mica utiliza a fun????o de produ????o para desenvolver an??lises microecon??micas e macroecon??micas, respectivamente ao n??vel da rma e riqueza das na????es. Este trabalho de tese prop??e uma nova modelagem para aplica????o em fun????o de produ????o, introduzindo o instrumental de c??pulas para modelagem da estrutura de depend??ncia em vari??veis aleat??rias, para an??lise do comportamento assim??trico e dos eventos extremos. No primeiro ensaio a fun????o de produ????o tradicional ?? modelada utilizando vine-c??pulas para as rmas dos EUA, Alemanha e Inglaterra, assim foi poss??vel a constru????o de distribui????es multivariadas e estima????o dos pares de vari??veis produto-capital (y????k); produto-trabalho (y????l); capital-trabalho dado uma produ????o (k????ljy); e conhecer as estruturas de depend??ncias associadas, avaliando a depend??ncia n??o linear e as caudas pesadas. Para o segundo ensaio, segue os dados e procedimentos metodol??gicos do primeiro ensaio, utilizando vine-c??pulas e a constru????o dos par??metros e coe cientes de depend??ncia s??o os mesmos. A diferen??a est?? na modelagem de fronteira estoc??stica e na abordagem de e ci??ncia utilizando esse novo instrumento estat??stico. As distribui????es marginais foram estimadas utilizando a fun????o gamma e a fun????o normal, foram encontradas v??rias fam??lias de c??pulas, entre elas Gumbel, Student-t, BB7, Normal e Clayton que implica em mudan??as no comportamento e na depend??ncia de cada par de vari??veis ao longo dos anos e em cada pa??s, al??m de mostrar as caudas pesadas onde as rmas de grande e/ou pequeno porte se concentram. Essa mudan??a nas fam??lias con rmam o comportamento estoc??stico nos dados e da fun????o de produ????o, em contraste com o uso inadequado de fun????es deterministicas e hip??tese de normalidade multivariada.