16Nov/100
Non-parametric analysis of the relationship between inflation and unemployment
Summary : The equation of Phillips reviewed successively by Samuelson-Solow and Akerlof Friedman-Lucas was the main macroeconomic debate of the past 50 years. The nonparametric econometrics developed after the 70s, the tools used here were developed after most of the publications of the authors mentioned above. The aim is to show wich of these theorists is right by a study of the joint density of unemployment-inflation relationship, leading then to a non-parametric estimation by local polynomials and splines. Contrary to all of these economists, we will not make any assumptions about the functional form of the Phillips curve.
Download the note (french only)
Attention, les données ne sont pas fournies.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 | #Importation des données xls chomage= year ) { return(tserie[( year - start(tserie)[1] +1),]) }else{ print("Pas d'année comprise dans la série") } }else{ if( year == start(tserie)[1] ){ return(tserie[as.vector((year*frequency(tserie))-(start(tserie)[1]*frequency(tserie))+start(tserie)[2]):((year*frequency(tserie))-(start(tserie)[1]*frequency(tserie))+frequency(tserie)),]) }else if( year == end(tserie)[1] ){ return(tserie[as.vector((year*frequency(tserie))-(start(tserie)[1]*frequency(tserie))+1):((year*frequency(tserie))-(start(tserie)[1]*frequency(tserie))+end(tserie)[2]),]) }else{ return(tserie[as.vector((year*frequency(tserie))-(start(tserie)[1]*frequency(tserie))+1):((year*frequency(tserie))-(start(tserie)[1]*frequency(tserie))+frequency(tserie)),]) } } } #et une fonction qui nous donne les données sur plusieurs années tperiode = function( tserie, mini, maxi ){ matrice = as.vector(na.omit(c(tselect(tserie,mini)))) for (i in as.numeric(mini+1):maxi) { matrice = c(matrice,as.vector(na.omit(c(tselect(tserie,i))))) } return(matrice) } na.omit(tselect(tinflation, 2009)) tselect(tinflation,2010) ## On écrit affiche l'équation year = 1960 myU = na.omit(tselect(tU[,-28],year)) myW = na.omit(tselect(tW[,-28],year)) if( length(myU) == length(myW) ) print("Bonne bdd") plot(myW~myU) reg = lm(myW~myU) lines(reg) plot(lm(myW~myU)) #sélectionner la balise borne1 = 1955 borne2 = 1967 borne1 = 1973 borne2 = 1979 borne1 = 2008 borne2 = 2010 borne1 = 1955 borne2 = 2010 myW = tperiode(tinflation, borne1, borne2) myU = tperiode(tchomage, borne1, borne2) plot(myW~myU) # tentative d'un echantillon pur sans inflation importée myW = c(tperiode(tinflation, 1955, 1972),tperiode(tinflation, 1985, 2010)) myU = c(tperiode(tchomage, 1955, 1972),tperiode(tchomage, 1985, 2010)) plot(myW~myU, xlim=c(0,20), ylim=c(-1,20)) ## II. Densité non paramétrique #on détermine le paramètre de bande par validation croisée avec la fonction Hscv library(ks) #param = cbind(dnorm(myU, mean = mean(myU), sd = sd(myU)),dnorm(myW, mean = mean(myW), sd = sd(myW))) param = cbind(myU,myW) H.scv |