Output from computer model e3mg detailing the depth of the recession and its length as a function of four exogenous parameters
e3mg is a matrix with 843 rows and 6 columns.
Four of the columns are exogenous variables (
and two are model outputs:
rec_len, the length (in years) of
the recession, and
dep_rec, the depth of the recession.
e3mg_LoF is a list of functions suitable for use with
The data comprises 843 runs of the e3mg econometric model, used to predict the recession precipitated by the banking crisis.
The depth of the recession is defined as the maximum difference between predicted post-crash GDP and GDP immediately pre-crash.
The length of the recession is defined as the time in years required for GDP to return to pre-crash levels.
Data kindly provided by Cambridge Econometrics
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
data(e3mg) a <- lm(rec_len~oil.price*direct.tax + direct.tax*saving.ratio + investment,data=data.frame(e3mg)) b <- lm(rec_dep~oil.price*direct.tax + direct.tax*saving.ratio + investment,data=data.frame(e3mg)) plot(residuals(a),residuals(b)) # correlated! # define an experiment object and find optimal prarams e3mg_expt <- apart(e3mg[1:20,],6:7) opt <- optimal_params(e3mg_expt, e3mg_LoF, option='c') # now a point in parameter space: center <- get_mdm(e3mg_expt)[c(1,40),] rownames(center) <- c('center_dep','center_len') xold(center) <- 0 #now predict the behaviour at the center: multem(center, e3mg_expt, hp=opt, e3mg_LoF, give = TRUE)