Changing expectations via function overloading

In this tutorial we will illustrate how to experiment with different expectations of the agents in the model.

import BeforeIT as Bit
using Random, Plots

Import standard parameters and initial conditions

par = Bit.AUSTRIA2010Q1.parameters
init = Bit.AUSTRIA2010Q1.initial_conditions

Dict{String, Any} with 21 entries:
  "sb_inact" => 2.23847
  "C_G"      => [11482.3; 11556.2; … ; 14578.2; 14714.8;;]
  "E_CB"     => 1.0618e5
  "sb_other" => 0.590286
  "D_RoW"    => 0.0
  "L_I"      => 236919.0
  "omega"    => 0.85
  "r_bar"    => 0.00164593
  "N_s"      => [123.0; 18.0; … ; 5.0; 60.0;;]
  "C_E"      => [17384.5; 17752.9; … ; 38786.7; 38617.8;;]
  "D_H"      => 219841.0
  "K_H"      => 4.05377e5
  "L_G"      => 2.32611e5
  "pi"       => [-0.00749736; -0.00789543; … ; 0.00530319; 0.00100796;;]
  "w_UB"     => 4.06546
  "Y"        => [1.04531e5; 1.05062e5; … ; 1.35709e5; 1.34636e5;;]
  "Y_I"      => [19804.6; 19944.1; … ; 36883.2; 36575.9;;]
  "Y_EA"     => 2.35485e6
  "D_I"      => 54049.0
  ⋮          => ⋮

Set the seed, initialise the model and run one simulation

Random.seed!(1234)
model = Bit.Model(par, init);
Bit.run!(model, 40);

Now we can experiment with changing expectations of the agents in the model. We will change the function estimate_next_value to make the agents expect the last value of the time series (so to represent backward looking expectations)

function Bit.estimate_next_value(data)
    return data[end]
end

Run the model again, with the same seed

Random.seed!(1234)
model2 = Bit.Model(par, init);
Bit.run!(model, 40);

Plot the results, comparing the two cases as different lines

p1 = plot(model.data.real_gdp, title = "gdp", titlefont = 10, label = "forward looking")
plot!(p1, model2.data.real_gdp, titlefont = 10, label = "backward looking")

p2 = plot(model.data.real_household_consumption, title = "consumption", titlefont = 10)
plot!(p2, model2.data.real_household_consumption, titlefont = 10, label = "backward looking")

plot(p1, p2, layout = (2, 1), legend = true)

Plot all time series

p1 = plot(model.data.real_gdp, title = "gdp", titlefont = 10)
plot!(p1, model2.data.real_gdp, titlefont = 10)
p2 = plot(model.data.real_household_consumption, title = "household cons.", titlefont = 10)
plot!(p2, model2.data.real_household_consumption, titlefont = 10)
p3 = plot(model.data.real_government_consumption, title = "gov. cons.", titlefont = 10)
plot!(p3, model2.data.real_government_consumption, titlefont = 10)
p4 = plot(model.data.real_capitalformation, title = "capital form.", titlefont = 10)
plot!(p4, model2.data.real_capitalformation, titlefont = 10)
p5 = plot(model.data.real_exports, title = "exports", titlefont = 10)
plot!(p5, model2.data.real_exports, titlefont = 10)
p6 = plot(model.data.real_imports, title = "imports", titlefont = 10)
plot!(p6, model2.data.real_imports, titlefont = 10)
p7 = plot(model.data.wages, title = "wages", titlefont = 10)
plot!(p7, model2.data.wages, titlefont = 10)
p8 = plot(model.data.euribor, title = "euribor", titlefont = 10)
plot!(p8, model2.data.euribor, titlefont = 10)
p9 = plot(model.data.nominal_gdp ./ model.data.real_gdp, title = "gdp deflator", titlefont = 10)
plot!(p9, model2.data.nominal_gdp ./ model2.data.real_gdp, titlefont = 10)

plot(p1, p2, p3, p4, p5, p6, p7, p8, p9, layout = (3, 3), legend = false)

Note that, importantly, once the function estimate_next_value has been changed, the model will use the new expectations in all the simulations, unless the function is changed again. To restore the original expectations you could close the Julia session.