Computation of the steady state and dimensionality reduction

Note

Most of the code of this section is in the folder 4_HetAgentsFcns, except for prepare_linearization() which is in 5_LinearizationFunctions.

The model features uninsured income shocks $y$ (by assumption, all workers supply the same efficiency units of labor [BBL], so idiosyncratic productivity shocks translate to income shocks) and two assets, bonds $m$ and illiquid capital $k$. Entrepreneurs (last income-state) receive no labor income, but firm profits, while workers additionally receive labor union profits.

The steady state equilibrium contains marginal value functions $V_m$ and $V_k$ on a three-dimensional grid $(m \times k \times y)$ and the ergodic joint distribution over these idiosyncratic states. We do dimensionality reduction [BL] by applying the Discrete Cosine Transformation to the marginal value functions and approximating the joint distribution with a copula and state-dependent marginals.

The main functions are HANKEstim.find_steadystate() and HANKEstim.prepare_linearization():

Overview of find_steadystate

HANKEstim.find_steadystateFunction
find_steadystate(m_par)

Find the stationary equilibrium capital stock.

Returns

  • KSS: steady-state capital stock
  • VmSS, VkSS: marginal value functions
  • distrSS::Array{Float64,3}: steady-state distribution of idiosyncratic states, computed by Ksupply()
  • n_par::NumericalParameters,m_par::ModelParameters
source

The function takes the parameter struct ModelParameters as input m_par (see Parameters).

To find the stationary equilibrium, we proceed in roughly the following steps:

  1. instantiate the parameter struct NumericalParameters as n_par (see Parameters). Within the struct, we set the number of income states [ny] and use the HANKEstim.Tauchen() method to obtain a grid and a transition matrix of income, given the autocorrelation of the income process [m_par.ρ_h]. Then, include entrepreneurial state.
  2. find equilibrium capital stock (by finding a root of HANKEstim.Kdiff()), where the supply of capital by households is calculated in HANKEstim.Ksupply(), which uses the Endogenous Grid Method (see HANKEstim.EGM_policyupdate) to iteratively obtain optimal policies and marginal value functions

Overview of prepare_linearization

HANKEstim.prepare_linearizationFunction
prepare_linearization(KSS, VmSS, VkSS, distrSS, n_par, m_par)

Compute a number of equilibrium objects needed for linearization.

Arguments

  • KSS: steady-state capital stock
  • VmSS, VkSS: marginal value functions
  • distrSS::Array{Float64,3}: steady-state distribution of idiosyncratic states, computed by Ksupply()
  • n_par::NumericalParameters,m_par::ModelParameters

Returns

  • XSS::Array{Float64,1}, XSSaggr::Array{Float64,1}: steady state vectors produced by @writeXSS()
  • indexes, indexes_aggr: structs for accessing XSS,XSSaggr by variable names, produced by @make_fn(), @make_fnaggr()
  • compressionIndexes::Array{Array{Int,1},1}: indexes for compressed marginal value functions ($V_m$ and $V_k$)
  • Copula(x,y,z): function that maps marginals x,y,z to approximated joint distribution, produced by HANKEstim.myinterpolate3()
  • n_par::NumericalParameters,m_par::ModelParameters
  • CDF_SS, CDF_m, CDF_k, CDF_y: cumulative distribution functions (joint and marginals)
  • distrSS::Array{Float64,3}: steady state distribution of idiosyncratic states, computed by Ksupply()
source

We first calculate other equilibrium quantities and produce distributional summary statistics (HANKEstim.distrSummaries()). Next, we reduce the dimensionality:

  1. compute coefficients of the Chebyshev polynomials that serve as basis functions for $V_m$ and $V_k$, using the Discrete Cosine Transformation (Julia-package FFTW), and retain those that explain the most of its variance, up to 100*(1-n_par.reduc) percent. Save their indices in compressionIndexes
  2. prepare a node mesh on which the time-varying linear interpolant of the copula is defined. The grid in each $m$, $k$, and $y$ dimension is selected such that each resulting bin holds approximately the same share of the respective aggregate variable.

Lastly, we collect the steady state values of all model variables in the vector XSS (see @writeXSS). The state variables consist of the marginal distributions over $m$, $k$ and $y$ and the aggregate state variables (collected in state_names). The control variables consist of the steady state marginal value functions (over the full grid) and the aggregate control variables (collected in control_names; these vectors are defined in the main script HANKEstim.jl).

While the steady state marginal value functions have full dimensionality, in the vectors that collect deviations from steady state (in HANKEstim.Fsys(), those are X and XPrime) only the coefficients of the most important Chebyshev polynomials are saved. Additionally, the deviations of the marginal distributions are saved with one entry short of the grid size, since the marginals are restricted to sum up to 1. We manage this by creating the struct indexes (using @make_fn), that has two fields for each variable: steady state value and deviation.

We also construct the vector XSSaggr and the struct indexes_aggr, which are similar to the above but only store (and manage) aggregate variables. This is useful for differentiating only with respect to aggregate variables in the estimation part (see HANKEstim.SGU_estim()).

Warning

Be sure that you edit prepare_linearization() and not prepare_linearization_generated() which will be overwritten by the model parser based on prepare_linearization().

Parameters

The model parameters for the steady state have to be calibrated. We set them in the struct ModelParameters. It also contains all other parameters that are estimated, including the stochastic process-parameters for the aggregate shocks.

HANKEstim.ModelParametersType

ModelParameters()

Collect all model parameters with calibrated values / priors for estimation in a struct.

Uses packages Parameters, FieldMetadata, Flatten. Boolean value denotes whether parameter is estimated.

Example

julia> m_par = ModelParameters();
julia> # Obtain vector of prior distributions of parameters that are estimated.
julia> priors = collect(metaflatten(m_par, prior))
source

The numerical parameters contain the grid (and the meshes) on which the stationary equilibrium is solved, discretization results of HANKEstim.find_steadystate() like the transition matrix of income and the joint distribution, and other parameters that determine the numerical approximation or solution technique, like reduc or sol_algo.

HANKEstim.NumericalParametersType

NumericalParameters()

Collect parameters for the numerical solution of the model in a struct.

Use package Parameters to provide initial values.

Example

julia> n_par = NumericalParameters(mmin = -6.6, mmax = 1000)
source

Find stationary equilibrium: functions

HANKEstim.TauchenFunction
Tauchen(rho,N,sigma,mue)

Generate a discrete approximation to an AR(1) process, following Tauchen (1987).

Uses importance sampling: each bin has probability 1/N to realize

Arguments

  • rho: autocorrelation coefficient
  • N: number of gridpoints
  • sigma: long-run variance
  • mue: mean of the AR(1) process

Returns

  • grid_vec: state vector grid
  • P: transition matrix
  • bounds: bin bounds
source
HANKEstim.KdiffFunction
Kdiff(K_guess,n_par,m_par)

Calculate the difference between the capital stock that is assumed and the capital stock that prevails under that guessed capital stock's implied prices when households face idiosyncratic income risk (Aiyagari model).

Requires global functions employment(K,A,m_par), interest(K,A,N,m_par), wage(K,A,N,m_par), output(K,TFP,N,m_par), and Ksupply().

Arguments

  • K_guess::Float64: capital stock guess
  • n_par::NumericalParameters, m_par::ModelParameters
source
HANKEstim.KsupplyFunction
Ksupply(RB_guess,R_guess,w_guess,profit_guess,n_par,m_par)

Calculate the aggregate savings when households face idiosyncratic income risk.

Idiosyncratic state is tuple $(m,k,y)$, where $m$: liquid assets, $k$: illiquid assets, $y$: labor income

Arguments

  • R_guess: real interest rate illiquid assets
  • RB_guess: nominal rate on liquid assets
  • w_guess: wages
  • profit_guess: profits
  • n_par::NumericalParameters
  • m_par::ModelParameters

Returns

  • K,B: aggregate saving in illiquid (K) and liquid (B) assets
  • TransitionMat,TransitionMat_a,TransitionMat_n: sparse transition matrices (average, with [a] or without [n] adjustment of illiquid asset)
  • distr: ergodic steady state of TransitionMat
  • c_a_star,m_a_star,k_a_star,c_n_star,m_n_star: optimal policies for consumption [c], liquid [m] and illiquid [k] asset, with [a] or without [n] adjustment of illiquid asset
  • V_m,V_k: marginal value functions
source
HANKEstim.EGM_policyupdateFunction
EGM_policyupdate(EVm,EVk,Qminus,πminus,RBminus,Tshock,inc,n_par,m_par,warnme)

Find optimal policies, given marginal continuation values EVm, EVk, today's prices [Qminus, πminus,RBminus], and income [inc], using the Endogenous Grid Method.

Optimal policies are defined on the fixed grid, but optimal asset choices (m and k) are off-grid values.

Returns

  • c_a_star,m_a_star,k_a_star,c_n_star,m_n_star: optimal (on-grid) policies for consumption [c], liquid [m] and illiquid [k] asset, with [a] or without [n] adjustment of illiquid asset
source
HANKEstim.distrSummariesFunction
distrSummaries(distr,c_a_star,c_n_star,n_par,inc,incgross,m_par)

Compute distributional summary statistics, e.g. Gini indexes, top-10% income and wealth shares, and 10%, 50%, and 90%-consumption quantiles.

Arguments

  • distr: joint distribution over bonds, capital and income $(m \times k \times y)$
  • c_a_star,c_n_star: optimal consumption policies with [a] or without [n] capital adjustment
  • n_par::NumericalParameters, m_par::ModelParameters
  • inc: vector of (on grid-)incomes, consisting of labor income (scaled by $\frac{\gamma-\tau^P}{1+\gamma}$, plus labor union-profits), rental income, liquid asset income, capital liquidation income, labor income (scaled by $\frac{1-\tau^P}{1+\gamma}$, without labor union-profits), and labor income (without scaling or labor union-profits)
  • incgross: vector of (on grid-) pre-tax incomes, consisting of labor income (without scaling, plus labor union-profits), rental income, liquid asset income, capital liquidation income, labor income (without scaling or labor union-profits)
source

Collect variables: macros

HANKEstim.@writeXSSMacro
@writeXSS()

Write all single steady state variables into vectors XSS / XSSaggr.

Requires

(module) globals state_names, control_names, aggr_names

source
HANKEstim.@make_fnMacro
@make_fn(fn_name)

Create function fn_name that returns an instance of struct IndexStruct (created by @make_struct), mapping states and controls to indexes inferred from numerical parameters and compression indexes.

Requires

(module) global state_names, control_names

source
HANKEstim.@make_fnaggrMacro
@make_fnaggr(fn_name)

Create function fn_name that returns an instance of struct IndexStructAggr (created by @make_struct_aggr), mapping aggregate states and controls to values 1 to length(aggr_names) (both steady state and deviation from it).

Requires

(module) global aggr_names

source
HANKEstim.@make_structMacro
@make_struct(struct_name)

Make struct struct_name with two fields for every variable name in s_names (state variables) and c_names (control variables), together with fields for distribution-states and marginal value function-controls.

Requires

(module) globals state_names, control_names

source
HANKEstim.@make_struct_aggrMacro
@make_struct_aggr(struct_name)

Make struct struct_name with two fields for every variable name in aggr_names (for steady state value and for deviation from it).

Requires

(module) global aggr_names

source