Exemple #1
0
 def updateIncomeProcessAlt(self):
     tax_rate = (self.income_unemploy*self.p_unemploy)/(self.l_bar*(1.0-self.p_unemploy))
     xi_dist = deepcopy(calculateMeanOneLognormalDiscreteApprox(self.xi_N,self.xi_sigma[0]))
     xi_dist[0] = np.insert(xi_dist[0]*(1.0-self.p_unemploy),0,self.p_unemploy)
     xi_dist[1] = np.insert(self.l_bar*xi_dist[1]*(1.0-tax_rate),0,self.income_unemploy)
     psi_dist = calculateMeanOneLognormalDiscreteApprox(self.psi_N,self.psi_sigma[0])
     self.income_distrib = [createFlatStateSpaceFromIndepDiscreteProbs(psi_dist,xi_dist)]
     if self.income_unemploy > 0:
         self.p_zero_income = [0.0]
     else:
         self.p_zero_income = [self.p_unemploy]
     if not 'income_distrib' in self.time_vary:
         self.time_vary.append('income_distrib')
     if not 'p_zero_income' in self.time_vary:
         self.time_vary.append('p_zero_income')
Exemple #2
0
 def updateIncomeProcess(self):
     '''
     Updates this agent's income process based on his own attributes.  The
     function that generates the discrete income process can be swapped out
     for a different process.
     '''
     original_time = self.time_flow
     self.timeFwd()
     perm_shocks = calculateMeanOneLognormalDiscreteApprox(self.psi_N, self.psi_sigma)
     temp_shocks = addDiscreteOutcomeConstantMean(calculateMeanOneLognormalDiscreteApprox(self.psi_N, self.xi_sigma),self.income_unemploy,self.p_unemploy)
     self.income_distrib = createFlatStateSpaceFromIndepDiscreteProbs(perm_shocks,temp_shocks)
     self.p_zero_income = np.sum(temp_shocks[0][temp_shocks[1] == 0])
     if not 'income_distrib' in self.time_vary:
         self.time_vary.append('income_distrib')
     if not 'p_zero_income' in self.time_vary:
         self.time_vary.append('p_zero_income')
     if not original_time:
         self.timeRev()
Exemple #3
0
def constructLognormalIncomeProcessUnemploymentFailure(parameters):
    """
    Generates a list of discrete approximations to the income process for each
    life period, from end of life to beginning of life.  The process is identical
    to constructLognormalIncomeProcessUnemployment but for a very tiny possibility
    that unemployment benefits are not provided.

    Parameters:
    -----------
    psi_sigma:    [float]
        Array of standard deviations in _permanent_ income uncertainty during
        the agent's life.
    psi_N:      int
        The number of approximation points to be used in the equiprobable
        discrete approximation to the permanent income shock distribution.
    xi_sigma      [float]
        Array of standard deviations in _temporary_ income uncertainty during
        the agent's life.
    xi_N:       int
        The number of approximation points to be used in the equiprobable
        discrete approximation to the permanent income shock distribution.
    p_unemploy:             float
        The probability of becoming unemployed
    p_unemploy_retire:      float
        The probability of not receiving typical retirement income in any retired period
    T_retire:       int
        The index value i equal to the final working period in the agent's life.
        If final_work_index < 0 then there is no retirement.
    income_unemploy:         float
        Income received when unemployed. Often zero.
    income_unemploy_retire:  float
        Income received while "unemployed" when retired. Often zero.
    T_total:       int
        Total number of non-terminal periods in this consumer's life.

    Returns
    =======
    income_distrib:  [income distribution]
        Each element contains the joint distribution of permanent and transitory
        income shocks, as a set of vectors: psi_shock, xi_shock, and pmf. The
        first two are the points in the joint state space, and final vector is
        the joint pmf over those points. For example,
               psi_shock[20], xi_shock[20], and pmf[20]
        refers to the (psi, xi) point indexed by 20, with probability p = pmf[20].
    p_zero_income: [float]
        A list of probabilities of receiving exactly zero income in each period.

    """
    # Unpack the parameters from the input
    psi_sigma = parameters.psi_sigma
    psi_N = parameters.psi_N
    xi_sigma = parameters.xi_sigma
    xi_N = parameters.xi_N
    T_total = parameters.T_total
    p_unemploy = parameters.p_unemploy
    p_unemploy_retire = parameters.p_unemploy_retire
    T_retire = parameters.T_retire
    income_unemploy = parameters.income_unemploy
    income_unemploy_retire = parameters.income_unemploy_retire

    # Set a small possibility of unemployment benefit failure
    p_fail = 0.01

    income_distrib = []  # Discrete approximation to income process
    p_zero_income = []  # Probability of zero income in each period of life

    # Fill out a simple discrete RV for retirement, with value 1.0 (mean of shocks)
    # in normal times; value 0.0 in "unemployment" times with small prob.
    if T_retire >= 0:
        if p_unemploy_retire > 0:
            retire_perm_income_values = np.array(
                [1.0, 1.0]
            )  # Permanent income is deterministic in retirement (2 states for temp income shocks)
            retire_income_values = np.array([
                income_unemploy_retire,
                (1.0 - p_unemploy_retire * income_unemploy_retire) /
                (1.0 - p_unemploy_retire)
            ])
            retire_income_probs = np.array(
                [p_unemploy_retire, 1.0 - p_unemploy_retire])
            if income_unemploy_retire > 0:
                temp_dist = addDiscreteOutcomeConstantMean(
                    [retire_income_values, retire_income_probs],
                    p=(p_fail * p_unemploy_retire),
                    x=0)
                retire_income_values = temp_dist[0]
                retire_income_probs = temp_dist[1]
                retire_perm_income_values = np.array([1.0, 1.0, 1.0])
        else:
            retire_perm_income_values = np.array([1.0])
            retire_income_values = np.array([1.0])
            retire_income_probs = np.array([1.0])
        income_dist_retire = [
            retire_perm_income_values, retire_income_values,
            retire_income_probs
        ]

    # Loop to fill in the list of income_distrib random variables.
    for Tminust in range(
            T_total
    ):  # Iterate over all periods, counting back from the final period.

        if T_retire >= 0 and Tminust < T_retire:
            # Then we are in the "retirement period" and add a retirement income object.
            income_distrib.append(deepcopy(income_dist_retire))
            if income_unemploy_retire == 0:
                p_zero_income.append(p_unemploy_retire)
            else:
                p_zero_income.append(p_fail * p_unemploy_retire)
        else:
            # We are in the "working life" periods.
            temp_xi_dist = calculateMeanOneLognormalDiscreteApprox(
                N=xi_N, sigma=xi_sigma[Tminust])
            if p_unemploy > 0:
                temp_xi_dist = addDiscreteOutcomeConstantMean(
                    temp_xi_dist, p=p_unemploy, x=income_unemploy)
                if income_unemploy > 0:
                    temp_xi_dist = addDiscreteOutcomeConstantMean(
                        temp_xi_dist, p=(p_unemploy * p_fail), x=0)
            temp_psi_dist = calculateMeanOneLognormalDiscreteApprox(
                N=psi_N, sigma=psi_sigma[Tminust])
            income_distrib.append(
                createFlatStateSpaceFromIndepDiscreteProbs(
                    temp_psi_dist, temp_xi_dist))
            if p_unemploy > 0:
                if income_unemploy == 0:
                    p_zero_income.append(p_unemploy)
                else:
                    p_zero_income.append(p_unemploy * p_fail)
            else:
                p_zero_income.append(0)

    return income_distrib, p_zero_income