Merge pull request #144 from sbenthall/sb-0.10.8b

fixes for 0.10.b

notebooks/IncExpectationExample.py

@@ -8,8 +8,8 @@
 #     text_representation:
 #       extension: .py
 #       format_name: percent
-#       format_version: '1.2'
-#       jupytext_version: 1.2.4
+#       format_version: '1.3'
+#       jupytext_version: 1.6.0
 #   kernelspec:
 #     display_name: Python 3
 #     language: python
@@ -23,7 +23,35 @@
 #     name: python
 #     nbconvert_exporter: python
 #     pygments_lexer: ipython3
-#     version: 3.6.9
+#     version: 3.7.5
+#   latex_envs:
+#     LaTeX_envs_menu_present: true
+#     autoclose: false
+#     autocomplete: true
+#     bibliofile: biblio.bib
+#     cite_by: apalike
+#     current_citInitial: 1
+#     eqLabelWithNumbers: true
+#     eqNumInitial: 1
+#     hotkeys:
+#       equation: Ctrl-E
+#       itemize: Ctrl-I
+#     labels_anchors: false
+#     latex_user_defs: false
+#     report_style_numbering: false
+#     user_envs_cfg: false
+#   toc:
+#     base_numbering: 1
+#     nav_menu: {}
+#     number_sections: true
+#     sideBar: true
+#     skip_h1_title: false
+#     title_cell: Table of Contents
+#     title_sidebar: Contents
+#     toc_cell: false
+#     toc_position: {}
+#     toc_section_display: true
+#     toc_window_display: false
 # ---
 
 # %% [markdown]
@@ -60,7 +88,7 @@
 from HARK.utilities import getLorenzShares, calcSubpopAvg
 mystr = lambda number : "{:.4f}".format(number)
 
-# %% {"code_folding": [1]}
+# %% {"code_folding": []}
 # This cell makes a subclass of PersistentShockConsumerType including the MPC 
 class PersistentShockConsumerTypeX(PersistentShockConsumerType):
     def getControls(self):
@@ -70,10 +98,10 @@ def getControls(self):
             these = t == self.t_cycle
             cLvlNow[these] = self.solution[t].cFunc(self.state_now["mLvlNow"][these],self.state_now["pLvlNow"][these])
             MPCnow[these]  =self.solution[t].cFunc.derivativeX(self.state_now["mLvlNow"][these],self.state_now["pLvlNow"][these])
-        self.cLvlNow = cLvlNow
+        self.controls["cLvlNow"] = cLvlNow
         self.MPCnow  = MPCnow
 
-# %% {"code_folding": [1]}
+# %% {"code_folding": []}
 # This cell defines a dictionary to make an instance of "persistent shocks" consumer for the infinite horizon model
 BaselineDict = {
         "CRRA": 2.0,                           # Coefficient of relative risk aversion
@@ -114,7 +142,7 @@ def getControls(self):
         "PrstIncCorr": 0.99,                   # Serial correlation coefficient for persistence of income
     }
 
-# %% {"code_folding": [1]}
+# %% {"code_folding": []}
 # This cell defines a function to solve and simulate a consumer misperceiving the correlation of persistent income shocks
 def runRoszypalSchlaffmanExperiment(CorrAct, CorrPcvd, DiscFac_center, DiscFac_spread):
     '''
@@ -178,7 +206,7 @@ def runRoszypalSchlaffmanExperiment(CorrAct, CorrPcvd, DiscFac_center, DiscFac_s
         type_list.append(ThisType)
 
     # Get the most recent simulated values of X = cLvlNow, MPCnow, aLvlNow, pLvlNow for all types   
-    cLvl_all = np.concatenate([ThisType.cLvlNow for ThisType in type_list])
+    cLvl_all = np.concatenate([ThisType.controls["cLvlNow"] for ThisType in type_list])
     aLvl_all = np.concatenate([ThisType.state_now["aLvlNow"] for ThisType in type_list])
     MPC_all = np.concatenate([ThisType.MPCnow for ThisType in type_list])
     pLvl_all = np.concatenate([ThisType.state_now["pLvlNow"] for ThisType in type_list])

notebooks/KeynesFriedmanModigliani.py

@@ -8,8 +8,8 @@
 #     text_representation:
 #       extension: .py
 #       format_name: percent
-#       format_version: '1.2'
-#       jupytext_version: 1.2.4
+#       format_version: '1.3'
+#       jupytext_version: 1.6.0
 #   kernelspec:
 #     display_name: Python 3
 #     language: python
@@ -23,7 +23,7 @@
 #     name: python
 #     nbconvert_exporter: python
 #     pygments_lexer: ipython3
-#     version: 3.6.9
+#     version: 3.7.5
 #   latex_envs:
 #     LaTeX_envs_menu_present: true
 #     autoclose: false
@@ -40,14 +40,26 @@
 #     latex_user_defs: false
 #     report_style_numbering: false
 #     user_envs_cfg: false
+#   toc:
+#     base_numbering: 1
+#     nav_menu: {}
+#     number_sections: true
+#     sideBar: true
+#     skip_h1_title: false
+#     title_cell: Table of Contents
+#     title_sidebar: Contents
+#     toc_cell: false
+#     toc_position: {}
+#     toc_section_display: true
+#     toc_window_display: false
 # ---
 
 # %% [markdown]
 # ## Introduction: Keynes, Friedman, Modigliani
 #
 # [![badge](https://img.shields.io/badge/Launch%20using%20-Econ--ARK-blue)](https://econ-ark.org/materials/keynesfriedmanmodigliani#launch)
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Some initial setup
 import sys
 import os
@@ -64,11 +76,6 @@
 from copy  import deepcopy
 
 import pandas_datareader.data as web
-# As of 09/03/2019 the latest available version of pandas-datareader
-# has conflicts with the latest version of pandas. We temporarily fix
-# this by loading data from files.
-# This should not be necessary when pandas-datareader>0.7 becomes available.
-from io import StringIO
 
 from HARK.ConsumptionSaving.ConsIndShockModel import PerfForesightConsumerType
 from HARK.utilities import plotFuncsDer, plotFuncs
@@ -97,7 +104,7 @@
 # %% [markdown]
 # #### The Keynesian Consumption Function
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 class KeynesianConsumer:
     """
     This class represents consumers that behave according to a
@@ -124,7 +131,7 @@ def __init__(self):
         self.a1 = self.cFunc(1) - self.cFunc(0)
 
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Plot cFunc(Y)=Y against the Keynesian consumption function
 # Deaton-Friedman consumption function is a special case of perfect foresight model
 
@@ -143,7 +150,7 @@ def __init__(self):
 plt.legend()
 plt.show()
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # This looks like the first of the three equations, consumption as a linear function of income!
 # This means that even in a microfounded model (that HARK provides), the consumption function can match Keynes reduced form
 # prediction (given the right parameterization).
@@ -271,7 +278,7 @@ def __init__(self):
 #
 # We begin by creating a class that class implements the Friedman PIH consumption function as a special case of the [Perfect Foresight CRRA](http://econ.jhu.edu/people/ccarroll/courses/choice/lecturenotes/consumption/PerfForesightCRRA) model.
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 class FriedmanPIHConsumer:
     """
     This class represents consumers that behave according to
@@ -386,7 +393,7 @@ def __init__(self, Rfree=1.001, CRRA = 2):
 #
 # Consider quarterly differences first:
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Lets use the data from FRED that we used before.
 
 # Using quarterly data (copying from above), we had:
@@ -406,7 +413,7 @@ def __init__(self, Rfree=1.001, CRRA = 2):
 # %% [markdown]
 # And now consider longer time differences, 20 quarters for instance, where the changes in permanent income should dominate transitory effects
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Using longer differences
 df_diff_long = df.diff(periods = 20) #create dataframe of differenced values
 df_diff_long.columns = ['cons', 'inc']

notebooks/Nondurables-During-Great-Recession.py

@@ -8,8 +8,8 @@
 #     text_representation:
 #       extension: .py
 #       format_name: percent
-#       format_version: '1.2'
-#       jupytext_version: 1.2.4
+#       format_version: '1.3'
+#       jupytext_version: 1.6.0
 #   kernelspec:
 #     display_name: Python 3
 #     language: python
@@ -23,7 +23,35 @@
 #     name: python
 #     nbconvert_exporter: python
 #     pygments_lexer: ipython3
-#     version: 3.6.9
+#     version: 3.7.5
+#   latex_envs:
+#     LaTeX_envs_menu_present: true
+#     autoclose: false
+#     autocomplete: true
+#     bibliofile: biblio.bib
+#     cite_by: apalike
+#     current_citInitial: 1
+#     eqLabelWithNumbers: true
+#     eqNumInitial: 1
+#     hotkeys:
+#       equation: Ctrl-E
+#       itemize: Ctrl-I
+#     labels_anchors: false
+#     latex_user_defs: false
+#     report_style_numbering: false
+#     user_envs_cfg: false
+#   toc:
+#     base_numbering: 1
+#     nav_menu: {}
+#     number_sections: true
+#     sideBar: true
+#     skip_h1_title: false
+#     title_cell: Table of Contents
+#     title_sidebar: Contents
+#     toc_cell: false
+#     toc_position: {}
+#     toc_section_display: true
+#     toc_window_display: false
 # ---
 
 # %% [markdown]
@@ -33,7 +61,7 @@
 #
 # <p style="text-align: center;"><small><small><small>Generator: QuARK-make/notebooks_byname</small></small></small></p>
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Initial imports and notebook setup, click arrow to show
 
 import matplotlib.pyplot as plt
@@ -74,7 +102,7 @@
 # With this basic setup, HARK's IndShockConsumerType is the appropriate subclass of $\texttt{AgentType}$. So we need to prepare the parameters to create instances of that class.
 #
 
-# %% {"code_folding": [1]}
+# %% {"code_folding": []}
 # Choose some calibrated parameters that roughly match steady state 
 init_infinite = {
     "CRRA":1.0,                    # Coefficient of relative risk aversion 
@@ -124,7 +152,7 @@
 #
 # First, let's create a list with seven copies of our baseline type.
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # A list in python can contain anything -- including consumers
 num_consumer_types   = 7 # declare the number of types we want
 ConsumerTypes = [] # initialize an empty list
@@ -138,7 +166,7 @@
 # %% [markdown]
 # Now we can give each of the consumer types their own discount factor. (This approximates the distribution of parameters estimated in ["The Distribution of Wealth and the Marginal Propensity to Consume"](http://econ.jhu.edu/people/ccarroll/papers/cstwMPC)). 
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Seven types is enough to approximate the uniform distribution (5 is not quite enough)
 from HARK.distribution import Uniform
 
@@ -159,7 +187,7 @@
 #
 # The cell below does both of those tasks, looping through the consumer types. For each one, it solves that type's infinite horizon model, then simulates 1000 periods to generate an approximation to the long run distribution of wealth.
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # tqdm presents a pretty bar that interactively shows how far the calculations have gotten
 for ConsumerType in tqdm(ConsumerTypes):
     ## We configured their discount factor above.  Now solve
@@ -175,7 +203,7 @@
 #
 # First, let's define a simple function that merely calculates the average consumption level across the entire population in the most recent simulated period.
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # We just merge the cNrm and pNrm lists already constructed for each ConsumerType
 def calcAvgC(ConsumerTypes):
     """
@@ -184,7 +212,7 @@ def calcAvgC(ConsumerTypes):
     """
     # Make arrays with all types' (normalized) consumption and permanent income level
     # The brackets indicate that the contents will be a list (in this case, of lists)
-    cNrm = np.concatenate([ThisType.cNrmNow for ThisType in ConsumerTypes])
+    cNrm = np.concatenate([ThisType.controls["cNrmNow"] for ThisType in ConsumerTypes])
     pLvl = np.concatenate([ThisType.state_now["pLvlNow"] for ThisType in ConsumerTypes])
 
     # Calculate and return average consumption level in the economy
@@ -205,7 +233,7 @@ def calcAvgC(ConsumerTypes):
 #  6. Calculate the new average consumption level as percentage change vs the prior level.
 # 3. Return the list of percentage changes
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Whenever you define a function, you should describe it (with a "docstring")
 def calcConsChangeAfterUncertaintyChange(OriginalTypes,NewVals,ParamToChange):
     '''
@@ -258,7 +286,7 @@ def calcConsChangeAfterUncertaintyChange(OriginalTypes,NewVals,ParamToChange):
 # %% [markdown]
 # Our counterfactual experiment function takes three inputs-- consumer types, counterfactual values, and the name of the parameter we want to change. For the sake of convenience, let's define small functions to run the experiment for each parameter with just a single input.
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Trivial functions can be useful in making the logic of your program clear
 def calcConsChangeAfterPermShkChange(newVals):
     return calcConsChangeAfterUncertaintyChange(ConsumerTypes,newVals,"PermShkStd")
@@ -272,7 +300,7 @@ def calcConsChangeAfterUnempPrbChange(newVals):
 # %% [markdown]
 # Now we can finally run our experiment.  In the cell below, we generate a plot of the change in aggregate consumption vs the (underlying) standard deviation of permanent income shocks.
 
-# %% {"code_folding": [0]}
+# %% {"code_folding": []}
 # Calculate the consequences of an "MIT shock" to the standard deviation of permanent shocks
 ratio_min = 0.8 # minimum number to multiply uncertainty parameter by
 TargetChangeInC = -6.3 # Source: FRED

notebooks/Structural-Estimates-From-Empirical-MPCs-Fagereng-et-al.ipynb

@@ -237,7 +237,7 @@
     "    # Calculate the MPC for each of the four lottery sizes for all agents\n",
     "    for ThisType in EstTypeList:\n",
     "        ThisType.simulate(1)\n",
-    "        c_base = ThisType.cNrmNow\n",
+    "        c_base = ThisType.controls[\"cNrmNow\"]\n",
     "        MPC_this_type = np.zeros((ThisType.AgentCount,4))\n",
     "        for k in range(4): # Get MPC for all agents of this type\n",
     "            Llvl = lottery_size[k]\n",
@@ -369,7 +369,7 @@
       "         Current function value: 0.500319\n",
       "         Iterations: 40\n",
       "         Function evaluations: 78\n",
-      "Time to estimate is 130.09294414520264 seconds.\n",
+      "Time to estimate is 182.82315731048584 seconds.\n",
       "Finished estimating for scaling factor of 1.0 and \"splurge amount\" of $0.0\n",
       "Optimal (beta,nabla) is [0.79054883 0.15918858], simulated MPCs are:\n",
       "[[0.77397136 0.68429879 0.56361473 0.41012588]\n",

notebooks/Structural-Estimates-From-Empirical-MPCs-Fagereng-et-al.py

@@ -223,7 +223,7 @@ def FagerengObjFunc(center,spread,verbose=False):
     # Calculate the MPC for each of the four lottery sizes for all agents
     for ThisType in EstTypeList:
         ThisType.simulate(1)
-        c_base = ThisType.cNrmNow
+        c_base = ThisType.controls["cNrmNow"]
         MPC_this_type = np.zeros((ThisType.AgentCount,4))
         for k in range(4): # Get MPC for all agents of this type
             Llvl = lottery_size[k]