Merge pull request #35 from nikhilwoodruff/microsim-fixes

Microsimulation fixes

Makefile

@@ -0,0 +1,5 @@
+all: test
+format:
+	black . -l 79
+test:
+	pytest .

examples/plot.py

@@ -0,0 +1,153 @@
+from ogcore.utils import safe_read_pickle
+from ogcore import txfunc
+import numpy as np
+import pandas as pd
+from argparse import ArgumentParser
+import plotly.express as px
+
+
+def main(year: int = 2018):
+    year = str(year)
+    # read in tax function parameters from pickle
+    tax_funcs_base = safe_read_pickle(
+        "./examples/OG-UK-Example/OUTPUT_BASELINE/TxFuncEst_baseline.pkl"
+    )
+    try:
+        tax_funcs_reform = safe_read_pickle(
+            "./examples/OG-UK-Example/OUTPUT_REFORM/TxFuncEst_policy.pkl"
+        )
+    except:
+        tax_funcs_reform = tax_funcs_base
+
+    # read in micro data from pickle
+    micro_data = safe_read_pickle(
+        "./examples/OG-UK-Example/OUTPUT_BASELINE/micro_data_baseline.pkl"
+    )
+
+    for ftype, name in zip(
+        ("etr", "mtrx", "mtry"), ("etr", "mtr_labinc", "mtr_capinc")
+    ):
+        tax = get_tax_fn(
+            year,
+            year,
+            [tax_funcs_base, tax_funcs_reform],
+            age=None,
+            tax_func_type=["DEP"],
+            rate_type=ftype,
+            over_labinc=True,
+            other_inc_val=1000,
+            max_inc_amt=100000,
+            data_list=[micro_data],
+            labels=["Baseline", "Reform"],
+            title=f"Rate type: DEP, over labour income",
+            path=None,
+        )
+
+        df = pd.DataFrame(
+            {
+                "Labour income": micro_data[year]["total_labinc"],
+                "Capital income": micro_data[year]["total_capinc"],
+                name: micro_data[year][name],
+                "Type": "Actual",
+            }
+        )
+        second_df = df.copy()
+
+        second_df[name] = tax(df["Labour income"], df["Capital income"])
+        second_df["Type"] = "Fitted"
+
+        df = pd.concat([df, second_df])
+
+        px.scatter_3d(
+            df,
+            x="Labour income",
+            y="Capital income",
+            z=name,
+            opacity=0.1,
+            color="Type",
+        ).update_layout(
+            title=f"{name} function in {year}",
+            template="plotly_white",
+        ).show()
+
+
+def get_tax_fn(
+    year,
+    start_year,
+    tax_param_list,
+    age=None,
+    tax_func_type=["DEP"],
+    rate_type="etr",
+    over_labinc=True,
+    other_inc_val=1000,
+    max_inc_amt=1000000,
+    data_list=None,
+    labels=["1st Functions"],
+    title=None,
+    path=None,
+):
+    # Check that inputs are valid
+    year, start_year = int(year), int(start_year)
+    assert year >= start_year
+    # if list of tax function types less than list of params, assume
+    # all the same functional form
+    if len(tax_func_type) < len(tax_param_list):
+        tax_func_type = [tax_func_type[0]] * len(tax_param_list)
+    for i, v in enumerate(tax_func_type):
+        assert v in ["DEP", "DEP_totalinc", "GS", "linear"]
+    assert rate_type in ["etr", "mtrx", "mtry"]
+    assert len(tax_param_list) == len(labels)
+
+    # Set age and year to look at
+    if age is not None:
+        assert isinstance(age, int)
+        s = age - 21
+    else:
+        s = 0  # if not age-specific, all ages have the same values
+    t = year - start_year
+
+    # create rate_key to correspond to keys in tax func dicts
+    rate_key = "tfunc_" + rate_type + "_params_S"
+
+    # Set income range to plot over (min income value hard coded to 5)
+    inc_sup = np.exp(np.linspace(np.log(5), np.log(max_inc_amt), 100))
+    # Set income value for other income
+    inc_fix = other_inc_val
+
+    if over_labinc:
+        key1 = "total_labinc"
+        X = inc_sup
+        Y = inc_fix
+    else:
+        key1 = "total_capinc"
+        X = inc_fix
+        Y = inc_sup
+
+    # get tax rates for each point in the income support and plot
+    for i, tax_params in enumerate(tax_param_list):
+        rates = txfunc.get_tax_rates(
+            tax_params[rate_key][s, t, :],
+            X,
+            Y,
+            None,
+            tax_func_type[i],
+            rate_type,
+            for_estimation=False,
+        )
+
+    return lambda x, y: txfunc.get_tax_rates(
+        tax_params[rate_key][s, t, :],
+        x,
+        y,
+        None,
+        tax_func_type[i],
+        rate_type,
+        for_estimation=False,
+    )
+
+
+if __name__ == "__main__":
+    parser = ArgumentParser()
+    parser.add_argument("--year", type=int, default=2018)
+    args = parser.parse_args()
+    main(args.year)

examples/run_oguk.py

@@ -18,29 +18,11 @@
 
 
 def get_default_reform():
-    from openfisca_uk.api import YEAR, Person, BASELINE_VARIABLES, reforms
+    from openfisca_tools.reforms import set_parameter
 
-    # Costs £68.8bn in year 1
-    UBI_AMOUNT = 20 * 52
-
-    class UBI(Variable):
-        value_type = float
-        entity = Person
-        definition_period = YEAR
-
-        def formula(person, period, parameters):
-            return 20 * 52
-
-    class tax(BASELINE_VARIABLES.tax):
-        def formula(person, period, parameters):
-            return BASELINE_VARIABLES.tax.formula(
-                person, period, parameters
-            ) - person("UBI", period)
-
-    ubi_reform = reforms.structural.new_variable(
-        UBI
-    ), reforms.structural.restructure(tax)
-    return ubi_reform
+    return set_parameter(
+        "tax.income_tax.rates.uk[0].rate", 0.3, "year:2018:10"
+    )
 
 
 start_time = time.time()
@@ -49,7 +31,9 @@ def formula(person, period, parameters):
 from ogcore.parameters import Specifications
 
 
-def main(reform=get_default_reform()):
+def main(reform=None):
+    if reform is None:
+        reform = get_default_reform()
     # Define parameters to use for multiprocessing
     client = Client()
     num_workers = min(multiprocessing.cpu_count(), 7)
@@ -114,18 +98,6 @@ def main(reform=get_default_reform()):
     Run reform policy
     ------------------------------------------------------------------------
     """
-    # Create a parametric reform for OpenFisca-UK
-    def lower_pa(parameters):
-        parameters.tax.income_tax.allowances.personal_allowance.amount.update(
-            period="2018", value=10000
-        )
-        return parameters
-
-    class lower_personal_tax_allowance(Reform):
-        def apply(self):
-            self.modify_parameters(modifier_function=lower_pa)
-
-    reform = lower_personal_tax_allowance
 
     # create new Specifications object for reform simulation
     p2 = Specifications(
@@ -216,13 +188,16 @@ def apply(self):
         description="A script to run the main OG-UK routine on a reform."
     )
     parser.add_argument(
-        "reform",
-        default="small_ubi_reform.ubi_reform",
+        "--reform",
         help="The Python reform object to use as a reform (if `reform` is defined in `reform_file.py`, then use `reform_file.reform`)",
     )
     args = parser.parse_args()
-
-    reform_path = args.reform.split(".")
-    python_module, object_name = ".".join(reform_path[:-1]), reform_path[-1]
-    reform = getattr(__import__(python_module), object_name)
-    main(reform)
+    reform = args.reform
+    if reform is not None:
+        reform_path = reform.split(".")
+        python_module, object_name = (
+            ".".join(reform_path[:-1]),
+            reform_path[-1],
+        )
+        reform = getattr(__import__(python_module), object_name)
+    main(reform=reform)