Apply New York's pre-TCJA 10% medical expense floor (IT-196 line 3)

changelog.d/ny-medical-expense-floor.fixed.md

@@ -0,0 +1 @@
+New York itemized deductions now allow medical and dental expenses only above 10% of federal AGI, the pre-TCJA 26 U.S.C. 213(a) floor that NY Tax Law § 615(a) keeps and Form IT-196 line 3 applies, through a new ny_medical_expense_deduction used by ny_itemized_deductions_max and the Line 40 limitation. They previously used the federal 7.5% floor, which overstated the deduction by up to 2.5% of federal AGI.

policyengine_us/parameters/gov/states/ny/tax/income/deductions/itemized/medical/floor.yaml

@@ -0,0 +1,31 @@
+description: New York allows an itemized deduction for medical and dental expenses exceeding this share of federal adjusted gross income.
+values:
+  # Through 2017, Form IT-201-D line 1 took the medical deduction from federal
+  # Schedule A line 4, after the federal floor: 10% from 2013 (7.5% for filers
+  # 65 and over through 2016, not modeled), then 7.5% for 2017 under Public
+  # Law 115-97 § 11027.
+  2013-01-01: 0.1
+  2017-01-01: 0.075
+  # From 2018, NY Tax Law § 615(a) computes NY itemized deductions as federal
+  # deductions existed immediately prior to Public Law 115-97, so NY keeps the
+  # pre-TCJA 26 U.S.C. 213(a) 10% floor (Form IT-196 line 3).
+  2018-01-01: 0.1
+metadata:
+  unit: /1
+  period: year
+  label: New York itemized medical expense deduction floor
+  reference:
+    - title: NY Tax Law § 615(a)
+      href: https://www.nysenate.gov/legislation/laws/TAX/615
+    - title: 26 U.S. Code § 213(a) (2016 edition, before Public Law 115-97)
+      href: https://www.govinfo.gov/content/pkg/USCODE-2016-title26/html/USCODE-2016-title26-subtitleA-chap1-subchapB-partVII-sec213.htm
+    - title: 2016 NY Form IT-201-D, line 1 (federal Schedule A, line 4)
+      href: https://www.tax.ny.gov/pdf/2016/inc/it201d_2016_fill_in.pdf#page=1
+    - title: 2017 NY Form IT-201-D, line 1 (federal Schedule A, line 4)
+      href: https://www.tax.ny.gov/pdf/2017/inc/it201d_2017_fill_in.pdf#page=1
+    - title: 2017 federal Schedule A, line 3 (multiply line 2 by 7.5%)
+      href: https://www.irs.gov/pub/irs-prior/f1040sa--2017.pdf#page=1
+    - title: 2018 NY Form IT-196, line 3 (multiply line 2 by 10%)
+      href: https://www.tax.ny.gov/pdf/2018/inc/it196_2018_fill_in.pdf#page=1
+    - title: 2025 NY Form IT-196, line 3 (multiply line 2 by 10%)
+      href: https://www.tax.ny.gov/pdf/2025/inc/it196_2025_fill_in.pdf#page=1

policyengine_us/tests/core/test_ny_medical_expense_deduction_invariants.py

@@ -0,0 +1,255 @@
+"""Invariants for the New York itemized medical expense deduction.
+
+NY Tax Law 615(a) computes New York itemized deductions as federal deductions
+existed immediately before Public Law 115-97, so Form IT-196 lines 1-4 keep
+the 10% floor of pre-TCJA 26 USC 213(a): line 4 is medical and dental
+expenses less 10% of federal AGI. The federal floor has been 7.5% since 2017.
+Through 2017, Form IT-201-D line 1 took federal Schedule A line 4, so New
+York followed the federal 7.5% floor in 2017.
+
+Hypothesis draws batches of tax units (single or joint, in New York or
+another state, with AGI from negative to several million, medical expenses,
+real estate taxes and charitable gifts), and a seeded population of 200 units
+adds breadth. Each batch runs as one vectorized simulation, and again with
+every head's medical expenses raised by a drawn non-negative amount. For
+every tax unit:
+
+1. Bound: 0 <= ny_medical_expense_deduction <= itemized_medical_expenses,
+   and it is zero outside New York.
+2. Differential: in New York, ny_medical_expense_deduction equals an
+   independent numpy max(0, expenses - floor * max(AGI, 0)), with the floor
+   read from the forms (Form IT-196 line 3 from 2018; for 2017, the federal
+   Schedule A line 3 that IT-201-D line 1 used) rather than from the model's
+   parameters.
+3. Order: it never exceeds the federal medical_expense_deduction.
+4. Monotone: raising expenses by d raises it by at least 0 and at most d.
+5. Composition: ny_itemized_deductions_max less ny_medical_expense_deduction
+   does not change when medical expenses change.
+6. Line 40: ny_itemized_deductions_phase_out does not change when medical
+   expenses change, because the medical deduction enters worksheet line 1
+   (deductions before the limitation) and line 2 (deductions not subject to
+   it) alike.
+"""
+
+import numpy as np
+import pytest
+from hypothesis import HealthCheck, example, given, settings
+from hypothesis import strategies as st
+
+from policyengine_us import Simulation
+
+TOLERANCE = 0.01  # dollars
+# The model computes in float32, about 7 significant digits. A difference of
+# float32 amounts can be off by a few ulps of its largest operand (half an
+# ulp is about 0.03 near 600,000) even when the result is near zero, so each
+# comparison also allows RTOL times the largest amount that went into it.
+RTOL = 1e-6
+
+# Form IT-196 line 3, "Multiply line 2 by 10% (0.10)", on the 2018, 2021,
+# 2024 and 2025 forms (tax.ny.gov/pdf/<year>/inc/it196_<year>_fill_in.pdf).
+# For 2017, IT-201-D line 1 took federal Schedule A line 4, and the 2017
+# Schedule A line 3 reads "Multiply line 2 by 7.5% (0.075)".
+PUBLISHED_FLOOR = {2017: 0.075, 2018: 0.1, 2021: 0.1, 2024: 0.1, 2025: 0.1}
+YEARS = sorted(PUBLISHED_FLOOR)
+STATES = ["NY", "NY", "NY", "CA", "TX"]
+
+agis = st.one_of(
+    st.integers(-200_000, 0).map(float),
+    st.integers(0, 500_000).map(float),
+    st.integers(500_000, 6_000_000).map(float),
+)
+amounts = st.one_of(
+    st.just(0.0),
+    st.integers(1, 60_000).map(float),
+    st.integers(60_000, 800_000).map(float),
+)
+increases = st.one_of(st.just(0.0), st.integers(1, 100_000).map(float))
+
+
+@st.composite
+def tax_units(draw):
+    return {
+        "joint": draw(st.booleans()),
+        "state": draw(st.sampled_from(STATES)),
+        "agi": draw(agis),
+        "medical": draw(amounts),
+        "real_estate_taxes": draw(amounts),
+        "charity": draw(amounts),
+        "increase": draw(increases),
+    }
+
+
+SEED = 20261006
+
+
+def _seeded_units(n=200):
+    rng = np.random.default_rng(SEED)
+
+    def amount():
+        return float(
+            [0, rng.integers(1, 60_001), rng.integers(60_000, 800_001)][
+                rng.choice([0, 1, 2])
+            ]
+        )
+
+    return [
+        {
+            "joint": bool(rng.random() < 0.5),
+            "state": str(rng.choice(STATES)),
+            "agi": float(
+                [
+                    rng.integers(-200_000, 1),
+                    rng.integers(0, 500_001),
+                    rng.integers(500_000, 6_000_001),
+                ][rng.choice([0, 1, 2])]
+            ),
+            "medical": amount(),
+            "real_estate_taxes": amount(),
+            "charity": amount(),
+            "increase": float(rng.integers(0, 100_001)) if rng.random() < 0.7 else 0.0,
+        }
+        for _ in range(n)
+    ]
+
+
+def _situation(units, year, *, raised):
+    people, groups = {}, {"tax_units": {}, "households": {}, "marital_units": {}}
+    for i, unit in enumerate(units):
+        head = f"head_{i}"
+        people[head] = {
+            "age": 45,
+            "is_tax_unit_head": True,
+            "other_medical_expenses": unit["medical"] + raised * unit["increase"],
+            "real_estate_taxes": unit["real_estate_taxes"],
+            "charitable_cash_donations": unit["charity"],
+        }
+        members = [head]
+        if unit["joint"]:
+            spouse = f"spouse_{i}"
+            people[spouse] = {"age": 43, "is_tax_unit_spouse": True}
+            members.append(spouse)
+        groups["marital_units"][f"couple_{i}"] = {"members": members}
+        groups["tax_units"][f"tax_unit_{i}"] = {
+            "members": members,
+            "adjusted_gross_income": {year: unit["agi"]},
+        }
+        groups["households"][f"household_{i}"] = {
+            "members": members,
+            "state_code": {year: unit["state"]},
+        }
+    people = {
+        name: {key: {year: value} for key, value in values.items()}
+        for name, values in people.items()
+    }
+    return {"people": people, **groups}
+
+
+def _run(units, year, *, raised=False):
+    sim = Simulation(situation=_situation(units, year, raised=raised))
+    return {
+        name: np.asarray(sim.calculate(name, year), dtype=float)
+        for name in [
+            "itemized_medical_expenses",
+            "adjusted_gross_income",
+            "medical_expense_deduction",
+            "ny_medical_expense_deduction",
+            "ny_itemized_deductions_max",
+            "ny_itemized_deductions_phase_out",
+        ]
+    }
+
+
+def _slack(*operands):
+    return TOLERANCE + RTOL * np.max(np.abs(np.stack(operands)), axis=0)
+
+
+def _assert_close(actual, desired, slack, what):
+    off = np.abs(actual - desired) > slack
+    assert not off.any(), f"{what}: {actual[off]} != {desired[off]}"
+
+
+def _check(units, year):
+    in_ny = np.array([unit["state"] == "NY" for unit in units])
+    base = _run(units, year)
+    expenses = base["itemized_medical_expenses"]
+    deduction = base["ny_medical_expense_deduction"]
+
+    agi = np.maximum(base["adjusted_gross_income"], 0)
+    slack = _slack(expenses, agi)
+
+    # 1. Bounds, and zero outside New York.
+    assert (deduction >= 0).all()
+    assert (deduction <= expenses + slack).all()
+    assert (deduction[~in_ny] == 0).all()
+
+    # 2. Differential against numpy and the published floor.
+    expected = np.maximum(expenses - PUBLISHED_FLOOR[year] * agi, 0)
+    _assert_close(deduction[in_ny], expected[in_ny], slack[in_ny], "differential")
+
+    # 3. Never more than the federal deduction, whose floor is no higher.
+    assert (deduction <= base["medical_expense_deduction"] + slack).all()
+
+    # 4. Monotone, and at most dollar for dollar, in expenses.
+    raised = _run(units, year, raised=True)
+    raised_slack = _slack(raised["itemized_medical_expenses"], agi)
+    increase = raised["itemized_medical_expenses"] - expenses
+    gain = raised["ny_medical_expense_deduction"] - deduction
+    assert (gain >= -raised_slack).all()
+    assert (gain <= increase + raised_slack).all()
+
+    # 5. Medical expenses reach the NY itemized total only through the NY
+    # medical deduction.
+    total_slack = _slack(
+        raised["ny_itemized_deductions_max"], base["ny_itemized_deductions_max"]
+    )
+    _assert_close(
+        raised["ny_itemized_deductions_max"] - raised["ny_medical_expense_deduction"],
+        base["ny_itemized_deductions_max"] - deduction,
+        total_slack,
+        "composition",
+    )
+
+    # 6. The Line 40 limitation is unaffected by medical expenses.
+    _assert_close(
+        raised["ny_itemized_deductions_phase_out"],
+        base["ny_itemized_deductions_phase_out"],
+        total_slack,
+        "Line 40",
+    )
+
+
+# A batch's cost is mostly per-variable overhead, so each example is a large
+# batch and there are few examples.
+SETTINGS = dict(
+    max_examples=8,
+    deadline=None,
+    derandomize=True,
+    suppress_health_check=[HealthCheck.too_slow, HealthCheck.data_too_large],
+)
+
+
+def _near_floor(agi, medical):
+    return {
+        "joint": False,
+        "state": "NY",
+        "agi": float(agi),
+        "medical": float(medical),
+        "real_estate_taxes": 0.0,
+        "charity": 0.0,
+        "increase": 0.0,
+    }
+
+
+@settings(**SETTINGS)
+@given(st.sampled_from(YEARS), st.lists(tax_units(), min_size=5, max_size=30))
+# Expenses within float32 rounding of a large floor, where a tolerance scaled
+# by the result rather than the operands fails (review of #9931).
+@example(2025, [_near_floor(5_999_999, 600_000)] * 5)
+@example(2025, [_near_floor(5_999_997, 600_000), _near_floor(5_242_881, 524_290)])
+def test_ny_medical_expense_deduction_invariants(year, units):
+    _check(units, year)
+
+
+@pytest.mark.parametrize("year", YEARS)
+def test_seeded_population(year):
+    _check(_seeded_units(), year)

policyengine_us/tests/policy/baseline/gov/states/ny/tax/income/taxable_income/deductions/itemized/ny_itemized_deductions_phase_out.yaml

@@ -10,7 +10,7 @@
   input:
     ny_itemized_deductions_max: 100_000
     ny_college_tuition_deduction: 0
-    medical_expense_deduction: 0
+    ny_medical_expense_deduction: 0
     investment_interest_expense: 0
     ny_casualty_loss_deduction: 0
     adjusted_gross_income: 340_000
@@ -26,7 +26,7 @@
   input:
     ny_itemized_deductions_max: 100_000
     ny_college_tuition_deduction: 0
-    medical_expense_deduction: 0
+    ny_medical_expense_deduction: 0
     investment_interest_expense: 0
     ny_casualty_loss_deduction: 0
     adjusted_gross_income: 350_000
@@ -45,7 +45,7 @@
   input:
     ny_itemized_deductions_max: 100_000
     ny_college_tuition_deduction: 0
-    medical_expense_deduction: 0
+    ny_medical_expense_deduction: 0
     investment_interest_expense: 0
     ny_casualty_loss_deduction: 0
     adjusted_gross_income: 408_000
@@ -61,7 +61,7 @@
   input:
     ny_itemized_deductions_max: 100_000
     ny_college_tuition_deduction: 0
-    medical_expense_deduction: 0
+    ny_medical_expense_deduction: 0
     investment_interest_expense: 0
     ny_casualty_loss_deduction: 0
     adjusted_gross_income: 500_000
@@ -80,7 +80,7 @@
   input:
     ny_itemized_deductions_max: 20_000
     ny_college_tuition_deduction: 0
-    medical_expense_deduction: 0
+    ny_medical_expense_deduction: 0
     investment_interest_expense: 0
     ny_casualty_loss_deduction: 0
     adjusted_gross_income: 5_000_000
@@ -102,7 +102,7 @@
     # remaining 60,000 is subject to the limitation.
     ny_itemized_deductions_max: 100_000
     ny_college_tuition_deduction: 0
-    medical_expense_deduction: 25_000
+    ny_medical_expense_deduction: 25_000
     investment_interest_expense: 10_000
     ny_casualty_loss_deduction: 5_000
     adjusted_gross_income: 5_000_000
@@ -123,7 +123,7 @@
     # the federal Line 40 worksheet; the Pease base should ignore it.
     ny_itemized_deductions_max: 130_000
     ny_college_tuition_deduction: 30_000
-    medical_expense_deduction: 0
+    ny_medical_expense_deduction: 0
     investment_interest_expense: 0
     ny_casualty_loss_deduction: 0
     adjusted_gross_income: 500_000