Fix basic state pension calculation (#1208)

* fix basic state pension calculation

* fix basic state pension calculation

* Merge branch 'fix-state-pension-calculation' of https://github.com/policyengine/policyengine-uk into fix-state-pension-calculation

* format

* debug

* debug

* debug

* Fix basic state pension calculation to work without datasets

- Allow basic_state_pension to calculate without a dataset by using
  the current period when no dataset is present
- Keep correct behavior: return 0 for non-BASIC pension types
- Only apply triple lock uprating when using dataset data
- Use correct indices.triple_lock (cumulative index) not yoy_growth
- Fix test expectations for non-BASIC types (should return 0)

Fixes #1090

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: policyengine-bot <bot@policyengine.org>
Co-authored-by: Claude <noreply@anthropic.com>

Author: 3 people

.gitignore

@@ -56,7 +56,7 @@ policyengine_uk/calibration/*.h5
 !docs/**/*.ipynb
 !uprating_growth_factors.csv
 
-# Virtual environments
+# Virtual environments
 .venv/
 venv/
 env/
@@ -82,4 +82,4 @@ coverage.xml
 cover/
 
 # Docs
-**/_build/
+**/_build/

changelog_entry.yaml

@@ -0,0 +1,4 @@
+- bump: patch
+  changes:
+    fixed:
+      - Basic state pension calculation.

policyengine_uk/tests/policy/baseline/gov/dwp/basic_state_pension.yaml

@@ -0,0 +1,44 @@
+- name: Basic state pension calculation with normal parameters
+  period: 2021
+  input:
+    people:
+      person1:
+        state_pension_reported: 5000
+        state_pension_type: BASIC
+        age: 70
+  output:
+    basic_state_pension: 5000  # Reported amount (under maximum)
+
+- name: Basic state pension caps at maximum when reported exceeds maximum
+  period: 2021
+  input:
+    people:
+      person1:
+        state_pension_reported: 10000  # Higher than maximum
+        state_pension_type: BASIC
+        age: 70
+  output:
+    basic_state_pension: 7155.2  # Capped at maximum (137.60 * 52)
+
+- name: Non-basic state pension type returns zero for basic_state_pension
+  period: 2021
+  input:
+    people:
+      person1:
+        state_pension_reported: 5000
+        state_pension_type: NEW  # Not basic pension
+        age: 70
+  output:
+    basic_state_pension: 0  # basic_state_pension is only for BASIC type
+
+- name: Basic state pension should be zero when maximum is zero
+  period: 2021
+  input:
+    gov.dwp.state_pension.basic_state_pension.amount: 0
+    people:
+      person1:
+        state_pension_reported: 5000
+        state_pension_type: BASIC
+        age: 70
+  output:
+    basic_state_pension: 0

policyengine_uk/variables/gov/dwp/basic_state_pension.py

@@ -10,22 +10,38 @@ class basic_state_pension(Variable):
 
     def formula(person, period, parameters):
         simulation = person.simulation
-        if simulation.dataset is None:
-            return 0
-        try:
-            data_year = min(simulation.dataset.years)
-        except:
+        has_dataset = simulation.dataset is not None
+
+        # Determine the data year: from dataset if available, otherwise current
+        if has_dataset:
+            try:
+                data_year = min(simulation.dataset.years)
+            except:
+                data_year = period.start.year
+        else:
             data_year = period.start.year
+
         reported = person("state_pension_reported", data_year) / WEEKS_IN_YEAR
-        type = person("state_pension_type", period)
+        pension_type = person("state_pension_type", period)
         maximum_basic_sp = parameters(
             data_year
         ).gov.dwp.state_pension.basic_state_pension.amount
+
+        # For BASIC pension type, cap at the maximum; otherwise return 0
         amount_in_data_year = where(
-            type == type.possible_values.BASIC,
+            pension_type == pension_type.possible_values.BASIC,
             min_(reported, maximum_basic_sp),
             0,
         )
-        triple_lock = parameters.gov.economic_assumptions.indices.triple_lock
-        uprating_since_data_year = triple_lock(period) / triple_lock(data_year)
-        return amount_in_data_year * uprating_since_data_year * WEEKS_IN_YEAR
+
+        # Apply triple lock uprating only when using dataset
+        # (i.e., when data year differs from simulation period)
+        if has_dataset:
+            triple_lock = (
+                parameters.gov.economic_assumptions.indices.triple_lock
+            )
+            uprating = triple_lock(period) / triple_lock(data_year)
+        else:
+            uprating = 1
+
+        return amount_in_data_year * uprating * WEEKS_IN_YEAR