API: Use lowest possible datetime64 resolution in PeriodIndex.to_timestamp

pandas/core/arrays/period.py

@@ -824,8 +824,10 @@ def to_timestamp(self, freq=None, how: str = "start") -> DatetimeArray:
 
         new_parr = self.asfreq(freq, how=how)
 
+        dt64_dtype = np.dtype("datetime64[ns]")
+
         new_data = libperiod.periodarr_to_dt64arr(new_parr.asi8, base)
-        dta = DatetimeArray._from_sequence(new_data, dtype=np.dtype("M8[ns]"))
+        dta = DatetimeArray._from_sequence(new_data, dtype=dt64_dtype)
 
         if self.freq.name == "B":
             # See if we can retain BDay instead of Day in cases where

pandas/tests/indexes/period/test_period.py

@@ -231,3 +231,29 @@ def test_dunder_array(array):
             np.array(obj, dtype=dtype)
         with pytest.raises(TypeError, match=msg):
             np.array(obj, dtype=getattr(np, dtype))
+
+
+def test_to_timestamp_monthly_resolution():
+    idx = PeriodIndex(["2011-01", "NaT", "2011-02"], freq="2M")
+    ts = idx.to_timestamp()
+    assert ts.dtype == np.dtype("datetime64[ns]")
+
+
+def test_to_timestamp_yearly_resolution():
+    idx = PeriodIndex(["2011", "2012"], freq="Y")
+    ts = idx.to_timestamp()
+    assert ts.dtype == np.dtype("datetime64[ns]")
+
+
+def test_to_timestamp_daily_resolution():
+    idx = PeriodIndex(["2011-01-01", "2011-01-02"], freq="D")
+    ts = idx.to_timestamp()
+    assert ts.dtype == np.dtype("datetime64[ns]")
+
+
+def test_to_timestamp_nanosecond_resolution():
+    idx = PeriodIndex(
+        ["2011-01-01 00:00:00.000000001", "2011-01-01 00:00:00.000000002"], freq="ns"
+    )
+    ts = idx.to_timestamp()
+    assert ts.dtype == np.dtype("datetime64[ns]")