From 8b12efd43bf24bf5bb87cbef14a80a82523f7520 Mon Sep 17 00:00:00 2001
From: Yao Xiao <108576690+Charlie-XIAO@users.noreply.github.com>
Date: Wed, 21 Jun 2023 01:15:34 +0800
Subject: [PATCH] BUG: resampling empty series loses time zone from dtype
 (#53736)

* BUG: resampling empty series loses time zone from dtype

* changelog added
---
 doc/source/whatsnew/v2.1.0.rst               |  1 +
 pandas/core/resample.py                      | 10 +++++++---
 pandas/tests/resample/test_datetime_index.py | 16 ++++++++++++++++
 3 files changed, 24 insertions(+), 3 deletions(-)

diff --git a/doc/source/whatsnew/v2.1.0.rst b/doc/source/whatsnew/v2.1.0.rst
index d0d98e08998f1..ac4fb6838b9ce 100644
--- a/doc/source/whatsnew/v2.1.0.rst
+++ b/doc/source/whatsnew/v2.1.0.rst
@@ -459,6 +459,7 @@ Plotting
 Groupby/resample/rolling
 ^^^^^^^^^^^^^^^^^^^^^^^^
 - Bug in :meth:`DataFrame.resample` and :meth:`Series.resample` in incorrectly allowing non-fixed ``freq`` when resampling on a :class:`TimedeltaIndex` (:issue:`51896`)
+- Bug in :meth:`DataFrame.resample` and :meth:`Series.resample` losing time zone when resampling empty data (:issue:`53664`)
 - Bug in :meth:`DataFrameGroupBy.idxmin`, :meth:`SeriesGroupBy.idxmin`, :meth:`DataFrameGroupBy.idxmax`, :meth:`SeriesGroupBy.idxmax` return wrong dtype when used on empty DataFrameGroupBy or SeriesGroupBy (:issue:`51423`)
 - Bug in weighted rolling aggregations when specifying ``min_periods=0`` (:issue:`51449`)
 - Bug in :meth:`DataFrame.groupby` and :meth:`Series.groupby`, where, when the index of the
diff --git a/pandas/core/resample.py b/pandas/core/resample.py
index a9faad4cbef6a..3fc0c47dd175e 100644
--- a/pandas/core/resample.py
+++ b/pandas/core/resample.py
@@ -1892,7 +1892,9 @@ def _get_time_bins(self, ax: DatetimeIndex):
             )
 
         if len(ax) == 0:
-            binner = labels = DatetimeIndex(data=[], freq=self.freq, name=ax.name)
+            binner = labels = DatetimeIndex(
+                data=[], freq=self.freq, name=ax.name, dtype=ax.dtype
+            )
             return binner, [], labels
 
         first, last = _get_timestamp_range_edges(
@@ -2023,8 +2025,10 @@ def _get_time_period_bins(self, ax: DatetimeIndex):
 
         freq = self.freq
 
-        if not len(ax):
-            binner = labels = PeriodIndex(data=[], freq=freq, name=ax.name)
+        if len(ax) == 0:
+            binner = labels = PeriodIndex(
+                data=[], freq=freq, name=ax.name, dtype=ax.dtype
+            )
             return binner, [], labels
 
         labels = binner = period_range(start=ax[0], end=ax[-1], freq=freq, name=ax.name)
diff --git a/pandas/tests/resample/test_datetime_index.py b/pandas/tests/resample/test_datetime_index.py
index 389db2d36474d..0f52f2d1c65ee 100644
--- a/pandas/tests/resample/test_datetime_index.py
+++ b/pandas/tests/resample/test_datetime_index.py
@@ -1968,3 +1968,19 @@ def test_long_rule_non_nano():
     )
     expected = Series([1.0, 3.0, 6.5, 4.0, 3.0, 6.5, 4.0, 3.0, 6.5], index=expected_idx)
     tm.assert_series_equal(result, expected)
+
+
+def test_resample_empty_series_with_tz():
+    # GH#53664
+    df = DataFrame({"ts": [], "values": []}).astype(
+        {"ts": "datetime64[ns, Atlantic/Faroe]"}
+    )
+    result = df.resample("2MS", on="ts", closed="left", label="left", origin="start")[
+        "values"
+    ].sum()
+
+    expected_idx = DatetimeIndex(
+        [], freq="2MS", name="ts", dtype="datetime64[ns, Atlantic/Faroe]"
+    )
+    expected = Series([], index=expected_idx, name="values", dtype="float64")
+    tm.assert_series_equal(result, expected)