diff --git a/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_expanding.py b/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_expanding.py
index 223adeaa48..634cbd7f0b 100644
--- a/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_expanding.py
+++ b/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_expanding.py
@@ -52,14 +52,17 @@ class OpsOnDiffFramesGroupByExpandingTest(PandasOnSparkTestCase, TestUtils):
         psdf = ps.from_pandas(pdf)
         kkey = ps.from_pandas(pkey)
 
+        # The behavior of GroupBy.expanding is changed from pandas 1.3.
         if LooseVersion(pd.__version__) >= LooseVersion("1.3"):
-            # TODO(SPARK-36367): Fix the behavior to follow pandas >= 1.3
-            pass
-        else:
             self.assert_eq(
                 getattr(psdf.groupby(kkey).expanding(2), f)().sort_index(),
                 getattr(pdf.groupby(pkey).expanding(2), f)().sort_index(),
             )
+        else:
+            self.assert_eq(
+                getattr(psdf.groupby(kkey).expanding(2), f)().sort_index(),
+                getattr(pdf.groupby(pkey).expanding(2), f)().drop("a", axis=1).sort_index(),
+            )
 
         self.assert_eq(
             getattr(psdf.groupby(kkey)["b"].expanding(2), f)().sort_index(),
diff --git a/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_rolling.py b/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_rolling.py
index 4f97769b8e..04ea448d80 100644
--- a/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_rolling.py
+++ b/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_rolling.py
@@ -50,14 +50,17 @@ class OpsOnDiffFramesGroupByRollingTest(PandasOnSparkTestCase, TestUtils):
         psdf = ps.from_pandas(pdf)
         kkey = ps.from_pandas(pkey)
 
+        # The behavior of GroupBy.rolling is changed from pandas 1.3.
         if LooseVersion(pd.__version__) >= LooseVersion("1.3"):
-            # TODO(SPARK-36367): Fix the behavior to follow pandas >= 1.3
-            pass
-        else:
             self.assert_eq(
                 getattr(psdf.groupby(kkey).rolling(2), f)().sort_index(),
                 getattr(pdf.groupby(pkey).rolling(2), f)().sort_index(),
             )
+        else:
+            self.assert_eq(
+                getattr(psdf.groupby(kkey).rolling(2), f)().sort_index(),
+                getattr(pdf.groupby(pkey).rolling(2), f)().drop("a", axis=1).sort_index(),
+            )
 
         self.assert_eq(
             getattr(psdf.groupby(kkey)["b"].rolling(2), f)().sort_index(),