diff --git a/quantecon/markov/gth_solve.py b/quantecon/markov/gth_solve.py
index b3257c7bf..76ca0438c 100644
--- a/quantecon/markov/gth_solve.py
+++ b/quantecon/markov/gth_solve.py
@@ -6,6 +6,9 @@
 import numpy as np
 from numba import jit
 
+from ..util.compat import copy_if_needed
+
+
 def gth_solve(A, overwrite=False, use_jit=True):
     r"""
     This routine computes the stationary distribution of an irreducible
@@ -52,7 +55,9 @@ def gth_solve(A, overwrite=False, use_jit=True):
        Simulation, Princeton University Press, 2009.
 
     """
-    A1 = np.array(A, dtype=float, copy=not overwrite, order='C')
+    copy = copy_if_needed if overwrite else True
+
+    A1 = np.array(A, dtype=float, copy=copy, order='C')
     # `order='C'` is for use with Numba <= 0.18.2
     # See issue github.com/numba/numba/issues/1103
 
diff --git a/quantecon/markov/tests/test_gth_solve.py b/quantecon/markov/tests/test_gth_solve.py
index f5275e869..16651f7e8 100644
--- a/quantecon/markov/tests/test_gth_solve.py
+++ b/quantecon/markov/tests/test_gth_solve.py
@@ -196,6 +196,13 @@ def test_matrices_with_C_F_orders():
     assert_array_equal(computed_F, stationary_dist)
 
 
+def test_unable_to_avoid_copy():
+    A = np.array([[0, 1], [0, 1]])  # dtype=int
+    stationary_dist = [0., 1.]
+    x = gth_solve(A, overwrite=True)
+    assert_array_equal(x, stationary_dist)
+
+
 def test_raises_value_error_non_2dim():
     """Test with non 2dim input"""
     assert_raises(ValueError, gth_solve, np.array([0.4, 0.6]))
diff --git a/quantecon/util/compat.py b/quantecon/util/compat.py
new file mode 100644
index 000000000..5037eec84
--- /dev/null
+++ b/quantecon/util/compat.py
@@ -0,0 +1,23 @@
+"""
+Utilities for compatibility
+
+"""
+from typing import Optional
+import numpy as np
+
+
+# From scipy/_lib/_util.py
+
+copy_if_needed: Optional[bool]
+
+if np.lib.NumpyVersion(np.__version__) >= "2.0.0":
+    copy_if_needed = None
+elif np.lib.NumpyVersion(np.__version__) < "1.28.0":
+    copy_if_needed = False
+else:
+    # 2.0.0 dev versions, handle cases where copy may or may not exist
+    try:
+        np.array([1]).__array__(copy=None)  # type: ignore[call-overload]
+        copy_if_needed = None
+    except TypeError:
+        copy_if_needed = False