diff --git a/sklearn/linear_model/coordinate_descent.py b/sklearn/linear_model/coordinate_descent.py
index 1b6c42844ab43..d3dfa1c80724b 100644
--- a/sklearn/linear_model/coordinate_descent.py
+++ b/sklearn/linear_model/coordinate_descent.py
@@ -1797,7 +1797,7 @@ def fit(self, X, y):
         X, y, X_offset, y_offset, X_scale = _preprocess_data(
             X, y, self.fit_intercept, self.normalize, copy=False)
 
-        if not self.warm_start or not hasattr(self, "coef_"):
+        if not self.warm_start or getattr(self, "coef_", None) is None:
             self.coef_ = np.zeros((n_tasks, n_features), dtype=X.dtype.type,
                                   order='F')
 
diff --git a/sklearn/linear_model/tests/test_coordinate_descent.py b/sklearn/linear_model/tests/test_coordinate_descent.py
index 630cfe558a350..d5ed17cce0042 100644
--- a/sklearn/linear_model/tests/test_coordinate_descent.py
+++ b/sklearn/linear_model/tests/test_coordinate_descent.py
@@ -405,7 +405,8 @@ def test_multi_task_lasso_and_enet():
     assert 0 < clf.dual_gap_ < 1e-5
     assert_array_almost_equal(clf.coef_[0], clf.coef_[1])
 
-    clf = MultiTaskElasticNet(alpha=1.0, tol=1e-8, max_iter=1)
+    clf = MultiTaskElasticNet(alpha=1.0, tol=1e-8, max_iter=1, warm_start=True)
+    clf.coef_ = None  # ensure that this is still supported with warm_start
     assert_warns_message(ConvergenceWarning, 'did not converge', clf.fit, X, Y)