BUG: NA are coerced in to NaN by concat

[davetapley]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

from pandas import NA, DataFrame, concat

a = DataFrame.from_dict({'val': [1], 'na': [NA]})
b = DataFrame.from_dict({'val': [2], 'na': [NA]})

print("a and b:")
print(a)
print(b)

print("concat:")
ab = concat([a, b])
print(ab)

Issue Description

NA are coerced in to NaN by concat:

a and b:
   val    na
0    1  <NA>
   val    na
0    2  <NA>
concat:
   val   na
0    1  NaN
0    2  NaN

Expected Behavior

The NA stay as NA.

Installed Versions

INSTALLED VERSIONS

commit : 0691c5c
python : 3.12.2
python-bits : 64
OS : Linux
OS-release : 6.8.0-1021-azure
Version : #25~22.04.1-Ubuntu SMP Thu Jan 16 21:37:09 UTC 2025
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : C.UTF-8
LOCALE : C.UTF-8

pandas : 2.2.3
numpy : 1.26.4
pytz : 2025.2
dateutil : 2.9.0.post0
pip : 24.0
Cython : None
sphinx : None
IPython : None
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : None
blosc : None
bottleneck : None
dataframe-api-compat : None
fastparquet : None
fsspec : None
html5lib : None
hypothesis : None
gcsfs : None
jinja2 : 3.1.6
lxml.etree : 5.2.2
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
psycopg2 : None
pymysql : None
pyarrow : 17.0.0
pyreadstat : None
pytest : None
python-calamine : None
pyxlsb : None
s3fs : None
scipy : 1.11.2
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlsxwriter : None
zstandard : None
tzdata : 2025.2
qtpy : None
pyqt5 : None

[davetapley]

I found these related, but none address this specific case:

• BUG: concat with empty dataframe with columns passed and nonempty dataframe coerces dtype to object #45637
• BUG: Concatenation of None with numerical values no longer converts None to Nan. #46922
• REGR: revert behaviour change for concat with empty/all-NaN data #47372

[arthurlw]

Thanks for raising this! Confirmed on main. This happens because DataFrame.from_dict({'na': [NA]}) gives the na column an object dtype, so concat falls back to NumPy’s object‐array logic, which doesn’t know about pd.NA and yields NaN instead. To preserve pd.NA you can first cast to a true nullable dtype. E.g.

a = pd.DataFrame.from_dict({'val': [1], 'na': pd.array([pd.NA], dtype='Int64')})
b = pd.DataFrame.from_dict({'val': [2], 'na': pd.array([pd.NA], dtype='Int64')})

This seems related to PDEP-16.

[yuanx749]

Note that in documentation:

Currently, pandas does not use those data types using NA by default in a DataFrame or Series, so you need to specify the dtype explicitly. An easy way to convert to those dtypes is explained in the conversion section.

So you can convert the df like a = a.convert_dtypes().

[rhshadrach]

Agreed with @arthurlw and @yuanx749 with the current workaround, but this coercion from NA to np.nan should also not happen. However work on this should wait for PDEP-16.