Why does column + column concatenation create arrays for some Windows accounts?

Question

When running the below Python code, I get different results depending on the user account/admin privileges that is used. The code is saved as test.py on a Windows Server 2016 machine.

import pandas as pd

data = {'A': ['a', 'b'], 'B': ['c', 'd']}

df = pd.DataFrame(data)

df['C'] = df['A'] + df['B']

print(df.head())

With df['C'] = df['A'] + df['B'], pandas should perform element-wise concatenation of columns A and B. Running python test.py in a CMD with admin gets the following correct results.

   A  B   C
0  a  c  ac
1  b  d  bd

But running the same command in a non-admin CMD gets the below results. In this case, pandas seems to not do element-wise concatenation.

   A  B         C
0  a  c  [ac, ad]
1  b  d  [bc, bd]

I have verified that the user account does have access to the Python, pandas and numpy folders. I would also expect the code to outright fail or error if this was a permission error, but it runs fine, just with incorrect results.

I have also verified that in both cases, the same python.exe is called.

Versions:

Python - 3.5.6
pandas - 0.23.4
numpy - 1.14.2

Why does pandas give different results when trying to concatenate two columns depending on if the code is run with or without admin privileges? Is there another file/folder the user account requires access to for pandas to perform this correctly?

UPDATE

I managed to resolve this by running the following command, found here, on the main Python folder.

ICACLS "<FolderPath>" /INHERITANCE:e /GRANT:r <UserName>:(F) /T /C

But this still does not explain why pandas does incorrect concatenation. Leaving the question up in hopes of one day getting an answer.