Strange pacf graph in statsmodels

[dm64...@gmail.com]

Hope someone can help here....running the following code produces a rather weird pacf plot:

df = pd.read_stata('D://junk//var24.dta')
index=pd.date_range('2000-01-01' , '2010-12-31', freq='M')
df.index=index


fig = plt.figure(figsize=(12,8))
ax1 = fig.add_subplot(211)
fig = sm.graphics.tsa.plot_acf(df['reading'].values.squeeze(), lags=25, ax=ax1)
ax2 = fig.add_subplot(212)
fig = sm.graphics.tsa.plot_pacf(df['reading'],lags=25, ax=ax2)

The acf is fine but the pacf is definitely not !!   Data attached

many thanks

Dan