Comparison of seasonal random walk,
seasonal random trend, and other ARIMA
models on deflated auto sales data

Model Comparison
----------------
Data variable: AUTOSALE/CPI
Number of observations = 263
Start index = 1/70            
Sampling interval = 1.0 month(s)
Length of seasonality = 12
Number of periods withheld for validation: 24

Models
------
(A) ARIMA(0,0,0)x(0,1,0)12 with constant     <--Seasonal random walk
(B) ARIMA(1,0,0)x(0,1,0)12 with constant     <--  ...with AR=1
(C) ARIMA(0,1,0)x(0,1,0)12                   <--Seasonal random trend
(D) ARIMA(0,1,1)x(0,1,1)12                   <--  ...with MA=1 and SMA=1

Estimation Period
Model  MSE          MAE          MAPE         ME           MPE
------------------------------------------------------------------------
(A)    8.85824      2.28831      9.41851      0.0188148    -0.706667    
(B)    4.20197      1.45045      5.70019      0.00586504   -0.265589    
(C)    4.84114      1.5869       6.19928      -0.00721427  -0.1331      
(D)    2.81782      1.25272      4.92789      0.0120522    -0.21509     

Model  RMSE         RUNS  RUNM  AUTO  MEAN  VAR
-----------------------------------------------
(A)    2.97628       ***   ***   ***   OK   OK   
(B)    2.04987       OK    *     ***   OK   ***  
(C)    2.20026       OK    *     ***   OK   ***  
(D)    1.67864       OK    OK    OK    OK   ***  

Validation Period
Model  MSE          MAE          MAPE         ME           MPE
------------------------------------------------------------------------
(A)    11.2222      3.01909      10.2064      -2.93276     -9.9272      
(B)    4.09013      1.67456      5.70802      -0.800644    -2.77393     
(C)    4.47723      1.60335      5.48813      -0.0117734   -0.127019    
(D)    1.85595      1.01423      3.47177      -0.354179    -1.36136     

Key:
RMSE = Root Mean Squared Error
RUNS = Test for excessive runs up and down
RUNM = Test for excessive runs above and below median
AUTO = Box-Pierce test for excessive autocorrelation
MEAN = Test for difference in mean 1st half to 2nd half
VAR = Test for difference in variance 1st half to 2nd half
OK = not significant (p >= 0.10)
* = marginally significant (0.05 < p <= 0.10)
** = significant (0.01 < p <= 0.05)
*** = highly significant (p <= 0.01)