Readme file for replication of results in Dorval and Smith,			
"Interwar Inflation, Unexpected Inflation, and Output Growth"


This folder contains 5 files:

1. quarterly_data.txt
2. panel.txt.
3. stata.zip
4. matlab.zip
5. readme.txt

1. quarterly_data.txt contains quarterly data on inflation and price levels:

country_r: quarterly retail prices data
country_inf_r: quarterly retail prices inflation data

country_w: quarterly wholesale prices data
country_inf_w: quarterly wholesale prices inflation data


2. panel.txt contains annual data: output growth (from Maddison), unexpected inflation and inflation (full sample estimations and recursive estimations):

Definition of variables:

y: output growth for each countries 
inf_r: actual retail prices inflation for each countries
inf_w: actual wholesale prices inflation for each countries
inf_full_r: retail prices inflation for each countries using full sample forecasts  (unexpected + expected = inflation)
inf_recur_r: retail prices inflation for each countries using recursive forecasts  (unexpected + expected = inflation)
inf_full_w: wholesale prices inflation for each countries using full sample forecasts  (unexpected + expected = inflation)
inf_recur_w: wholesale prices inflation for each countries using recursive forecasts  (unexpected + expected = inflation)
unexp_full_r: wholesale prices expected inflation using full sample forecasts
unexp_recur_r: retail prices unexpected inflation using recursive forecasts
unexp_full_w: retail prices expected inflation using full sample forecasts
unexp_recur_w: wholesale prices unexpected inflation using recursive forecasts
country: country name
_Icountry_2 - _Icountry_27: country dummy variables

3. stata.zip contains:

3.1 the .do files to replicate the results presented in Tables 2-5 

3.2 the .do files Full_sample_forecasts.do and Recursive_forecasts.do. These collapse the quarterly forecast and associated inflation rate
into annual data for full sample estimations and recursive estimations.
Note that the first actual inflation value for recursive or full sample might be different from the inflation series. This happens
because "the annual averaging always involves the same number of actual inflation rates and forecasted ones, 
and the annual averages of expected and unexpected inflation always add up to actual inflation for each year."

3.3 quarterly_data.txt and panel.txt but in .dta format

4. matlab.zip contains the files: 

4.1 murphy_topel.m contains the matlab code to generate the Murphy-Topel standard errors for the full-sample forecasts in Table 4 (wholesale and retail).
The code shows how the Topel-Murphy standard errors are similar to homoscedastic standard errors. This results can be generalized to
heteroscedastic standard errors.  

4.2 retail_murphy_topel.mat 
and
4.3 wholesale_murphy_topel.mat
contain the necessary matrices to generate the Murphy-Topel standard errors and are defined in murphy_topel.m


5. readme.txt