

This is a short documentation how to estimate and forecast the federal funds rate target
as it was done in Grammig and Kehrle (JEDC, 2008).

4 Folders:
DATA:
ESTIMATION:
FORECAST:
RESULTS:

------------------------------------------------------------------------------
DATA
------------------------------------------------------------------------------
READDATB4.PRG

descriptions: reads in data and produces event time variables
uses: newmat3.fmt
(= data)

columns in newmat3:
1: "TARGET": federal funds rate target
2: "CHANGE": federal funds rate target changes
3: "FOMC": FOMC meeting dummy = 1 for meeting, = 0 no meeting in week t
4: "TB6_WAV": six month treasury bill
5: "FF_WAV": effective federal funds rate target
6: "SP6MFF": spread variable not used in analysis
7: "spread": spread variable used in analysis
8: "event": event dummy  = 1 target change, = 0 no target change in week t
9: "chang": change variable identical to column 2
10: "kdate": date variable
11: "monthdum": dummy indicating the first week of the month
12: "duration": duration between target changes

------------------------------------------------------------------------------
ESTIMATION Estimation results for the  ACH, OP, ACM results
------------------------------------------------------------------------------
ACH RESULTS:
let run MAXSEEK.PRG twice
once with sample = 1; /* first  period from 1984-1989 */
once with sample = 2; /* second period from 1989-2001 */
NOTE: The program requires optmum source of GAUSS
uses: READDATB4.PRG
      MAXSEEK.SRC
produces: maxseek_par1.prg
          maxseek_par2.prg

OP RESULTS:
let run PROBSEEK.PRG 
uses: READDATB4.PRG
      PROBSEEK.SRC
produces: probseek_par5.prg


ACM RESULTS:
you need first a path definition: e.g. RESULTS_MYACM folder
NOTE: The program requires CML source of GAUSS
let run MYACM_BIG4.PRG
uses: MYACM_PROC.SRC
      READDATB4.PRG
produces: A.FMT, B.FMT, CT.FMT, D1.FMT, D2.FMT, & and for all SE = standard errors LOGLIK.FMT 
models in the paper
    mod9 = PSACM 
    mod2 = DACM
declare string models =   "mod2" "mod9";

------------------------------------------------------------------------------
FORECAST: Forecast results
------------------------------------------------------------------------------
descriptions:  produces the forecast simulation
/* IMPORTANT: this program needs the results of the estimation programs*/

let run TWICE: once with icontrol = 1 and icontrol = 2 
- myacm_forecon4.prg (uses: myacm_forepro4.prg)

requires:
#include readdatb4.prg;   
#include varproc2.prg;
#include myacm_forepro4.src;
#include probseek_par5.prg;

produces with icontrol = 1:
z_mat (needed for density forecast evaluation)
includes the continued z-series
        uniformity tests
	autocorrelations
	histograms
	autocorrelograms 
tar_i1
target forecast based on the information of the previous week

produces with icontrol = 2:
tar_i2 (needed for MSE comparisons)
target forecast based on the information of the previous months last week

------------------------------------------------------------------------------
FORECAST: Forecast evaluation 
------------------------------------------------------------------------------
- mymonth9.prg (uses monthch9.prg) 
/* IMPORTANT: this program needs the results of the run myacm_forecon4.prg.*/

descriptions: produces MSE, uniformity tests, autocorrelation tests,
autocorrelograms, histograms, all for each model

load  mvardat.csv; ffrmonthlycsv.csv;

various horizon possible / various models possible
model averaging models: it is not possible to produce histograms
and autocorrs...(i.e.
DENSITY EVALUATION   = no)

MSE EVALUATION       = yes/no
DENSITY EVALUATION   = yes/no
SAVE RESULTS ?       = yes/no /* results are saved to hard disk by fmt files*/

PATH SETTINGS
load path: lp     /* lp is loading the tarbase */
save path: svp_st /* starting path.. be careful, a folder structure must be on your computer*/

EXAMPLE FOR PATH SETTINGS
fide = "/";
lp    =fide$+"home"$+fide$+"kehrleke"$+fide$+"share"$+fide$+"gk2005"$+fide$+"results_five"$+fide; 
svp_st=fide$+"home"$+fide$+"kehrleke"$+fide$+"share"$+fide$+"gk2005"$+fide$+"results_five"$+fide;

MODEL SETTINGS

sample_index = {1 2 3 4}; /*   1: insample
			       2: first outsample 5/01-5/04
			       3: second outsample 6/04 - 1/06
			       4: outsample total 5/01-1/06 */

FOR THE  HISTOGRAMS 
number_of_bins = 20; /* only possible 10 or 20 otherwise the y-axis of the hist does not fit */
conf_interval = 0.95;

AUTOCORRELATION PLOTS
lagorder = 6;

UNIVORMITY TESTS
"Important choice is not allowed to be bigger than horizon !!!!!";
choice =  {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,20,24,26,28,30,40,50,52,60};
/* this are the chosen horizons to be written out in the evaluation */
there is an new uniformity test: lambda.
in the paper we report the old uniformity test!!!

FORMAT SETTING 
rounding = 9,10;
format /rd rounding; 
--------------------------------------------
RESULTS:
--------------------------------------------
- myacm_results2xls_homepage.prg
takes results of mymonth9.prg and the estimation results
and writes them into suitable table formats 

allsample = {1 2 3 4} /*  1: insample
			  2: first outsample 5/01-5/04
			  3: second outsample 6/04 - 1/06
			  4: outsample total 5/01-1/06 */
