run Bry Boschan program

function dating=brybos(X,F)

% function dating=brybos(X,F)

% This function applies the Bry-Boschan (1971) algorithm and determines the peaks and troughs of data

% matrix X with T time series observations and N time series. The output is a (TxN) matrix where 1 signifies a

% peak and -1 a trough. F is the frequency of the observations where monthly observations are the default.

% F=0: monthly, F=1: quarterly, F=2: annual.

% For monthly data, the minimum peak-to-trough (trough-to-peak) period is 5 months and peak-to-peak

% (trough-to-trough) is 15 months. For quarterly p-to-t is 2 quarters and p-to-p is 6 quarters. For annual

% data, p-to-t is 1 year and p-to-p is 2 years.

% The program calls on the M-functions alternate.m, check.m, dates.m, enforce.m, ma.m, mcd.m, outier.m,

% qcd.m, refine.m, spencer.m

% This program was writen by Robert Inklaar, University of Groningen (May 2003) and is an adaption of the

% programs for Gauss of Mark Watson.

% The algorithm is based on Bry and Boschan (1971), 'Cyclical analysis of time series: Selected procedures

% and computer programs', NBER: New York

if nargin == 1; F=0; end

if F == 0; D=5;

elseif F == 1; D=2;

elseif F == 2; D=1;

end

N=size(X,2);

if sum(sum(isnan(X)))>0; error('Data matrix contains empty values'); end

% I - Find outliers and replace them with the Spencer curve value X=outlier(X); Moved down this step and

% only use it in step II

% II - Peaks and troughs of one-year centered moving average (enforcing alternating peaks and troughs)

if F == 0; M=12; elseif F == 1; M=4; elseif F == 2; M=1; end

Xf=ma(outlier(X),M);

[peaks,troughs] = dates(Xf,D);

[peaks,troughs] = alternate(Xf,peaks,troughs);

if F == 0;

% III - Refine peaks and troughs with Spencer curve. Also enforce alternating peaks and troughs and a

% minimum p-to-p (t-to-t) period.

Xs=spencer(X);

[peaks,troughs] = check(peaks,troughs,D);

[peaks,troughs] = refine(Xs,peaks,troughs,D);

[peaks,troughs] = alternate(Xs,peaks,troughs);

[peaks,troughs] = enforce(Xs,peaks,troughs,D);

% IV - Refine peaks and troughs with moving average determined by the number of months/quarters of

% cyclical dominance (MCD). For annual data, the cyclical dominance is set to 1 year. Also enforce

% alternating peaks and troughs.

if F == 0; cdnum=mcd(X);

elseif F == 1; cdnum=qcd(X);

else cdnum=ones(1,N);

end

for i=1:N; Xf2(:,i)=ma(X(:,i),cdnum(i)); end

[peaks,troughs] = check(peaks,troughs,D);

[peaks,troughs] = refine(Xf2,peaks,troughs,D);

[peaks,troughs] = alternate(Xf2,peaks,troughs);

% V - Refine peaks and troughs with actual series. Also enforce

% alternating peaks and troughs and a minimum p-to-p (t-to-t)period.

span=max(4,cdnum);

for i=1:N

[peaks(:,i),troughs(:,i)] = check(peaks(:,i),troughs(:,i),span(i));

[peaks(:,i),troughs(:,i)] = refine(X(:,i),peaks(:,i),troughs(:,i),span(i));

[peaks(:,i),troughs(:,i)] = alternate(X(:,i),peaks(:,i),troughs(:,i));

[peaks(:,i),troughs(:,i)] = enforce(X(:,i),peaks(:,i),troughs(:,i),span(i));

[peaks(:,i),troughs(:,i)] = alternate(X(:,i),peaks(:,i),troughs(:,i));

end

dating=peaks-troughs;

else

[peaks,troughs] = refine(X,peaks,troughs,D);

[peaks,troughs] = check(peaks,troughs,D);

[peaks,troughs] = enforce(X,peaks,troughs,D);

[peaks,troughs] = alternate(X,peaks,troughs);

[peaks,troughs] = enforce(X,peaks,troughs,D);

dating=peaks-troughs;

end

.

.

.

the following message appears when I run the Bryboschan program:

>> brybos (X,0)

Undefined function or variable 'outlier'.

Error in brybos (line 27)

Xf=ma(outlier(X),M);

[Jishuang Yu]

hi ramzi, I am also finding the bry-boschan program code. Could you tell me where you found the program code?Thanks a lot.

Answer 1

[crane lee]

Adding outliers.m to your file will solve this error.