EXAMPLES OF RATS AND TSP

Basic Sample RATS Programs

1. Defining Sample Statement
2. Reading Data Set
3. Transforming Data Set
4. Basic Program
5. Advanced Program

Basic Sample TSP Programs

1. Defining Sample Statement
2. Reading Data Set
3. Transforming Data Set
4. Basic Program
5. Advanced Program

Copyright @1997 Irene Chan
Last Modified: January 27, 1997
If you have any comments, please feel free to email me chani@qed.econ.queensu.ca

Defining Sample Statements

• RATS supports every time series frequency, including annual, semi-annual, monthly, quarterly, weekly, and daily, as well as panel data (pooled cross section time series data) and daily data with multiple observations per day.

• CALENDAR sets the dating scheme and specifies the starting date for the data. ALLOCATE sets the default length of the data series (the CALENDAR instruction is omitted when using cross-sectional data).

1. Reading annual data set starting from 1922 to 1941. CALENDAR 1922 ALLOCATE 1941:1

2. Monthly data, from February, 1965 through December, 1995: CALENDAR 1965 2 12 ALLOCATE 1995:12
   In the Calendar statement, 2 means February (starting month/period), 12 means periods per year. It will be 4 for Quarterly data.

   Daily data (Mondays through Fridays only), from January 2nd, 1995 through December 29th, 1995: CALENDAR(DAILY) 1995 1 2 ALLOCATE 1995:12:29

   As above, but seven days per week: CALENDAR(SEVENDAY) 1995 1 2 ALLOCATE 1995:12:29

   Data with 24 periods per year, starting in 1950: CALENDAR 1950 1 24 ALL 1995:24

Reading Data Set

• For the DOS user, you can import ascii/spreadsheet/RATS data file easily by using RATSDATA utility.

1. The FOODDATA.RAT is a sample RATS time series format data file in which each column is the series's whole data set. FORMAT=RATS denotes the data file is RATS format and org=obs denotes the data format is time series, i.e.data are organized horizontally by obserations. For cross-sectional data fromat, use org=vars, i.e. data are organized horizontally by variables. UNIX users can find this sample data file in /usr/local/rats410/examples CALENDAR 1922 ALLOCATE 1941:1 OPEN DATA C:\RATS\FOODDATA.RAT DATA(FORMAT=RATS,ORG=OBS) / FOODCONS PRRETAIL DISPINC It is very important to have space between the / and each series as RATS is very sensitive to spaces. There is no space between DATA command and its options listed in the parenthesis.

2. Read an ascii time series data format file without labels at the top of each series in each column . OPEN DATA FOODDATA.DAT DATA(FORMAT=FREE,ORG=OBS) / FOODCONS PRRETAIL DISPINC
3. use $ for the continuation of the commands to the next line DATA(FORMAT=FREE,ORG=OBS) / FOODCONS $ PRRETAIL DISPINC
   

Transforming Data Set

• In RATS, set is the command in managing the time series. whereas compute is the comamnd in managing all kind of data incluind integer, real number, matrix.

• RATS has two different ways in making comments:
  Insert a * before a single line comment, e.g. * This is my only line of the comment

  use /* and */ for multiple lines comment, e.g. /* This is the beginning of the comment This is the second line of the comment This is the end of the comment */

1. To create a linear time trend series: SET TREND = T
   
2. To create a MA(2) average series, {1} indicates a one period lag, use the negative number for a one period lead, {-1} : SET AVGPRICES = (PRFARM+PRFARM{1})/2.0
   
3. Taking First-difference of Y: Two different ways of taking the first-difference of Y: SET DY = Y - Y{1} DIFF Y / DY
   
4. Two ways of taking the natural log of X: SET LX = LOG(X) LOG X / LX
   
5. Basic Functions: set x = y**2 /* Square of Y */ /* inverse and transpose of Y iff x and y are the same type of matrix, e.g. rectangular, symmetric, etc. */ comp x = inv(y*tr(y)) set x = sqrt(y) /* Square root of Y */

Basic Program

Advanced Program

Defining Sample Statements

• Unlike RATS, TSP doesn't support every time series frequency. It supports annual, quarterly, monthly, weekly and panel data. Nonetheless you can use convert option to manipulate the frequency yourself.

• In TSP, you need both FREQ and SMPL statements to define the sample range. Unlike RATS, TSP requires semicolon follows your last command

1. Reading annual data set starting from 1922 to 1941. FREQ A; SMPL 1922, 1941;

2. Monthly data, from February, 1965 through December, 1995: FREQ M; SMPL 1965:2, 1995:12;

3. Weekly data, from 15th week of 1965 through 52th week of 1995: FREQ W; SMPL 1965:15, 1995:52;

Reading Data Set

• TSP can read an ascii or spreadsheet (Excel or Lotus) data format beside its own TSP database.

1. The DATA.TXT is an ascii file which contains data organized horizontally by observation (the usual Time Series data format). DATA.TLB is an TSP's own binary data format file. FREQ A; SMPL 61,75; ? or you can write smpl 1961 1975; READ(FILE='DATA.TXT',BYOBS) GNP CONS I;
2. There is an alternative way to read the data if you don't like to have data stored in a separate file: DATA.TSP

3. You can store the series in TSP's data base. FREQ A; SMPL 61,75; READ(FILE='DATA.TXT',BYOBS) GNP CONS I; genr logc = log(cons); out cons; ? a new TSP data file is being created and names as "cons.tlb" keep logc; ? I only want to store the new series in this data base

4. To read this TSP data fille FREQ A; SMPL 61,75; in cons; ? loading the data base into this tsp session print cons;

Transforming Data Set

• In TSP, set is the command in managing the scalar or matrix whereas genr is the comamnd in managing a dynamic time series (e.g. AR(1) process).

• TSP has only one way in making comments:
  Insert a ? before a single line comment, e.g. ? This is my only line of the comment

1. To create a linear time trend series: TREND AA; ? AA is the trend series
   
2. To create a MA(2) average series, (-1) indicates a one period lag, use the positive number for a one period lead, (+1) : GENR AVGPRICES = (PRFARM+PRFARM(-1))/2.0 ;
   
3. Taking First-difference of Y: GENR DY = Y - Y(-1);
   
4. Taking the natural log of X: GENR LX = LOG(X);
   
5. Basic Functions: genr x = y**2; ? Square of Y mat x = y'; ? transpose of y matrix mat x = y"; ? inverse of y matrix genr x = y**(1/2) ; ? Square root of Y time series

Basic Program