Seminar 2

Construction of Linear Regression Model (LRM)

1. Theoretical background
The traditional econometric approach might be divided into following steps that we have
to deal with:
(i) Economic theory
(ii) Algebraic Economic model
(iii) Econometric Model
(iv) Economic data
(v) Estimation
(vi) Economic interpretation
(vii) Statistical verification
(viii) Application – policy analysis (ex-post analysis), simulation and
forecasting

Econometric Model
Econometric model may contain two types of equations:
o Stochastic and
o Identity equations

The equations contain following variables:

o Endogenous variables – are generated within the model.
o Exogenous variables – are generated outside the model.
o Predetermined variables – are used to explain endogenous variables. Predetermined
variables are: exogenous variables, lagged exogenous variables and endogenous
variables.
o Error (Disturbance) terms.
2. Exercises

Data table
C PM
CP PM CP BM CP ChM Income
Year (kg/person/
(CZK/kg) (CZK/kg) (CZK/kg) (CZK) *
year)
Variable
1995 8,04 84.2 94,81 52,32 55 578,0
1996 8,87 90,42 102,12 62,77 64 114,0
1997 8,74 92,11 104,82 70,64 70 968,0
1998 10,36 86,39 110,16 73,31 77 942,0
1999 9,78 80,47 107,80 56,51 80 771,0
2000 8,94 90,04 111,53 61,83 83 422,0
2001 9,05 101,66 112,56 71,28 90 167,0
2002 9,55 89,84 112,99 62,40 93 153,0
2003 10,14 82,74 108,02 60,67 98 102,0
2004 9,97 85,36 112,84 62,55 102 217,0
2005 11,18 85,30 117,73 62,73 116 573,5
Average 9,51 88,43 108,67 63,36 84 818,9
* Net income per household

Correlation matrix
Variable C PM CP PM CP BM CP ChM In
C PM 1
CP PM -0,369659 1
CP BM 0,751629 0,165662 1
CP ChM 0,209048 0,626810 0,416617 1
In 0,814929 -0,069005 0,892922 0,144044 1

Economic model
Pork meat consumption depends on pork meat consumer price, beef meat consumer price,
chicken meat consumer price and income.

Assumptions (based on the economic theory):

• pork meat and chicken meat are substitutes;
• pork meat and beef meat are substitutes;
• income increase causes increase of pork meat consumption.

Tasks

1. Declare all variables in the data table (i.e. use symbols y and x with their relevant
indexes).

2. Write values of the following variables:

a. y1,6 =
b. x2,10 =
c. x4,1 =

3. Write an algebraic form of the model defined above. Assume linear relationship
among endogenous and exogenous variables.
4. Write the model based on the power function.

5. Modify the economic model (linear form of the model) to econometric model.

6. Explain term u1t.

7. Write the procedure of u1,5 calculation.

8. Check the multicollinearity occurrence and show possibilities of its elimination.

9. Data transformation:
a. Include intercept term (constant) into model. How may it be included in data
table?

b. What is the problem of extreme values within the data set? What is the
possible process of their detection? Propose any solution of the problem of
extreme values occurrence.

c. Which technique may be used to solve the problem of missing values?

d. Modify data into form suitable for parameters’ estimation in Microsoft Excel.

10. Define dynamic model (results should be written into the following table):
a. Based on the time vector.

b. Based on the lagged value (one period) of endogenous variable. What is the
impact on the data table? Define vector of one period lagged values of
variable y1t, i.e. values of vector y1(t-1).
 c. Based on the variables in form of 1st differences. Rewrite variables y1t, x1t a x2t
in form of 1st differences.

Table of other variables

UV y1t x1t x2t
Year Time vector y1(t-1)
(intercept) (differences) (differences) (differences)

Variable
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Average

Modified data table

C PM
CP PM CP BM CP ChM Income
Year (kg/person/
(CZK/kg) (CZK/kg) (CZK/kg) (CZK) *
year)
Variable
1996 8,87 90,42 102,12 62,77 8,536
1997 8,74 92,11 104,82 70,64 6,854
1998 10,36 86,39 110,16 73,31 6,974
1999 9,78 80,47 107,80 56,51 2,829
2000 8,94 90,04 111,53 61,83 2,651
2001 9,05 101,66 112,56 71,28 6,745
2002 9,55 89,84 112,99 62,40 2,986
2003 10,14 82,74 108,02 60,67 4,949
2004 9,97 85,36 112,84 62,55 4,115
2005 11,18 85,30 117,73 62,73 14,357
Average 9,66 88,43 110,06 64,47 6,100
* Net income per household

Modified correlation matrix

Variable C PM CP PM CP BM CP ChM In
C PM 1
CP PM -0,602642 1
CP BM 0,614154 0,005568 1
CP ChM -0,163091 0,628116 -0,017416 1
In 0,456484 0,091175 0,220278 0,279075 1
Individual exercises

1. Based on the data table included in supplement 1 define one-equation linear model,
which describes that households’ expenditures depend on households’ expenditures in
previous period, inflation, interest rate and income. Define assumptions about the
relations among the variables (based on the economic theory).

2. Declare all variables in the data table (i.e. use symbols y and x with their relevant
indexes).

3. Write values of the following variables:

a. y1,4 =
b. x2,8 =
c. x3,2 =

4. Write an algebraic form of the model defined above. Assume linear relationship
among endogenous and exogenous variables.

5. Write the model based on the power function.

6. Modify the economic model (linear form of the model) to econometric model.

7. Write the procedure of u1,3 calculation.

8. Check the multicollinearity occurrence and show possibilities of its elimination.

9. Data transformation:
a. Include intercept term (constant) into model.

b. Modify data into form suitable for parameters’ estimation in Microsoft Excel.

10. Check whether the model is static or dynamic. If the model is static, propose the
possible ways to transform it into the dynamic form.