8/4/2019 Ch11D3
1/24
Why do some people go to college while others do not?
Why do some women enter the labor force while others
do not?
Why do some people buy houses while others rent?
Why do some people migrate while others stay?
1
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
Financial Economists are often interested in the factors behind the decision-making ofindividuals or enterprises, examples being shown above.
8/4/2019 Ch11D3
2/24
Why do some people go to college while others do not?
Why do some women enter the labor force while others
do not?
Why do some people buy houses while others rent?
Why do some people migrate while others stay?
2
The models that have been developed for this purpose are known as qualitative response orbinary choice models, with the outcome, which we will denote Y, being assigned a value of
1 if the event occurs and 0 otherwise.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
3/24
Why do some people go to college while others do not?
Why do some women enter the labor force while others
do not?
Why do some people buy houses while others rent?
Why do some people migrate while others stay?
3
Models with more than two possible outcomes have also been developed, but we willconfine our attention to binary choice models.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
4/24
4
The simplest binary choice model is the linear probability model where, as the nameimplies, the probability of the event occurring, p, is assumed to be a linear function of a set
of explanatory variables.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1(
8/4/2019 Ch11D3
5/24
5
XXi
1
0
1 +2Xi
y, p
Graphically, the relationship is as shown, if there is just one explanatory variable.
iii XYpp 21)1(
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
1
8/4/2019 Ch11D3
6/24
6
Of course pis unobservable. One has data on only the outcome, Y. In the linear probabilitymodel this is used like a dummy variable for the dependent variable.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1(
8/4/2019 Ch11D3
7/24
Why do some people graduate from high school while
others drop out?
7
As an illustration, we will take the question shown above. We will define a variableGRADwhich is equal to 1 if the individual graduated from high school, and 0 otherwise.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
8/24
. g GRAD=0
. replace GRAD=1 if S>11
(523 real changes made)
. reg GRAD ASVABC
Source | SS df MS Number of obs = 570
---------+------------------------------ F( 1, 568) = 112.59
Model | 7.13422753 1 7.13422753 Prob > F = 0.0000
Residual | 35.9903339 568 .063363264 R-squared = 0.1654
---------+------------------------------ Adj R-squared = 0.1640Total | 43.1245614 569 .07579009 Root MSE = .25172
------------------------------------------------------------------------------
GRAD | Coef. Std. Err. t P>|t| [95% Conf. Interval]
---------+--------------------------------------------------------------------
ASVABC | .0121518 .0011452 10.611 0.000 .0099024 .0144012
_cons | .3081194 .0583932 5.277 0.000 .1934264 .4228124
------------------------------------------------------------------------------
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8
The Stata output above shows the construction of the variable GRAD. It is first set to 0 forall respondents, and then changed to 1 for those who had more than 11 years of schooling.
8/4/2019 Ch11D3
9/24
. g GRAD=0
. replace GRAD=1 if S>11
(523 real changes made)
. reg GRAD ASVABC
Source | SS df MS Number of obs = 570
---------+------------------------------ F( 1, 568) = 112.59
Model | 7.13422753 1 7.13422753 Prob > F = 0.0000
Residual | 35.9903339 568 .063363264 R-squared = 0.1654
---------+------------------------------ Adj R-squared = 0.1640Total | 43.1245614 569 .07579009 Root MSE = .25172
------------------------------------------------------------------------------
GRAD | Coef. Std. Err. t P>|t| [95% Conf. Interval]
---------+--------------------------------------------------------------------
ASVABC | .0121518 .0011452 10.611 0.000 .0099024 .0144012
_cons | .3081194 .0583932 5.277 0.000 .1934264 .4228124
------------------------------------------------------------------------------
9
Here is the result of regressing GRADon ASVABC. It suggests that every additional pointon the ASVABCscore increases the probability of graduating by 0.012, that is, 1.2%.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
10/24
. g GRAD=0
. replace GRAD=1 if S>11
(523 real changes made)
. reg GRAD ASVABC
Source | SS df MS Number of obs = 570
---------+------------------------------ F( 1, 568) = 112.59
Model | 7.13422753 1 7.13422753 Prob > F = 0.0000
Residual | 35.9903339 568 .063363264 R-squared = 0.1654
---------+------------------------------ Adj R-squared = 0.1640Total | 43.1245614 569 .07579009 Root MSE = .25172
------------------------------------------------------------------------------
GRAD | Coef. Std. Err. t P>|t| [95% Conf. Interval]
---------+--------------------------------------------------------------------
ASVABC | .0121518 .0011452 10.611 0.000 .0099024 .0144012
_cons | .3081194 .0583932 5.277 0.000 .1934264 .4228124
------------------------------------------------------------------------------
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
10
The intercept has no sensible meaning. Literally it suggests that a respondent with a 0ASVABCscore has a minus 31% probability of graduating. However a score of 0 is not
possible.
8/4/2019 Ch11D3
11/24
11
Unfortunately, the linear probability model has some serious shortcomings. First, there areproblems with the disturbance term.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1(
8/4/2019 Ch11D3
12/24
12
As usual, the value of the dependent variable Yiin observation ihas a nonstochasticcomponent and a random component. The nonstochastic component depends on Xiand
the parameters. The random component is the disturbance term.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1( iii uYEY )(
8/4/2019 Ch11D3
13/24
13
The nonstochastic component in observation iis its expected value in that observation.This is simple to compute, because it can take only two values. It is 1 with probability piand
0 with probability (1 - pi) The expected value in observation iis therefore 1 + 2Xi.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1( iii uYEY )(
iiiii XpppYE 21)1(01)(
O O O O
8/4/2019 Ch11D3
14/24
14
This means that we can rewrite the model as shown.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1( iii uYEY )(
iiiii XpppYE 21)1(01)(
iii uXY 21
BINARY CHOICE MODELS LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
15/24
XXi
1
0
1 +2Xi
Y, p iii XYpp 21)1( BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
15
The probability function is thus also the nonstochastic component of the relationshipbetween Yand X.
1
BINARY CHOICE MODELS LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
16/24
16
iiiii XpppYE 21)1(01)(
iii uXY 21 iii XuY 2111
iii XuY 210
In observation i, for Yi to be 1, uimust be (1 - 1 - 2Xi). For Yi to be 0, uimust be (- 1 - 2Xi).
iii uYEY )(
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
iii XYpp 21)1(
BINARY CHOICE MODELS LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
17/24
XXi
1
0
1 +2Xi
Y, p iii XYpp 21)1( BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
1
17
The two possible values, which give rise to the observations A and B, are illustrated in thediagram. Since udoes not have a normal distribution, the standard errors and test
statistics are invalid. Its distribution is not even continuous.
A1 - 1 - 2Xi
B
1 + 2Xi
BINARY CHOICE MODELS LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
18/24
XXi
1
0
1 +2Xi
Y, p
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
1
A1 - 1 - 2Xi
B
1 + 2Xi
18
Further, it can be shown that the population variance of the disturbance term in observation
iis given by (1 + 2Xi)(1 - 1 - 2Xi). This changes with Xi, and so the distribution isheteroscedastic.
)1)(( 21212
iiu XXi
BINARY CHOICE MODELS LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
19/24
XXi
1
0
1 +2Xi
Y, p
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
1
A1 - 1 - 2Xi
B
1 + 2Xi
19
Yet another shortcoming of the linear probability model is that it may predict probabilities ofmore than 1, as shown here. It may also predict probabilities less than 0.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
20/24
20
The Stata command for saving the fitted values from a regression is predict, followed by thename that you wish to give to the fitted values. We are calling them PROB.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL. g GRAD=0
. replace GRAD=1 if S>11
(523 real changes made)
. reg GRAD ASVABC
Source | SS df MS Number of obs = 570
---------+------------------------------ F( 1, 568) = 112.59
Model | 7.13422753 1 7.13422753 Prob > F = 0.0000
Residual | 35.9903339 568 .063363264 R-squared = 0.1654
---------+------------------------------ Adj R-squared = 0.1640Total | 43.1245614 569 .07579009 Root MSE = .25172
------------------------------------------------------------------------------
GRAD | Coef. Std. Err. t P>|t| [95% Conf. Interval]
---------+--------------------------------------------------------------------
ASVABC | .0121518 .0011452 10.611 0.000 .0099024 .0144012
_cons | .3081194 .0583932 5.277 0.000 .1934264 .4228124
------------------------------------------------------------------------------
. predict PROB
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
21/24
. tab PROB if PROB>1
PROB | Freq. Percent Cum.
------------+-----------------------------------
1.000773 | 33 18.75 18.75
1.012925 | 16 9.09 27.84
1.025077 | 21 11.93 39.77
1.037229 | 26 14.77 54.55
1.04938 | 35 19.89 74.43
1.061532 | 20 11.36 85.80
1.073684 | 16 9.09 94.89
1.085836 | 3 1.70 96.591.097988 | 6 3.41 100.00
------------+-----------------------------------
Total | 176 100.00
21
tab is the Stata command for tabulating the values of a variable, and for cross-tabulatingtwo or more variables. We see that there are 176 observations where the fitted value is
greater than 1.
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
BINARY CHOICE MODELS: LINEAR PROBABILITY MODEL
8/4/2019 Ch11D3
22/24
. tab PROB if PROB>1
PROB | Freq. Percent Cum.
------------+-----------------------------------
1.000773 | 33 18.75 18.75
1.012925 | 16 9.09 27.84
1.025077 | 21 11.93 39.77
1.037229 | 26 14.77 54.55
1.04938 | 35 19.89 74.43
1.061532 | 20 11.36 85.80
1.073684 | 16 9.09 94.89
1.085836 | 3 1.70 96.591.097988 | 6 3.41 100.00
------------+-----------------------------------
Total | 176 100.00
. tab PROB if PROB
8/4/2019 Ch11D3
23/24
. tab PROB if PROB>1
PROB | Freq. Percent Cum.
------------+-----------------------------------
1.000773 | 33 18.75 18.75
1.012925 | 16 9.09 27.84
1.025077 | 21 11.93 39.77
1.037229 | 26 14.77 54.55
1.04938 | 35 19.89 74.43
1.061532 | 20 11.36 85.80
1.073684 | 16 9.09 94.89
1.085836 | 3 1.70 96.591.097988 | 6 3.41 100.00
------------+-----------------------------------
Total | 176 100.00
. tab PROB if PROB
8/4/2019 Ch11D3
24/24
. tab PROB if PROB>1
PROB | Freq. Percent Cum.
------------+-----------------------------------
1.000773 | 33 18.75 18.75
1.012925 | 16 9.09 27.84
1.025077 | 21 11.93 39.77
1.037229 | 26 14.77 54.55
1.04938 | 35 19.89 74.43
1.061532 | 20 11.36 85.80
1.073684 | 16 9.09 94.89
1.085836 | 3 1.70 96.591.097988 | 6 3.41 100.00
------------+-----------------------------------
Total | 176 100.00
. tab PROB if PROB