Sunteți pe pagina 1din 13

Path Analysis

Human Capital vs Homeownership


Gaetan Guy Lion
April 2009

Introduction
Path Analysis is a way to decompose
correlations between different variables, in
this case Human Capital vs Homeownership
rate.
Human Capital is % of population over 25
with a college degree.

The Rosetta Stone in Path Analysis

With standardized variables within a single relationship the


Correlation is equal to the Slope.

Correlation = COVAR (X, Y)/(s X)(s Y)


Slope = COVAR (X, Y)/VAR X
If both s = 1. Correlation = Slope

The Path Analysis Diagram


Independent
variable

Intermediary
variables
Housing
affordability

Human
Capital

Dependent
variable

+
-

Demographic
(youth)

Home
ownership

Unemployment

The Path Analysis Diagram defines our hypothesis. Human Capital has an impact on:
Home Affordability (-) as highly educated wage earners bid up prices of homes,
Demographic/youth (-) more youth fewer older people with degrees, and
Unemployment (-) as Human Capital lowers unemployment.

In turn, those intermediary variables impact Homeownership rate:


Housing Affordability (+), if homes are more affordable homeownership goes up.
Demographic-Youth (% of population between 20 and 29), (-) as younger people can
ill afford homes, and
4
Unemployment (-) as unemployed lack the income to buy homes.

The Actual Correlations


Correlations

Human
Capital

-0.176
-0.181

Housing
affordability

0.573

-0.064

Youth

-0.249

-0.250

Unemploy.

0.064

Home
ownership

We embedded the correlations within the diagram. We also added a correlation


directly from Human Capital to Home ownership. Most correlation signs support
the hypothesis except Unemployment.

The Path Coefficients


Path Coefficients
-0.130

Human
Capital

-0.181

Housing
affordability

0.581

-0.064

Youth

-0.228

-0.250

Unemploy.

-0.181

Home
ownership

Given that the variables are standardized, all bivariate correlations already
represent Path coefficients (in white). Well calculate the Path coefficients in
yellow with a regression model.
SUMMARY OUTPUT
Regression Statistics
Multiple R
0.633
R Square
0.401
Adjusted R Square
0.375
Standard Error
0.785
Observations
111

Dependent variable is Homeownership rate


Human capital
Housing affordability
Young population
Unemployment

Coefficients Stand. Error


-0.130
0.078
0.581
0.079
-0.228
0.077
-0.181
0.081

t Stat
-1.665
7.334
-2.976
-2.227

P-value
0.099
0.000
0.004
0.028

Correlations vs Path Coefficients


Correlations

Human
Capital

-0.176
-0.181

Housing
affordability

0.573

-0.064

Youth

-0.249

-0.250

Unemploy.

0.064

Home
ownership

Path Coefficients
-0.130

Human
Capital

-0.181

Housing
affordability

0.581

-0.064

Youth

-0.228

-0.250

Unemploy.

-0.181

Home
ownership

Correlations reflect the relationship between just two variables. The Path
coefficients reflect the effect one variable has on another when controlled
for the other three variables. Now the Path coefficient of Unemployment
rate is negative.

Direct and Indirect Effects


Correlation

Direct Effect

Indirect Effect

Causal Effect

The Correlation of the independent variable can be decomposed into its


Direct Effect and Indirect Effect on the dependent variable. The Causal
Effect is the sum of the mentioned Effects and should equal the
Correlation.
8

Human Capital
Direct and Indirect Effects
Path Coefficients
-0.130

Human
Capital

-0.181

Housing
affordability

0.581

-0.064

Youth

-0.228

-0.250

Unemploy.

-0.181

Decomposing correlations into indirect and direct effects


Human Capital indirect effect on Homeownership
Human Cap. -> Housing affordability -> Homeownership
Human Cap. -> Youth -> Homeownership
Human Cap. -> Unemployment -> Homeownership

Human Capital causal effect


(-0.176) on Homeownership
equals its correlation.

Home
ownership

A
-0.181
-0.064
-0.250

B
0.581
-0.228
-0.181

AxB
-0.105
0.015
0.045
-0.045

Human Capital direct effect on Homeownership

-0.130

Human Capital Causal effect on Homeownership


a) Indirect effect
b) Direct effect
Total causal effect

-0.045
-0.130
-0.176

Assessing the Model Fit


Assessing the model fit entails:
1. Reconstructing all the correlations by using the
path coefficients; and
2. Assessing the closeness of the fit between the
reconstructed correlations and actual ones.

10

Reconstructing correlations
Path Coefficients
-0.130

Human
Capital

-0.181

Housing
affordability

0.581

-0.064

Youth

-0.228

-0.250
Unemploy.
Reconstructing the other correlations vs Homeownership

Home
ownership

-0.181

Housing affordability
-- Homeownership
-- Human capital -- Homeownership
-- Human capital -- Youth -- Homeownership
-- Human capital -- Unemploy.-- Homeownership

A
0.581
-0.181
-0.181
-0.181

Youth
-- Homeownership
-- Human capital -- Homeownership
-- Human capital -- Housing afford. -- Homeownership
-- Human capital -- Unemployment -- Homeownership

-0.228
-0.064
-0.064
-0.064

Unemployment
-- Homeownership
-- Human capital -- Homeownership
-- Human capital -- Housing afford. -- Homeownership
-- Human capital -- Youth -- Homeownership

-0.181
-0.250
-0.250
-0.250

-0.130
-0.064
-0.250

-0.228
-0.181

-0.130
-0.181
-0.250

-0.130
-0.181
-0.064

0.581
-0.181

0.581
-0.228

AxBxC
0.581
0.024
-0.003
-0.008
0.593
-0.228
0.008
0.007
-0.003
-0.216
-0.181
0.033
0.026
-0.004
-0.126

11

Assessing closeness of fit with RMSE

Root Mean Square Error (RMSE)


Standard Error of Prediction
Correlation vs Homeownership
Human capital
Housing affordability
Youth
Unemployment rate

Actual R
-0.176
0.573
-0.249
0.064

Predicted R
-0.176
0.593
-0.216
-0.126

Error
0.000
-0.020
-0.033
0.190
RMSE

Error2
0.0000
0.0004
0.0011
0.0361
0.0969

12

Path Analysis next step


We test other models and check if their fit is
superior to the original model.

13

S-ar putea să vă placă și