Political Party & Unemployment
Group 5:
Richard Baker, Michael Jelmini, Kenneth Morino,
Cambria Price, You Ren, James Young
Econ 240A
Professor Llad Phillips
Objective
Why?• Upcoming graduation
• Recent recession
• Recent increase in unemployment
What?• Is there a correlation between unemployment and presidential party
affiliation?
How?• Time series data of national unemployment rate
• Bernoulli distribution of party affiliation
Hypothesis
H0 : When a republican is in presidential office,
the unemployment rate is higher than when a democrat is in office.
Ha : There is no correlation between the political
affiliation of the president and the unemployment rate.
Collected Data
• Unemployment Data from 1947-2002-Bureau of Labor Statistics
http://stats.bls.gov
• Presidential Party Affiliations-The White House
http://www.whitehouse.gov/history/presidents
Meet the Presidents
Presidential Party Affiliation
Time Series Data of Unemployment Rates
Unemployed Percent of Labor Force vs Year
0
2
4
6
8
10
12
1947 1957 1967 1977 1987 1997
Year
Un
emp
loye
d P
erce
nt
of
Lab
or
Fo
rce
Unemployment vs Presidential Party Affiliation
0
2
4
6
8
10
12
1947
1951
1955
1959
1963
1967
1971
1975
1979
1983
1987
1991
1995
1999
Year
Une
mpl
oym
ent (
%)
Democrat
Republican
Probability of Democrat Given Unemployment Rate
y = -0.1175x + 1.1041
R2 = 0.128
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12
Unemployment Rate (%)
Pro
bab
ilit
y P
resi
den
t is
Dem
ocr
at
Data Points
Linear Model
Linear Probability Model
Regression Results
Dependent Variable: DEMOCRAT Method: Least Squares Date: 11/20/03 Time: 14:34 Sample: 1947 2002 Included observations: 56
Variable Coefficient Std. Error t-Statistic Prob.
UNEMPLOYMENT -0.117549 0.041758 -2.815034 0.0068 C 1.104073 0.242009 4.562109 0.0000
R-squared 0.127969 Mean dependent var 0.446429 Adjusted R-squared 0.111820 S.D. dependent var 0.501621 S.E. of regression 0.472744 Akaike info criterion 1.374535 Sum squared resid 12.06828 Schwarz criterion 1.446869 Log likelihood -36.48697 F-statistic 7.924418 Durbin-Watson stat 0.626808 Prob(F-statistic) 0.006793
Linear Probability Model Analysis
• Problem:– LPM not constrained within 0<P(x)<1
• Solution:– model with Probit!
Model with Probit
Probability of Democrat President Given Unemployment Rate
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12
Unemployment Rate (%)
Pro
ba
bili
ty P
res
ide
nt
is D
em
oc
rat
Data Points
Logit Model
Probit Regression Results
Dependent Variable: DEMOCRAT Method: ML - Binary Probit (Quadratic hill climbing) Date: 11/20/03 Time: 14:35 Sample: 1947 2002 Included observations: 56 Convergence achieved after 4 iterations Covariance matrix computed using second derivatives
Variable Coefficient Std. Error z-Statistic Prob.
UNEMPLOYMENT -0.348600 0.132693 -2.627109 0.0086 C 1.783351 0.741412 2.405344 0.0162
Mean dependent var 0.446429 S.D. dependent var 0.501621 S.E. of regression 0.473574 Akaike info criterion 1.304778 Sum squared resid 12.11073 Schwarz criterion 1.377112 Log likelihood -34.53379 Hannan-Quinn criter. 1.332822 Restr. log likelihood -38.49420 Avg. log likelihood -0.616675 LR statistic (1 df) 7.920804 McFadden R-squared 0.102883 Probability(LR stat) 0.004887
Obs with Dep=0 31 Total obs 56 Obs with Dep=1 25
Further Analysis
• Is there an unemployment rate overlap due to previous presidential party affiliation?
• If so, how much overlap?– 1 year– 2 years
• Will these models produce a better/worse correlation?
0
2
4
6
8
10
12
1947
1951
1955
1959
1963
1967
1971
1975
1979
1983
1987
1991
1995
1999
Year
Un
emp
loym
ent
(%)
Democrat
Republican
0
2
4
6
8
10
12
1947
1951
1955
1959
1963
1967
1971
1975
1979
1983
1987
1991
1995
1999
Year
Un
emp
loym
ent
(%)
Democrat
Republican
One Year Forward Lag
BeforeAfter
Probability of Democrat President Given Unemployment Rate (One Year Lag)
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12
Unemployment Rate (%)
Pro
bab
ilit
y P
resi
den
t is
D
emo
crat
Data Points
Probit Model
Probit Model of One Year Lag
One Year Lag Results
Dependent Variable: LAG1 Method: ML - Binary Probit Date: 11/20/02 Time: 21:14 Sample: 1947 2002 Included observations: 56 Convergence achieved after 4 iterations Covariance matrix computed using second derivatives
Variable Coefficient Std. Error z-Statistic Prob.
UNEMPLOYMENT -0.590586 0.162483 -3.634743 0.0003 C 3.162821 0.905592 3.492545 0.0005
Mean dependent var 0.464286 S.D. dependent var 0.503236 S.E. of regression 0.430697 Akaike info criterion 1.131294 Sum squared resid 10.01698 Schwarz criterion 1.203628 Log likelihood -29.67623 Hannan-Quinn criter. 1.159338 Restr. log likelihood -38.67326 Avg. log likelihood -0.529933 LR statistic (1 df) 17.99407 McFadden R-squared 0.232642 Probability(LR stat) 2.22E-05
Obs with Dep=0 30 Total obs 56 Obs with Dep=1 26
0
2
4
6
8
10
12
1947
1951
1955
1959
1963
1967
1971
1975
1979
1983
1987
1991
1995
1999
Year
Un
emp
loym
ent
(%)
Democrat
Republican
Two Year Forward Lag
BeforeAfter
0
2
4
6
8
10
12
1947
1951
1955
1959
1963
1967
1971
1975
1979
1983
1987
1991
1995
1999
Year
Une
mpl
oym
ent (
%)
Democrat
Republican
Probability of Democrat President Given Unemployment Rate (Two Year Lag)
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12
Unemployment Rate (%)
Pro
bab
ilit
y P
resi
den
t is
D
emo
crat
Data Points
Probit Model
Probit Model of Two Year Lag
Two Year Lag ResultsDependent Variable: LAG2 Method: ML - Binary Probit Date: 11/20/02 Time: 21:16 Sample: 1947 2002 Included observations: 56 Convergence achieved after 3 iterations Covariance matrix computed using second derivatives
Variable Coefficient Std. Error z-Statistic Prob.
UNEMPLOYMENT -0.411472 0.126972 -3.240644 0.0012 C 2.257277 0.733633 3.076846 0.0021
Mean dependent var 0.482143 S.D. dependent var 0.504203 S.E. of regression 0.448275 Akaike info criterion 1.246607 Sum squared resid 10.85131 Schwarz criterion 1.318941 Log likelihood -32.90499 Hannan-Quinn criter. 1.274650 Restr. log likelihood -38.78052 Avg. log likelihood -0.587589 LR statistic (1 df) 11.75107 McFadden R-squared 0.151507 Probability(LR stat) 0.000608
Obs with Dep=0 29 Total obs 56 Obs with Dep=1 27
Conclusions
• A significant correlation between unemployment rate and political party affiliation was discovered
• A stronger correlation was found when applying a one year lag, and to a lesser degree, a two year lag, on this relationship
• Overall, we conclude from our data that unemployment is significantly greater when a Republican president is in office when compared to a Democrat