Our project focuses on generating models by weighted linear regression to predict the net income of JP Morgan for the next nine quarters. Model generation and stress testing have played an integral role after the financial crisis of 2008. Hence, we have further conducted a stress test datasets of 2016 Dodd-Frank Act Stress Test (DFAST) for our final model, which determines the efficiency and reliability of our prediction and the standing of the bank.
3. Focus
• Predicting the net income of banks for the next
nine quarters, namely JP Morgan
• Conduct a stress test to measure the
effectiveness of the models created for the
prediction
3
Yiqing Liu @ Rutgers University
4. 4
• Financial crises of
2008
• To develop model
• To predict the
efficiency of assets
and other domestic
factors
• To provide deeper
insight into making a
stronger foundation
for future shocks
Importance
Yiqing Liu @ Rutgers University
5. Independent Variables
• Dodd-Frank Act Stress Test 2016
• 16 domestic factors
• 1st quarter in 2001 to 1st quarter in 2019.
5
Training Testing Stress testing
2002,4-2013,3 2013,4-2015,4 2015,4-2019,1
44 9 10
Yiqing Liu @ Rutgers University
6. • Economic activity and prices: This category would include six variables
namely; percentage change in real and nominal gross domestic product,
the unemployment rate, percentage change in real and nominal disposable
personal income and the percentage change in the consumer price index.
• Aggregate measures of asset prices or financial conditions: This
category includes four variables namely; indexes of house prices,
commercial property prices, equity prices and the US stock market volatility.
• Interest rates: This category includes six variables namely; the rate on the
3-month Treasury bill, the yield on the 5-year Treasury bond, the yield on the
10-year Treasury bond, the yield on a 10-year BBB corporate security, the
interest rate associated with a conforming, conventional, fixed-rate 30-year
mortgage and, the prime rate.
6
Independent variables
Yiqing Liu @ Rutgers University
7. 7
• Net incomes
• JPMORGAN
CHASE BANK,
NATIONAL
ASSOCIATION
• 2002, 4th
to 2016, 3rd
Responsible Variable
Yiqing Liu @ Rutgers University
17. 17
Clustering by K-means
Cluster R2
1 0.9915
2 0.8718
3 0.6867
• Limited data set
• Good for Predict
• Bad for Stress
Testing
Yiqing Liu @ Rutgers University
18. 18
• Inspired by Prime Rate
• Using inverse 10 years Treasury Yield
Weighted Model
Yiqing Liu @ Rutgers University
24. • Our project includes generating various different models to predict
the net income of JP Morgan by methods used in class, which is
followed by stress testing to affirm the efficiency of our final model.
• In order to predict results more accurately, we have added less
weight to the observations before the financial crises and more
weight to the observations after the 2008 crises.
• We have classified our observations into good and bad years and
generated models pertaining to the type of year. More research can
be conducted on this topic with more observations in the future.
• JP Morgan seems to portray a good trend in predicted net income
and the stress test supports our argument based on our final model.
Conclusion
24
Yiqing Liu @ Rutgers University