Predicting the DJIA: Going up the Income Statement

1
Predicting the DJIAGoing up the Income
Statement

• BSD
• Roberto Alatorre
• Prakash Arya
• Brian Chin
• Philipp Schmahl
• John Withrow

2
Agenda

• Data Collection
• Data Organization
• Results
• Conclusions

3
Data Collection

• Collected data on the 30 companies that make up
  the DJIA
• Quarterly Data
• Q1 1979 through Q4 1999
• DJIA composition changes in
• - Q2 1979 - Q4 1985
• - Q1 1991 - Q2 1997
• - Q4 1998 - Q4 1999

4
Data Organization

• Income Statement data on a per share basis
• EBITDA Yield for DJIA
• Dividend Yield for DJIA
• Earnings for DJIA
• Divisor Calculation
• (? Share Prices) / (Actual DJIA)
• Base case and adjusting for stock splits, etc.
• Modeling the DJIA

5
BSD-DJIA vs. DJIA
6
Hypothesis

• The further you go up the Income Statement, the
  better the variables will be to predict the
  returns for the DJIA
• Rationale
• Dividends are at managers discretion
• EBITDA is hard to manipulate

7
Regression Results
Dividend Model DJIAreturns - 0.054 -
5.016CrSprdLag 2.273TrmSprdLag
1.255TyldLag 5.966DivLag
Earnings Model DJIAreturns 0.067
4.639CrSprdLag 2.479TrmSprdLag
0.788TyldLag 0.924EarnLag
EBITDA Model DJIAreturns 0.008
4.018CrSprdLag 2.449TrmSprdLag
1.014TyldLag 1.002EbitLag
8
Regression Results
Model Constant Credit Spread Term Spread T-Bond 5yr Category Yield
Dividend -0.0538 -5.0156 2.2728 1.2551 5.9663
Earnings 0.0667 -4.6387 2.4787 0.7878 0.9243
EBITDA 0.0076 -4.0177 2.4492 1.0137 1.0021
9
Regression Results
p-Values Dividend Model
0.0071 Earnings Model 0.1034 EBITDA Model
0.0078
10
DJIA In-Sample Results
11
DJIA Out-Sample Results
12
Conclusions

• For DOW 30 Blue Chip Companies, EBITDA Yield is
  as good a predictor as Dividend Yield
• Confidence interval on predictions from Earning
  Yield is less than 90
• ARCH underestimates DJIA in sample and in
  validation period