1 Economics 240A Economics 240A Power Eight Power Eight
Dec 21, 2015
11
Economics 240AEconomics 240A
Power EightPower Eight
22
OutlineOutline Lab FourLab Four Maximum Likelihood EstimationMaximum Likelihood Estimation The UC Budget AgainThe UC Budget Again Regression ModelsRegression Models The Income Generating Process for an The Income Generating Process for an
Asset Asset
33
UCBUDGSH(t) = a + b*t + e(t)UCBUDGSH(t) = a + b*t + e(t)UC Budget Share of CA General Fund Expenditure:1968-69 through 2007-08
2007-08
68-69 y = -0.0009x + 0.0704
R2 = 0.8663
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
7.00%
8.00%
Fiscal year
Pe
rce
nt
44
UCBUDSH(t) = a + b*t + e(t)UCBUDSH(t) = a + b*t + e(t)UC Budget Share of CA General Fund Expenditure:68-69 through 2007-08
y = -0.0009x + 0.0694
R2 = 0.8663
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
7.00%
8.00%
0 5 10 15 20 25 30 35 40 45
Time
Pe
rce
nt
5.105
19.5
55
UC Budget Share of General Fund Expenditure, 1968-69 through 2005-06
1968-69
2005-06
y = -0.0009x + 0.0691
R2 = 0.8449
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
7.00%
8.00%
0 5 10 15 20 25 30 35 40
Year
Pe
rce
nt
means: 5.22%, 18.5 yr.
UCBUDSH(t) = a + b*t + e(t)UCBUDSH(t) = a + b*t + e(t)
66
How to Find a-hat and b-hat?How to Find a-hat and b-hat?
MethodologyMethodology grid searchgrid search differential calculusdifferential calculus likelihood functionlikelihood function
motivation: the likelihood function connects the topics motivation: the likelihood function connects the topics of of probability probability (especially independence), the practical (especially independence), the practical application of application of random samplingrandom sampling, the , the normal normal distributiondistribution, and the derivation of estimators, and the derivation of estimators
77
Likelihood functionLikelihood function
The joint density of the estimated residuals The joint density of the estimated residuals can be written as:can be written as:
If the sample of observations on the If the sample of observations on the dependent variable, y, and the independent dependent variable, y, and the independent variable, x, is random, then the observations variable, x, is random, then the observations are independent of one another. If the errors are independent of one another. If the errors are also identically distributed, f, i.e. i.i.d, are also identically distributed, f, i.e. i.i.d, thenthen
)ˆ.....ˆˆˆ( 1210 neeeeg
88
Likelihood functionLikelihood function Continued: If i.i.d., thenContinued: If i.i.d., then
If the residuals are normally distributed:If the residuals are normally distributed:
This is one of the assumptions of linear This is one of the assumptions of linear regression: errors are i.i.d normalregression: errors are i.i.d normal
then the joint distribution or likelihood then the joint distribution or likelihood function, L, can be written as:function, L, can be written as:
)ˆ()...ˆ(*)ˆ()ˆ...ˆˆ( 110110 nn efefefeeeg
2]/)0ˆ[(2/12 )2/1(),0(~)ˆ( iei eNef
99
Likelihood functionLikelihood function
and taking natural logarithms of both sides, where and taking natural logarithms of both sides, where the logarithm is a monotonically increasing the logarithm is a monotonically increasing function so that if lnL is maximized, so is L:function so that if lnL is maximized, so is L:
1
0
22
2
]ˆ[)2/1(2/2/2
]/)0ˆ[(2/11
0110
*)2/1(*)/1(
)2/1()ˆ...ˆˆ(
n
ii
i
enn
en
in
eL
eeeegL
1010
The Natural Logarithm Function
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
0 1 2 3 4 5 6
x
lnx
1111
Log-LikelihoodLog-Likelihood
Taking the derivative of lnL with respect to Taking the derivative of lnL with respect to either a-hat or b-hat yields the same either a-hat or b-hat yields the same estimators for the parameters a and b as with estimators for the parameters a and b as with ordinary least squares, except now we know ordinary least squares, except now we know the errors are normally distributed.the errors are normally distributed.
21
0
22
1
0
222
]*ˆˆ[)2/1()2ln(*)2/(]ln[*)2/(ln
ˆ)2/1()2ln(*)2/(]ln[*)2/(ln
i
n
ii
n
ii
xbaynnL
ennL
1212
Log-LikelihoodLog-Likelihood Taking the derivative of lnL with respect to Taking the derivative of lnL with respect to
sigma squared, we obtain an estimate for the sigma squared, we obtain an estimate for the variance of the errors:variance of the errors:
andand
in practice we divide by n-2 since we used up in practice we divide by n-2 since we used up two degrees of freedom in estimating a-hat and two degrees of freedom in estimating a-hat and b-hat. b-hat.
0ˆ)/1(*)2/1()/1(*)2/(/ln1
0
2422
n
iienL
nen
ii /]ˆ[ˆ
1
0
22
1313
Interpreting Excel OutputInterpreting Excel Output
1414
The sum of squared residuals (estimated)The sum of squared residuals (estimated)
2ie
1515
CA Size of Govt. Vs. SIze of Economy
y = 0.0657x - 1.0238
R2 = 0.9902
0
10
20
30
40
50
60
70
80
90
100
0 200 400 600 800 1000 1200 1400
CAPY, B Nom.$
Ge
n. F
un
d E
x. B
No
m. $
CAGFD(t) = a + b*CAPY(t) +e(t): CAGFD(t) = a + b*CAPY(t) +e(t): 1968-69 through 2005-061968-69 through 2005-06
1616
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.99510756
R Square 0.99023905
Adjusted R Square 0.98996792
Standard Error 2.52724563
Observations 38
ANOVA
df SS MS F Significance F
Regression 1 23326.28511 23326.29 3652.16735 8.58311E-38
Residual 36 229.9309379 6.38697
Total 37 23556.21605
Coefficients Standard Error t Stat P-value Lower 95% Upper 95%
Intercept -1.02377776 0.727626534 -1.40701 0.167999648 -2.499472762 0.451917
X Variable 1 0.06565026 0.001086328 60.43316 8.58311E-38 0.063447085 0.067853
2ie
Goodness of fit, R2
Number of Observations, n
Regress CA State General Fund Expenditures on CA Personal Income, Lab Four
1717
Estimated CoefficientsEstimated Coefficients
CoefficientsStandard
Error t Stat P-value Lower 95%Upper
95%
Intercept -1.02377776 0.727626534 -1.40701 0.167999648 -2.499472762 0.451917
X Variable 1 0.06565026 0.001086328 60.43316 8.58311E-38 0.063447085 0.067853
a
b
41.1727.0/)0204.1(ˆ/)]ˆ(ˆ[ ˆˆ aa aEat
1818
Appendix BTable 4p. B-9
2.5 % in the upper tail
From Power 6:Student’s t-distributionText: pp. 260-2
1919
Table of Analysis of VarianceTable of Analysis of Variance
ANOVA
Mean Mean
SquareSquare
=SS/df=SS/df
df SS MS F Significance F
Regression 1 23326.28511 23326.29 3652.16735 8.58311E-38
Residual 36 229.9309379 6.38697
Total 37 23556.21605
Degrees ofFreedom
Sum of Squares F1, 37 = EMS/UMS
2020
The Intuition Behind the Table of The Intuition Behind the Table of Analysis of Variance (ANOVA)Analysis of Variance (ANOVA)
y = a + b*x + ey = a + b*x + e the variation in the dependent variable, y, is the variation in the dependent variable, y, is
explained by either the regression, a + b*x, or by explained by either the regression, a + b*x, or by the error, ethe error, e
The sample sum of deviations in y:The sample sum of deviations in y:
21
0
][ yyn
ii
2121
Table of ANOVATable of ANOVA
Source Degrees ofFreedom
Sum ofSquares
MeanSquare
Regression(a + b*x
1
Error (e) n-2
Total (y) n-1
ANOVAdf SS MS F Significance F
Regression 1 23326.28511 23326.29 3652.167 8.58311E-38Residual 36 229.9309379 6.38697Total 37 23556.21605
21
0
][ yyn
ii
2ie
By difference
)2/(}ˆ{ 2 nei
2222
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.99510756
R Square 0.99023905
Adjusted R Square 0.98996792
Standard Error 2.52724563
Observations 38
ANOVA
df SS MS F Significance F
Regression 1 23326.28511 23326.29 3652.16735 8.58311E-38
Residual 36 229.9309379 6.38697
Total 37 23556.21605
Coefficients Standard Error t Stat P-value Lower 95% Upper 95%
Intercept -1.02377776 0.727626534 -1.40701 0.167999648 -2.499472762 0.451917
X Variable 1 0.06565026 0.001086328 60.43316 8.58311E-38 0.063447085 0.067853
2ie
Goodness of fit, R2
Number of Observations, n
Regress CA State General Fund Expenditures on CA Personal Income, Lab Four
)2/(ˆˆ 2 nee
2323
Test of the Significance of the Test of the Significance of the Regression: F-testRegression: F-test
FF1,n-2 1,n-2 = explained mean square/unexplained mean = explained mean square/unexplained mean
squaresquare example: Fexample: F1, 36 1, 36 = = 23326.29 / 6.387= 365223326.29 / 6.387= 3652
2424
Table 6,pp. B-11 throughB-16Text: pp.270-274
2525
The UC BudgetThe UC Budget
2626
The UC BudgetThe UC Budget
The UC Budget can be written as an The UC Budget can be written as an identity:identity:
UCBUD(t)= UC’s Gen. Fnd. Share(t)* The UCBUD(t)= UC’s Gen. Fnd. Share(t)* The Relative Size of CA Govt.(t)*CA Personal Relative Size of CA Govt.(t)*CA Personal Income(t)Income(t) where UC’s Gen. Fnd. Share=UCBUD/CA where UC’s Gen. Fnd. Share=UCBUD/CA
Gen. Fnd. ExpendituresGen. Fnd. Expenditures where the Relative Size of CA Govt.= CA Gen. where the Relative Size of CA Govt.= CA Gen.
Fnd. Expenditures/CA Personal IncomeFnd. Expenditures/CA Personal Income
2727
Long Run Political TrendsLong Run Political Trends
UC’s Share of CA General Fund ExpendituresUC’s Share of CA General Fund Expenditures
2828
The Regression Passes Through The Regression Passes Through the Means of y and xthe Means of y and x
UC Budget Share of General Fund Expenditure, 1968-69 through 2005-06
1968-69
2005-06
y = -0.0009x + 0.0691
R2 = 0.8449
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
7.00%
8.00%
0 5 10 15 20 25 30 35 40
Year
Pe
rce
nt
means: 5.22%, 18.5 yr.
3.27%
2929
UC’s Budget ShareUC’s Budget Share
UC’s share of California General Fund UC’s share of California General Fund expenditure shows a long run downward expenditure shows a long run downward trend. Like other public universities across trend. Like other public universities across the country, UC is becoming less public and the country, UC is becoming less public and more private. Perhaps the most “private” of more private. Perhaps the most “private” of the public universities is the University of the public universities is the University of Michigan. Increasingly, public universities Michigan. Increasingly, public universities are looking to build up their endowments are looking to build up their endowments like private universities.like private universities.
3030
Long Run Political Trends Long Run Political Trends
The Relative size of California GovernmentThe Relative size of California Government The Gann Iniative passed on the ballot in 1979. The Gann Iniative passed on the ballot in 1979.
The purpose was to limit the size of state The purpose was to limit the size of state government so that it would not grow in real government so that it would not grow in real terms per capita.terms per capita.
Have expenditures on public goods by the Have expenditures on public goods by the California state government grown faster than California state government grown faster than personal income?personal income?
3131
The Size of CA State Government Relative to the Economy
6.48%
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
7.00%
8.00%
1968
-69
1970
-71
1972
-73
1974
-75
1976
-77
1978
-79
1980
-81
1982
-83
1984
-85
1986
-87
1988
-89
1990
-91
1992
-93
1994
-95
1996
-97
1998
-99
2000
-01
2002
-03
2004
-05
Fiscal Year
Pe
rce
nt
3232
The Relative Size of CA State Govt.The Relative Size of CA State Govt.
California General Fund Expenditure was California General Fund Expenditure was growing relative to personal income until growing relative to personal income until the Gann initiative passed in 1979. Since the Gann initiative passed in 1979. Since then this ratio has declined, especially in the then this ratio has declined, especially in the eighties and early nineties. After recovery eighties and early nineties. After recovery from the last recession, this ratio recovered, from the last recession, this ratio recovered, but took a dive in 2003-04.but took a dive in 2003-04.
3333
Guessing the UC Budget for Guessing the UC Budget for 2005-062005-06
UC’s Budget Share, 05-06: 0.0327UC’s Budget Share, 05-06: 0.0327 Relative Size of CA State Govt.: 0.0648Relative Size of CA State Govt.: 0.0648 Forecast of CA Personal Income for 2006-07 Forecast of CA Personal Income for 2006-07
3434
California Personal Income, Billions of Nominal $, 1968-69 through 2005-06
2005-06, $1.324B
0
200
400
600
800
1000
1200
1400
68-6
9
70-7
1
72-7
3
74-7
5
76-7
7
78-7
9
80-8
1
82-8
3
84-8
5
86-8
7
88-8
9
90-9
1
92-9
3
94-9
5
96-9
7
98-9
9
00-0
1
02-0
3
04-0
5
Fiscal Year
Bill
ion
s o
f $
3535
3636
3737
3838
3939
4040
Guessing the UC Budget for 2005-06Guessing the UC Budget for 2005-06 UC’s Budget Share, 05-06: 0.0327UC’s Budget Share, 05-06: 0.0327 Relative Size of CA State Govt.: 0.0648Relative Size of CA State Govt.: 0.0648 Forecast of CA Personal Income for 2006-07: $ Forecast of CA Personal Income for 2006-07: $
1,406.5 B1,406.5 B UCBUD(06-07) = 0.0327*0.0648*$1,406.5BUCBUD(06-07) = 0.0327*0.0648*$1,406.5B UCBUD(06-07) = $ 2.98 BUCBUD(06-07) = $ 2.98 B compares to UCBUD(05-06) = $ 2.81 Bcompares to UCBUD(05-06) = $ 2.81 B An increase of $170 millionAn increase of $170 million
4141
UC Budget in Billions, 1968-69 through 2005-06
y = 0.0805x + 0.1147
R2 = 0.9344
0
0.5
1
1.5
2
2.5
3
3.5
4
0 5 10 15 20 25 30 35 40
Year
$
Forecast:$2.98 B
4242
The Relative Size of CA Govt.The Relative Size of CA Govt. Is it determined politically or by economic Is it determined politically or by economic
factors?factors? Economic Perspective: Engle Curve- the Economic Perspective: Engle Curve- the
variation of expenditure on a good or service variation of expenditure on a good or service with incomewith income
lnCAGenFndExp = a + b lnCAPersInc +e lnCAGenFndExp = a + b lnCAPersInc +e b is the elasticity of expenditure with incomeb is the elasticity of expenditure with income
bCAPersIncpCAGenFndEx ln/ln
4343
The elasticity of expenditures The elasticity of expenditures with respect to incomewith respect to income
Note:Note:
So, in the log-log regression, So, in the log-log regression, lny = a + b*lnx + e, the lny = a + b*lnx + e, the coefficient b is the elasticity of y with respect coefficient b is the elasticity of y with respect to x.to x.
)/1(*
)/(*)/1(
/ln
CAPersIncb
CAPersIncpCAGenFndExpCAGenFndEx
CAPersIncpCAGenFndEx
4444
Logarithms of California General Fund Expenditures and Personal Income, 1968-69 through 2005-06
y = 0.927x - 3.3845
R2 = 0.99
4
4.5
5
5.5
6
6.5
7
7.5
8 8.5 9 9.5 10 10.5 11 11.5 12
LnCAPY(t)
Ln
CA
Ge
nF
nd
Ex
(t)
Lncagenfndex(t) = a +b*lncapy(t) + e(t)Lncagenfndex(t) = a +b*lncapy(t) + e(t)
4545
80.30179.0/)1068.1(/)]ˆ(ˆ[
1:,1:
ˆˆ
0
bb
a
bEbt
bHbH
4646
Is the Income Elasticity of CA Is the Income Elasticity of CA State Public Goods >1?State Public Goods >1?
Step # 1: Formulate the HypothesesStep # 1: Formulate the Hypotheses HH0 0 : b = 1: b = 1
HHa a : b > 1: b > 1
Step # 2: choose the test statisticStep # 2: choose the test statistic
Step # 3: If the null hypothesis were true, Step # 3: If the null hypothesis were true, what is the probability of getting a t-statistic what is the probability of getting a t-statistic this big?this big?
8.30179.0/)1068.1(/)]ˆ(ˆ[ ˆ b
bEbstatt
4747
Appendix BTable 4p. B-9
5.0 % in the upper tail
t..050
35 1.69
4848
Regression ModelsRegression Models Trend AnalysisTrend Analysis
linear: y(t) = a + b*t + e(t)linear: y(t) = a + b*t + e(t) exponential: lny(t) = a + b*t + e(t)exponential: lny(t) = a + b*t + e(t) Y(t) =exp[a + b*t + e(t)]Y(t) =exp[a + b*t + e(t)]
Engle CurvesEngle Curves ln y = a + b*lnx + eln y = a + b*lnx + e
Income Generating ProcessIncome Generating Process
4949
Returns Generating ProcessReturns Generating Process
How does the rate of return on an asset vary How does the rate of return on an asset vary with the market rate of return?with the market rate of return?
rrii(t): rate of return on asset i(t): rate of return on asset i
rrff(t): risk free rate, assumed known for the (t): risk free rate, assumed known for the
period aheadperiod ahead rrMM(t): rate of return on the market(t): rate of return on the market
[r[rii(t) - r(t) - rff00(t)] = a +b*[r(t)] = a +b*[rMM(t) - r(t) - rff
00(t)] + e(t) (t)] + e(t)
5050
ExampleExample rrii(t): monthly rate of return on UC stock index (t): monthly rate of return on UC stock index
fund, Sept., 1995 - Sept. 2003fund, Sept., 1995 - Sept. 2003 rrff(t): risk free rate, assumed known for the period (t): risk free rate, assumed known for the period
ahead. Usually use Treasury Bill Rate. I used ahead. Usually use Treasury Bill Rate. I used monthly rate of return on UC Money Market monthly rate of return on UC Money Market Fund Fund http://atyourservice.ucop.edu/employees/retiremehttp://atyourservice.ucop.edu/employees/retirement/performance.htmlnt/performance.html
5151
Example (cont.)Example (cont.)
rrMM(t): rate of return on the market. I used the (t): rate of return on the market. I used the
monthly change in the logarithm of the total monthly change in the logarithm of the total return (dividends reinvested)*100. return (dividends reinvested)*100. http://research.stlouisfed.org/fred2/http://research.stlouisfed.org/fred2/
5252
Returns Generating Process Time Series Data
-20
-15
-10
-5
0
5
10
15
Se
p-9
5
Se
p-9
6
Se
p-9
7
Se
p-9
8
Se
p-9
9
Se
p-0
0
Se
p-0
1
Se
p-0
2
Se
p-0
3
Date
Mo
thly
Ra
te o
f R
etu
rn
UC Equity Fund
Standard & Poors 500
UC Money Market Fund
5353
Returns Generating Process, Sept. 95-Sept. 03
-20.00
-15.00
-10.00
-5.00
0.00
5.00
10.00
15.00
-15 -10 -5 0 5 10
Standard & Poors 500, Net
UC
Sto
ck In
dex
Fu
nd
, Net
5454
-13.35, 16.09;Ucnet,
S&Pnet
y = 1.0601x - 0.106
R2 = 0.9136
-20.00
-15.00
-10.00
-5.00
0.00
5.00
10.00
15.00
-15 -10 -5 0 5 10
Watch Excel on xy plots!
True x axis: UC Net
5555
5656
Returns Generating Process
y = 1.0601x - 0.106
R2 = 0.9136
-20.00
-15.00
-10.00
-5.00
0.00
5.00
10.00
15.00
-15 -10 -5 0 5 10
Standard & Poors 500, Net
UC
Sto
ck I
nd
ex F
un
d,
Net
Really the Regression of S&P on UC
5757
SUMMARY OUTPUT
Regression StatisticsMultiple R 0.95580613R Square 0.91356536Adjusted R Square 0.91265552Standard Error 1.31011043Observations 97
ANOVAdf SS MS F Significance F
Regression 1 1723.42 1723.42 1004.096 2.65348E-52Residual 95 163.057 1.716389Total 96 1886.477
CoefficientsStandard Error t Stat P-value Lower 95% Upper 95%Lower 95.0%Upper 95.0%Intercept 0.12800497 0.13335 0.959915 0.339535 -0.1367287 0.392739 -0.13673 0.392739X Variable 1 0.86177094 0.027196 31.68748 2.65E-52 0.807780204 0.915762 0.80778 0.915762
5858
Is the beta for the UC Stock Is the beta for the UC Stock Index Fund <1?Index Fund <1?
Step # 1: Formulate the HypothesesStep # 1: Formulate the Hypotheses HH0 0 : b = 1: b = 1
HHa a : b < 1: b < 1
Step # 2: choose the test statisticStep # 2: choose the test statistic
Step # 3: If the null hypothesis were true, Step # 3: If the null hypothesis were true, what is the probability of getting a t-statistic what is the probability of getting a t-statistic this big?this big?
4.6027.0/)1862.0(/)]ˆ(ˆ[ ˆ b
bEbstatt
5959
Appendix BTable 4p. B-9
5.0 % in the lower tail
t..050
95 1.66
6060
-15
-10
-5
0
5
10
-20 -10 0 10
SPNET
UC
ST
OC
KN
ET
Returns Generating ProcessEViews Chart
6161
Midterm 2001Midterm 2001
6262
1. (15 points) The following graph 4-1 shows the results of regressing California
General Fund expenditures, in billions of nominal dollars, against California Personal
Income, in billions of nominal dollars beginning in fiscal year1968-69 and ending in
fiscal year 2001-02.
a. How much of the variance in the dependent variable is explained by personal
income?
b. Interpret the estimated slope.
Table 4-1 follows with the estimated parameters and table of analysis of variance.
c. Is the slope significantly different from zero? What statistic do you use to
answer this question? What distribution do you use to answer this question?
What probability were you willing to accept for a Type I error?
Q. 4
d. What is the ratio of the explained mean square to the unexplained mean square?
6363
Calfifornia General Fund Expenditures Vs. California Personal Income, Billions of Nominal $
y = 0.066x - 1.1974
R2 = 0.981
0
10
20
30
40
50
60
70
80
90
0 200 400 600 800 1000 1200 1400
Personal Income
Gen
Fu
nd
Exp
end
itu
res
Q 4
Figure 4-1: California General Fund Expenditures Versus California Personal Income, both in Billions of Nominal Dollars
6464
Regression StatisticsMultiple R 0.9904673R Square 0.9810255Adjusted R Square 0.9804325Standard Error 2.9988336Observations 34
ANOVA
df SS MS F SignificanceF
Regression 1 14878.68965 14878.69 1654.47398 3.98668E-29Residual 32 287.7761003 8.993003Total 33 15166.46575
Coefficients Standard Error t Stat P-value Lower 95% Upper 95%Intercept -1.197411 0.927956018 -1.29037 0.20616709 -3.08759378 0.6927721X Variable 1 0.0659894 0.001622349 40.67523 3.9867E-29 0.062684796 0.069294
Q 4Table 4-1: Summary Output