Transcript Lies, Damn Lies, and Statistics

FIN 30220: Macroeconomic
Analysis
Using Economic Data
“There are three kinds of lies; Lies, Damn
Lies, and Statistics”
- Mark Twain
Principle #1: What are you trying to measure? How is your statistic
defined ? Is your statistic consistent with what you are trying to measure?
Example: Poverty in the US
Lets see how we compare to other parts of the world…
But, How would you define poverty?
Source: CIA Factbook
Poverty was defined by Mollie Orshansky of the SSA in 1964 as 3 times
the cost of the Dept. of Agriculture’s “Low cost food plan”
2014 POVERTY GUIDELINES
Family Size Threshold
One
$11,670
Two
$15,730
Three
$19,790
Four
$23,850
Five
$27,910
Six
$31,970
Seven
$36,030
Eight
$40,090
Nine
$44,150
That number has been indexed by inflation every year (The Johnson Administration later
substituted the “economy” food plan)
USDA Food Plans: 1964
$6.5052  $3383  $1014
$1014 1.05  $11, 627
(Poverty Line in 1964)
50
(Approximate Poverty Line Today)
Food Budget as a percentage of household income: 1964
Average
$20,000 - $24,999
$12,000 - $14,999
$8,000 - $9,999
$6,000 - $6,999
$4,000 - $4,999
Under $3,000
0
20
40
60
Lower Income Households spent around 33% on Food in
1964
Food Budget as a percentage of household income: 2003
16.5%
In Fact, No Family Spends Anything Near 33% on Food Today!!!
Suppose we use current food prices and the current budget share
$4352  $2,236  $2,2366  $13,416 (Poverty Line in 2014?)
Assuming Food is 16.5% of ones budget
Actual Calculation
$6.5052  $3383  $1014  $1014 1.05  $11, 627
50
So, which is it and why should we care?
15%
Difference
Redefined Household Budget (Current food prices)
$4352  $2,236  $2,2366  $13,416
A more important question: Is it “absolute” income that we really care
about?
The Simpsons have a
household income of
$35,000. Median
income is Springfield
is $50,000
The Griffins have a
household income of
$45,000. Median
income is Quahog is
$85,000
Which of these two families do you think is happier?
Relative poverty measures define poverty as a certain percentage of median
household income
Poverty Line for 3.5 Person Household= $20,000
(40% of Median)
Median Household Income = $50,000
Altering the definition of poverty can make a big difference when
comparing across countries!!
Absolute poverty rate
(threshold set at 40% of
U.S. median household
income)
Relative poverty rate (40% of Median Income)
Pretransfer
Post-transfer
Pre-transfer
Post-transfer
Sweden
23.7
5.8
14.8
4.8
Norway
9.2
1.7
12.4
4.0
Netherlands
22.1
7.3
18.5
11.5
Finland
11.9
3.7
12.4
3.1
Denmark
26.4
5.9
17.4
4.8
Germany
15.2
4.3
9.7
5.1
Switzerland
12.5
3.8
10.9
9.1
Canada
22.5
6.5
17.1
11.9
France
36.1
9.8
21.8
6.1
Belgium
26.8
6.0
19.5
4.1
Australia
23.3
11.9
16.2
9.2
United
Kingdom
16.8
8.7
16.4
8.2
United States 21.0
11.7
21.0
11.7
Italy
14.3
19.7
9.1
Country
30.7
It also makes a big difference when looking across time periods!
The common international poverty line has in the past been roughly $1
a day. In 2008, the World Bank came out with a revised figure of $1.25.
Percentage of Population Living on Less that $1.25/day
*Source: United Nations
Principle #2: How is your variable measured?
Example: U.S. Unemployment
1991 Recession
2001 Recession
2007 Recession
Each month, the Department of Labor surveys 60,000 households. Each
household is asked a series of questions:
1) “Are you currently
working?” (Note: no
mention of part time or
full time)
YES
You are employed
(147 Million)
No
2) “Have you looked for
a job in the past 30
days?”
YES
You are unemployed
(9 Million)
No
You are not in the labor force
(93 Million)
Unemployment =
Rate (UR)
=
Unemployed
Labor Force
9
= .057
147+ 9
(5.7%)
Over the same month, the Department of Labor surveys 400,000
businesses and asks one question.
1) “How many
employees are
currently on your
payroll?”
Total Non-Farm Payrolls
(140 Million)
Wait a minute, that’s not what
the household survey
reported??? Which is it ???
1) “Are you currently
working?” (Note: no
mention of part time or
full time)
YES
You are employed
(147 Million)
The two surveys track each other reasonably well, but there are noticeable
differences.
However, the establishment survey is subject to fairly large revisions
Revisions to Total Payroll Employment
Preliminary minus Current Estimate, thousands
1,500
Jo b s o ver est i mat ed
1,000
500
0
-500
-1,000
Jo b s und er est i mat ed
-1,500
1990
1992
1994
1996
Source: Bureau of Labor Statistics
1998
2000
2002
2004
The Establishment Survey Will often times “Double Count” Jobs
Suppose you quit your job at company A and find a new job at
company B – if this is done in the same payroll period (most payrolls
are bi-weekly) you will be counted twice!!
No Turnover
Turnover
Pay Period
Pay Period
Job at Company A
Job at Company B
count = 1 payroll job
count = 2 payroll jobs
In months with high job turnover, the establishment survey will overstate employment.
Average US Labor Market Turnover
8.11M jobs are gained
per quarter
FYI: Worstcase
estimates
predict
outsourcing
costs us
55,000 jobs
per quarter.
7.71M jobs arelost
per quarter
Household Survey vs. Establishment Survey
The household survey includes agricultural workers, self
employed workers and private household workers. The
establishment survey does not.
The household survey counts people on
unpaid leave as employed – the
establishment survey does not.
The household survey only counts people over the age
of 16 – the establishment survey is not limited by age.
Main problems with measuring the
unemployment rate




The unemployment rate doesn’t count
underemployment (those that would like to work full
time, but only work part time)
The “discouraged worker effect”: Those that have
given up trying to find a job are counted as not in the
labor force rather than unemployed
Selection bias: those that are unemployed are more
likely to answer the survey.
Moral hazard: due to unemployment insurance, it is
difficult to tell how hard individuals are trying to find
work
Principle #3: Is your variable in terms of current prices or fixed prices
(Real vs. Nominal)
Example: The Top 10 All Time Grossing Films (in Millions – US)
1)
Avatar (2009): $760
2)
Titanic(1997): $658
3)
Marvel’s the Avengers (2012): $588
4)
The Dark Knight (2008): $533
5)
Star Wars I: The Phantom Menace (1999)
$474
6)
Star Wars IV: A New Hope (1977): $460
7)
The Dark Knight Rises (2012) $449
8)
Shrek 2 (2011): $441
9)
E.T. The Extra-Terrestrial (1982): $435
10) The Hunger Games: Catching Fire (2013):
$424
Real vs. Nominal Variables
Nominal Variables are in terms of a current dollars. For
example, you’re starting salary after college might be
$50,000 per year.
Real variables are in terms of some fixed commodity. Real
variables measure purchasing power. If a gallon of gas costs
$2.00, then we can calculate your “real” income.
Real Income =
Nominal Income
$50,000
= 25,000
= $2.00
Price
In 2009, a gallon of gas cost $3.50
Real Gross =
Nominal Gross
$749M
= $3.50
Price
= 214M
(Gallons of Gas)
In 1977, a gallon of gas cost $.62
Real Income =
Nominal Gross
Price
$460M
= $.62
= 742M
(Gallons of Gas)
Usually, the “commodity” used for real
variables is a particular year’s dollars.
Suppose we want both grosses to reflect
1997 gas prices. (Gas was $1.26 in 1997)
Real X = Nominal X
$1.26
Real
= $749M $3.50
Gross
= $270M
( 1997 Dollars)
Target
Year
Current
Year
Target year Price
Current Year Price
$1.26
Real
= $460M $.62
Gross
= $935M
( 1997 Dollars)
These two dollar figures are comparable because they represent the same year’s dollars!
The Top 10 All Time Grossing Films– Inflation
Adjusted (Millions of 2000 Dollars)
1)
Gone With the Wind (1939): $1,689
2)
Star Wars Episode IV(1977): $960
3)
The Sound of Music(1965): $768
4)
ET: The Extraterrestrial(1982): $764
5)
The Ten Commandments (1956): $706
6)
Titanic (1997): $691
7)
Jaws (1975): $690
8)
Dr. Zhivago (1965): $669
9)
The Exorcist (1973): $596
10) Snow White (1937): $587
Notes: Avatar falls to #14 ($516), a movie ticket in 1939 was $0.23
Principle #4: Annualizing
Example: Treasury Yields
Suppose that you buy a
$1,000, 90 Day Treasury Bill
for $994
90 Days from now, you
receive $1,000 from the
government
$1,000 - $994
$994
X 100 = .6%
Alternatively, you could buy a
$1,000, 5 year bond for $902
5 years from now, you
receive $1,000 from the
government
$1,000 - $902
$902
X 100 =10.86%
Which of these two assets is paying a higher return?
Suppose that you could earn .6% interest every quarter
(90 days). How much would you have in a year?
$1
$1.006
$1.012
2
$1(1.006)
$1.018
3
$1(1.006)
$1.024
4
$1(1.006)
$1(1.006)
You earned 2.4% (Annualized)
For the 5 year bond, we do the same process in reverse (how much would
you have to earn per year to get a 10.86% return after 5 years)?
$1
$??
$1(1+i)
1  i 
5
$1(1+i)
$??
2
$??
$1(1+i)
3
$??
$1(1+i)
 1.1086  1  i  1.1086  1.02
1
5
4
$1.1086
$1(1+i)
5
You earned 2.0% (Annualized)
Principle #5: Economic data can be
can be broken into 3 components:
Trend (many years)
Business Cycle (1-5 years)
Seasonal (Months)
Recessions are periods of
below trend growth
GDP
Expansions are
periods of above
trend growth
Trend
Seasonal Cycle
Business Cycle
Time
Example: Tax Cuts, Tax Revenues and “VooDoo Economics”
The Bush Tax Cuts of 2001 & 2003 lowered marginal tax rates across the board,
lowered the capital gains tax, eliminated the estate tax, and lowered the
“marriage penalty
Bracket
Old Rate
New Rate
$0 - $6,000
15%
10%
$6,000 - $27,250
15%
15%
$27,251 - $67,550
28%
25%
$67,551 - $141,600
31%
28%
$141,601 - $307,300
36%
33%
$307,301 +
39.6%
35%
The tax cut was advertised as “the largest tax cut in history”
What do we mean by the “cost” of a tax cut,
anyways?
Suppose that the Griffin family has a household
income of $50,000. Currently, the income tax rate is
20% of all income earned
Under the current tax code, the Griffins
pay $10,000 per year in Taxes.
If the government cuts the tax rate to 10%,
then the Griffin’s tax bill falls to $5,000
The cost of the tax cut
is the $5,000 in lost
revenues
By this measure, the Bush Tax cuts have a price tag of around
$130 Billion per year!!
Let’s take a look at previous marginal tax rate changes to put the Bush tax cut
in a historical context.
Wilson
1917
Coolidge
1925
FDR
1933
Kennedy
1964
Reagan
1981
Bush
2001/2003
100
90
80
70
60
Bottom Rate
50
Top rate
40
30
20
10
0
1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003
Source: Congressional Budget Office
Given an income distribution in 1964, 1981, and
2001/2003, we can estimate the per year “cost” of the
three major tax cuts
Tax Bill
Cost in Dollars
(Billions)
Kennedy Tax Cut (1964) $11.5
Reagan Tax Cut (1981)
$38.3
Bush Tax Cut (2001)
$73.8
Bush Tax Cut (2003)
$60.8
What’s the problem with comparing these numbers?
When expressed in real terms as a percentage of GDP, the Bush tax
cuts aren’t so big after all!
Kennedy
1964
Cost (in 1964 dollars): $11.5B
CPI: 30.9
Real GDP: $2998.6B
175.1
$11.5B 30.9 = $67.6B (2.25% of GDP)
Reagan
1981
Cost (in 1981 dollars): $38.3B
CPI: 87.0
Real GDP: $5,291.7B
175.1
$38.3B 87.0 = $79.9B (1.5% of GDP)
Cost (in 2003 dollars): $134.6B
CPI: 175.1
Real GDP: $10,301B
175.1
$134.6B 175.1 =$134.65B (1.3% of GDP)
Bush
2001/2003
Let’s return to the Griffin family. The Griffin family
has a household income of $50,000. Currently, the
income tax rate is 20%.
Under the current tax code, the Johnsons
pay $10,000 per year in Taxes.
Suppose that a drop in the marginal rate
encourages Lois Griffin to go back to
work. With the two income earners, the
Griffin family income rises to $120,000.
At the 10% rate, their tax rises to
$12,000
Could this happen?
The drop in the tax rate
caused revenues to
increase rather than
decrease!
Tax Revenues = (Tax Rate) (Tax Base)
The basic logic behind the
Laffer Curve is that the tax base
should be negatively related to
the tax rate.
Tax
Revenues
Is there evidence of a Laffer
curve in practice?
Tax Rate
0%
Revenue
Maximizing Rate
100%
Tax
Revenues
History suggests that taxes are two high,
but be careful…
Tax Rate
0%
Revenue
Maximizing Rate
100%
Annual Federal
Year 0
Receipts (Billions)
Year 1
Year 2
Year 3
Year 4
Kennedy (1964)
$116.8
$130.8
$148.8
$153.0
$186.9
Reagan (1981)
$617.7
$600.5
$666.5
$734.0
$769.2
Bush (2001)
$1,853.4 $1,782.5
$1,880.2
$2,153.8 $2,402.7
Source: Congressional Budget Office
Once we account for price changes, the Laffer curve effect starts to disappear
Date
Revenues (Billions of
Dollars)
% Change
CPI
Real Revenues (Billions of
2003 Dollars)
%Change
1964
116.8
---
30.9
720.8
---
1965
130.8
11.3
31.2
799.4
10.3
1966
148.8
24.2
31.8
892.3
21.3
1967
153
27.0
32.9
886.8
20.7
1968
186.9
47.0
34.1
1045.2
37.2
1981
617.7
---
87.0
1353.9
---
1982
600.5
-2.82
94.3
1214.3
-10.9
1983
666.5
7.6
97.8
1299.6
-4.10
1984
734
17.25
101.9
1373.6
1.4
1985
769.2
21.9
105.5
1390.4
2.6
2001
1853.4
---
175.1
2018.5
---
2002
1782.5
-3.90
177.1
1919.4
-5.0
2003
1880.2
1.44
181.7
1973.3
-2.3
2004
2153.8
15.0
185.2
2217.7
9.4
2005
2402.7
25.9
190.7
2402.7
17.4
After correcting for price changes, it appears that empirical evidence suggests
the presence of a Laffer curve. However, we need to be careful.
50.00
40.00
30.00
20.00
10.00
0.00
-10.00
-20.00
% Change
%Change Real
Kennedy
1964
Recession
Reagan
1981
Bush
2001/2003
Expansion
100
90
80
70
60
50
Bottom Rate
Top rate
40
30
20
10
0
1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003
Each of these tax cuts was passed during a recession!
*Source: Congressional Budget Office
The Laffer effect of a tax cut should affect the trend, but not the cycle…
GDP
New Tax Code
Old Tax Code
This is what we have measured
This is what
we should be
measuring
Time
In other words, we have overstated the Laffer effect in the previous slides