1 Now Playing: The Biggest Hit in Economics: The Gross Domestic Product.
-
Upload
marshall-cummings -
Category
Documents
-
view
214 -
download
1
Transcript of 1 Now Playing: The Biggest Hit in Economics: The Gross Domestic Product.
What do these have in common?
• Real GDP• Consumer price index• Unemployment rate• Exchange rate of the dollar• Inflation rate• Real exchange rate
5
Answer….
They are all “indexes” that require some economic theory to construct.
Indeed, for most of human history (99.9%), we did not know how to construct them.
Understanding the construction of price and output indexes is our main analytical task today.
But first, some recent macro data….
6
9
-2
0
2
4
6
8
10
12
14
J-60
F-62
M-64
A-66
M-68
J-70
J-72
A-74
S-76
O-78
N-80
D-82
J-85
F-87
M-89
A-91
M-93
J-95
J-97
A-99
S-01
O-03
N-05
D-07
J-10
F-12
Inflation rate, price of personal consumption
An important inflation measure (corrected)
Fed target
Overview of national accounts
“While the GDP and the rest of the national income accounts may seem to be arcane concepts, they are truly among the great inventions of the twentieth century. Like a satellite that can view the weather across an entire continent, so the GDP can provide an overall picture of the state of the economy.”
A leading economics textbook.
10
11
Major concepts in national economic accounts
1. GDP measures final output of goods and services.2. Two ways of measuring GDP lead to identical results:
• Expenditure = income3. Savings = investment is an accounting identity.
• We will also see that it is an equilibrium condition.• Note the advanced version of this includes government
and foreign sector. 4. GDP v. GNP: differs by ownership of factors5. Constant v. current prices: correct for changing prices6. Value added: Total sales less purchases of intermediate
goods- Note that income-side GDP adds up value addeds
7. Net exports = exports – imports 8. Net v. gross investment:
• Net investment = gross investment minus deprecation
13
How to measure output growth?
Now take the following numerical example. • Suppose good 1 is computers and good 2 is shoes. • How would we measure total output and prices?
period 1 period 2
Ratio: period 2 to period 1
Real outputq1 1 100 100q2 1 1 1
Pricesp1 1 0.010 0.010p2 1 1.00 1.00
The growth picture for index numbers:the real numbers!
14
Source: Bureau of Economics Analysis
Output (billions 2005$)Sector 1960 2012Computers 0.0000337 87.94 Non computers 3,105.8 15,382.8
Price (2005 = 1)1960 2012
Computers 5,935.7 0.9006 Non computers 0.1749 1.0560
15
Some answers
• We want to construct a measure of real output, Q = f(q1,…,
qn ;p1,…, pn)
• How do we aggregate the qi to get total real, GDP(Q)?
– Old fashioned fixed weights: Calculate output using the prices of a given year, and then add up different sectors.
– New fangled chain weights: Use new “superlative” techniques
16
Old fashioned price and output indexes
Laspeyres (1871): weights with prices of base yearLt = ∑ wi,base year (Δq/q)i,t
Paasche (1874): use current (latest) prices as weights
Πt = ∑ wi,t (Δq/q)i,t
17
Start with Laspeyres and Paasche
HUGE difference!
What to do?
period 1 period 2
Ratio: period 2 to period 1
Real outputq1 1 100 100q2 1 1 1
Pricesp1 1 0.010 0.010p2 1 1.00 1.00
Nominal output
= ∑piqi 2.0 2.0 1.0Quantity indexes
Laspeyres (early p) 2.000 101.000 50.50Paasche (late p) 1.010 2.000 1.98
18
Solution
Brilliant idea: Ask how utility of output differs across different bundles.
How to implement: Let U(q1, q2) be the utility function. Assume have {qt} = {qt
1, qt2}. Then growth is:
g({qt}/{qt-1}) = U(qt)/U(qt-1).
For example, assume “Cobb-Douglas” utility function, Q = U = (q1)λ
(q2) 1- λ
Also, define the (logarithmic) growth rate of xt as g(xt) = ln(xt/xt-1). Then
Qt / Qt-1 =[(qt1)λ (qt
2) 1- λ]/[(qt-1
1)λ (qt-12)
1- λ]
g(Qt) = ln(Qt/Qt-1) = λ ln(qt1/qt-1
1) + (1-λ) ln(qt2/qt-1
2)
g(Qt) = λ g(qt1) + (1-λ) g(qt
2)
The class of 2nd order approximations is called “superlative.”This is a superlative index called the Törnqvist index.
18
19
What do we find?
1. L > Util > P [that is, Laspeyres overstates growth and Paasche understates relative to true.
period 1 period 2
Ratio: period 2 to period 1
Real outputq1 1 100 100q2 1 1 1
Pricesp1 1 0.010 0.010p2 1 1.00 1.00
Nominal output
= ∑piqi 2.0 2.0 1.0
Utility = (q1*q2)̂ .5 1.00 10.00 10.00
Quantity indexes
Laspeyres (early p) 2.000 101.000 50.50Paasche (late p) 1.010 2.000 1.98
20
Currently used “superlative” indexes
Fisher* Ideal (1922): geometric mean of L and P:Ft = (Lt × Πt )½
Törnqvist (1936): average geometric growth rate:
(ΔQ/Q)t = ∑ si,T (Δq/q)i,t, where si,T =average nominal share
of industry in 2 periods
(*Irving Fisher (YC 1888), America’s greatest macroeconomist)
21
Now we construct new indexes as above: Fisher and Törnqvist
Superlatives (here Fisher and Törnqvist) are exactly correct.
Usually very close to true.
period 1 period 2
Ratio: period 2 to period 1
Real outputq1 1 100 100q2 1 1 1
Pricesp1 1 0.010 0.010p2 1 1.00 1.00
Nominal output = ∑piqi 2.0 2.0 1.0
Utility = (q1*q2)̂ .5 1.00 10.00 10.00
Quantity indexes
Fisher (geo mean of L and P) 1.421 14.213 10.00
Törnqvist (wt. average growth rate) 1.000 10.000 10.00
22
Current approaches
• Most national accounts used Laspeyres until recently– Why Laspeyres? Primarily because the data
requirements are less stringent.• CPI uses Laspeyres index (sub-par approach!). • US moved to Fisher for national accounts in 1995• BLS has constructed “chained CPI” using Törnqvist since
2002• China still uses Laspeyres in its GDP.
– Who knows whether Chinese data are accurate?