Transcript Lecture VI Real Business Cycle Models

Lecture VI Real Business Cycle and Ed Prescott
What are RBC models?
• Macroeconomic models in which business
fluctuations to a large extent can be
accounted for by real rather than nominal
shocks
• RBC models assume the economy
remains at general equilibrium rather than
disequilibrium, as is the case for
Keynesian and monetarist models
Kydland and Prescott
• The theory was introduced in Kydland and
Prescott’s Time to Build And Aggregate
Fluctuations
• They envisioned the main source of business
cycle fluctuations are supply shocks, which
could then have long-lasting residual effects
• These supply shocks are changes to productivity
that may derive from technology shifts like
innovations, but may also derive from regulatory
changes, changes in input prices, weather
variability, etc.
Isolating trend from cycle
• The first step in the RBC process is to try to
decompose a macroeconomic series into its two
main components the secular trend and a
cyclical component
• Because a series may change its trend due to
technology or other supply shock, we cannot
assume the series’ trend will remain stable
• The methodology employed is a filter; the most
common one, designed by Prescott and Hodrick
is the Hodrick-Prescott or HP filter
LN Real US GDP 1947-2010 Q3
Ln Real GDP
9.7
9.2
8.7
8.2
7.7
7.2
Jan47
Jan52
Jan57
Jan62
Jan67
Jan72
Jan77
Jan82
Jan87
Jan92
Jan97
Jan02
Jan07
US GDP data Difference of Logs
DiffLnGDP US 1947-2010 Q3
0.04
0.03
First difference
0.02
0.01
0
7
-4
an
J
-0.01
Ja
52
n-
57
nJa
62
nJa
Ja
67
n-
Ja
72
n-
77
n-
Ja
Ja
-0.02
-0.03
-0.04
Period
82
n-
Ja
87
n-
n-
Ja
92
97
n-
Ja
02
n-
Ja
07
n-
Ja
Correlation in US GDP first
differenced data
• The correlation between DiffLnGDP and
its lagged value was 0.368363; r = 0
means that the next value has an equal
chance of being above or below
regardless of this period’s value and r = 1
means that the next value will definitely
be on the same side of 0
Filtering the GDP data
• When you look at the US first differenced data
you will note the periods that are mode
correlated; these are the long runs that are
above or below the long-term trend
• A filter will adjust the “level” by shifting upward of
downward from the 0 line to try to even out the
observations that are above and below the line
• In other words, the 0 line is adjusting to fit the
differenced data
• The part that is left is the “white noise” which is
uncorrelated with its lagged values and the trend
That’s about all I know
• In order to actually perform RBC model building,
you need to have a H-P filter routine, which can
be found on the internet
• The you need a model of the economy, which
also requires lots of data – good data
• You then need a set of estimate coefficients –
some of these apparently are considered more
reliable than others; you can tell the program
how much confience you have in the parameters
Run and Shock the Model
• Then you need software that can run the model,
add “shocks” to the various sectors
• You use Monte Carlo random generators to
generate these shocks
• You then observe how the model behaves under
this “stress test” environment
• You see if you observe the magnitude and
direction we observe in the real world as well as
the comovements of the various sectors
What are they looking for?
• They look at the following sectors: output, consumption,
investment, labor hours and productivity and capital stock
• Expected correlations and standard deviations are as
follows:
Standard Correlation
Sector
Deviation with Output
Output
1.76
1.00
Consumption
1.29
0.85
Investment
8.60
0.92
Capital Stock
0.63
0.04
Labor
1.66
0.76
Productivity
1.18
0.42
What can they not do?
• The shocks that are generated are of the
magnitude they suppose actually affect the
economy
• They are not intended to mimic actual shocks
because these are not observable
• They merely show that a general equilibrium
model subject to rand shocks can generate an
economy that looks like our economy; in other
words it is “well-behaved” and doesn’t (usually)
run of somewhere
• But when it does, you got a problem!
Some very unfortunate news
• Occasions such as the recent financial crisis, not to
mention the other “big one” 80 years ago do challenge
the notion that most business cycles and most of the
volatility in one are due to real and not nominal shocks
• Same for the early 1980’s recession; both RET and RBC
models don’t really like to face such realities
• As I have said before and will maintain for as long as I
live and breath, business cycles are not of any one type
• Understanding this fact is essential to understanding
how the economy works, or doesn’t work, and each
cycle should be explained based on our intellect and not
a fixed idea about them
Critiques
• Summers, Hoover, Sheffrin, Mankiw,
McCallum, Phelps, Eichenbaum, Stadler,
and Hartley, et. al. all had various
criticisms of New Classical theory
Critique 1
• The relative size of the substitution effect
versus the income effect implied by RBC
models is much greater than those found from
microeconomic studies of households, who
tend to maintain a rather steady amount of
effort
Critique 2
• Too much reliance on unobservable supply
shocks especially in recessions
• Critics regard the amount of shocks
necessary to obtain the kind of volatility
observed is “quite implausible”
(Muellbauer, 1997)
Critique 3
• Recessions as periods of “technological regress”
• RBC folks argue that, by technology shocks,
they include regulations that can reduce
productivity
• Muellbauer points to UK early 1990s which
clearly suffered from non-RBC factors: massive
interest rate increases, overvalued currency
while entering ERM, and a collapse in property
values, all non-RBC variables
Critique 4
• The idea that all unemployment is
voluntary seem totally implausible,
especially during the Great Depression
• Also quit rates are pro-cyclical rather than
counter-cyclical, if increases in
unemployment, which is counter-cyclical,
were truly voluntary
• And there are others