MACROECONOMETRICS

1
MACROECONOMETRICS

• LAB 6 CALIBRATION

2
ROADMAP

• What is the reality of business cycles?
• What are the business cycles?
• How can we try to model them
• Are we happy with ourselves?
• Nothing about STATA ?

3
Business cycles US data
YQ of peak No. of Q to recovery GDP change (to average)
19484 4 -1.1
19532 4 -3.1
19573 2 -2.2
19601 4 -0.8
19693 3 -0.9
19734 5 -4.1
19801 1 -1
19813 4 -2.8
19902 3 -1.6
4
Business cycles US data
Component of GDP Average share in GDP () Share of GDP during recession ()
Consumption (nondurables) 24.8 14,3
Consumption (durables) 6.9 6,7
Investment (in equipment) 10 21
Investment (residential structures) 5.1 14.7
Inventories 0.6 30,7
!!!
5
Business cycles US data
Variable Average change in recession ()
Real GDP -4.7
Employment -2.2
Average no. of hours (per week) -0.9
Productivity (output per worker) -1.4
Wages -0.5
Unemployment 2,1
6
Business cycles conclusions

• No regularity
• On average 5, but sometimes 9 and sometimes 1
• Uneven distribution
• From 1 do 5 quarters, longest arent hardest
• Nothing deterministic
• Kondratiev and others no applicability
• Shocks
• Seem to be random
• Have to have a mechanism to propagate over the
  economy

7
Business cycles - conclusions

• ANY (!) model has to reproduce
• Anti-cyclical unemployment
• Pro-cyclical employment, productivity AND
  slightly pro-cyclical wages (!)
• REMARK
• Neoclassical model has NO business cycles!
• Keynesian model WILL NOT WORK
• In Lucas we could, but assumptions not realistic

8
Kydland and Prescott (Minnesota)

• Cycles driven by technological shocks
• How do we get recessions? Sudden amnesia?

PROPAGATION-ACCELERATION MECHANISM

• Technological improvement in any sector
• How does it spread into whole economy?

GENERAL EQUILIBRIUM MODEL
9
Nature of the modelling

• Consumer equation
• IntERtemporal
• IntRAtemporal
• Producer equation
• Solow function works well, if A stochastic
• Market clearing condition
• Labour market
• Goods market
• Financial markets

10
New techniques and methods

• Separation variables
• State A, G, K optimal
• Decision L, C, I (gt w, r)
• Method
• Detrend
• Substract the deterministic trend
• Obtain a zero deterministic steady-state PROCESS
  (!)
• Analyse laws of motion

11
How well do we do?

• Method
• estimate moments on real economy
• simulate your model
• estimate moments on the modelled economy
• compare

12
How well do we do?
US real data Simulated models
sY 1.92 1.3
sG/sY 0.45 0.31
sI/sY 2.75 3.15
cov(L, Y/L) -0.14 0.93
sL/sY 0.96 0.49
13
How well do we do?
US real data Simulated models
sY 1.92 1.3
sG/sY 0.45 0.31
sI/sY 2.75 3.15
cov(L, Y/L) -0.14 0.93
sL/sY 0.96 0.49
14
How well do we do?
US real data Simulated models
sY 1.92 1.3
sG/sY 0.45 0.31
sI/sY 2.75 3.15
cov(L, Y/L) -0.14 0.93
sL/sY 0.96 0.49
15
How well do we do?
US real data Simulated models
sY 1.92 1.3
sG/sY 0.45 0.31
sI/sY 2.75 3.15
cov(L, Y/L) -0.14 0.93
sL/sY 0.96 0.49
16
How well do we do?
US real data Simulated models
sY 1.92 1.3
sG/sY 0.45 0.31
sI/sY 2.75 3.15
cov(L, Y/L) -0.14 0.93
sL/sY 0.96 0.49
17
How well do we do?
US real data Simulated models
sY 1.92 1.3
sG/sY 0.45 0.31
sI/sY 2.75 3.15
cov(L, Y/L) -0.14 0.93
sL/sY 0.96 0.49
!!!