Historical Statistics and Monetary Policy

1
Historical Statistics and Monetary Policy

• Pierre L. Siklos
• Wilfrid Laurier University and
• Viessmann Research Centre

2
What are we trying to analyze, and why?

• High Frequencies gt business cycle (short-run
• Noise in data is expected. Quality may matter
  less than econometric technique used to interpret
  data
• WHY?
• Focus is more on the role of expectations and
  markets and the pure reaction to information

3
What are we trying to analyze, and why?

• Middle Frequencies ? business cycle
• Quarterly or monthly data will do and their
  regimes may come into play. Quality may be
  important but consistency of definitions for
  variables becomes relatively more important
• WHY?
• Regimes are less frequent but policy
  reactions/changes are possibly more frequent

4
What are we trying to analyze, and why?

• Low Frequencies lt business cycle (long-run)
• Annual data are OK but Quality of data becomes
  paramount because measurement of variables can
  change significantly across time
• WHY?
• Rare events so we need a long span of data,
  statistical technique requires a long span of
  data, we want to know about changes in regimes

5
How Should Historical Series be Put Together?

• Is consistency more important than span of data?
  Is there a Trade-off?
• May be easier for some time series than for
  others (e.g., GDP VS money supply)
• Is there a danger that increasing span of data
  makes it easier to rewrite history?
• Construction may be by back casting VS from the
  ground up (e.g., price data)

6
What Can We Learn, what are the pitfalls of
analyzing monetary policy with historical Data?

• Example 1 Siklos (1993) 2 slides
• Example 2 Siklos-Burdekin (1998) 2slides
• Example 3 Romer 1 slide
• Example 4 Monetary Policy Rules and the Output
  Gap (Burdekin and Siklos 2004)
• Example 5 Siklos (2004?)

7
(No Transcript)
8
(No Transcript)
9
The "Phillips" Curve by Type of Income Measure
10
The "Phillips" Curve by Type of Unemployment
Measure
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
What Is the State of Play in the Use of
Historical Statistics? Selected Examples
17
Future Research?

• Part I Methodology How Historical Data are
  Used or Misused? A Meta-Analysis?
• Part II Econometric considerations How to test
  theories while remaining sensitive to data
  construction issues