The Name of the Rose

1
The Name of the Rose

• Classifying the International Monetary System

Scott UrbanSt Antonys College, Oxford
2
Motivation

• Ever since the United States suspended the
  convertibility of the dollar into gold in 1971,
  the major currencies have once again been
  permitted to fluctuate against one another much
  as they did in the 1930s.
• Golden Fetters, 395

3
Motivation

• because in the 1970s the world has embarked on
  a general float of exchange rates, I thought it
  might be interesting to know more about the last
  period when most exchange rates were unpegged.
• Drummond, The Floating Pound, 1

4
Motivation

• In the interwar period, the return to the gold
  standard was short-lived, ending with the Great
  Depression. The return was preceded by
  widespread floating as was the period following
  it.
• Bordo 2003, 6

5
Exchange-rate regime classificationorigins

• Bretton Woods (post-WW2)'par values' (official
  pegs to US)
• Nixon closes gold window (1971)second amendment
  (1976) 'exchange arrangements'
• De jure classifications
• Calvo and Reinhart (2000) 'Fear of
  Floating''Everybody pegs but nobody admits it!'

6
Exchange-rate regime classificationorigins

• Bretton Woods (post-WW2)'par values' (official
  pegs to US)
• Nixon closes gold window (1971)second amendment
  (1976) 'exchange arrangements'
• De jure classifications
• Calvo and Reinhart (2000) 'Fear of
  Floating''Everybody pegs but nobody admits it!'

7
Exchange-rate regime classificationorigins

• Bretton Woods (post-WW2)'par values' (official
  pegs to US)
• Nixon closes gold window (1971)second amendment
  (1976) 'exchange arrangements'
• De jure classifications
• Calvo and Reinhart (2000) 'Fear of
  Floating''Everybody pegs but nobody admits it!'

8
Exchange-rate regime classificationorigins

• Bretton Woods (post-WW2)'par values' (official
  pegs to US)
• Nixon closes gold window (1971)second amendment
  (1976) 'exchange arrangements'
• De jure classifications
• Calvo and Reinhart (2000) 'Fear of
  Floating''Everybody pegs but nobody admits it!'

9
Exchange-rate regime classificationapproaches

• Inference from policy reserves

10
Exchange-rate regime classificationapproaches

• Inference from policy reserves

11
Exchange-rate regime classificationapproaches

• Inference from policy reserves

12
Exchange-rate regime classificationapproaches

• Inference from policy reserves

13
Exchange-rate regime classificationapproaches

• Inference from policy reserves

14
Exchange-rate regime classificationapproaches

• Inference from policy reserves
• Reliability of reserves data (Thailand 1997)
• Valuation changes
• Management of reserve portfolio
• Interest-rate policy

15
Exchange-rate regime classificationapproaches

• Inference from policy reserves
• Reliability of reserves data (Thailand 1997)
• Valuation changes
• Management of reserve portfolio
• Interest-rate policy

16
Exchange-rate regime classificationapproaches

• Inference from policy reserves
• Reliability of reserves data (Thailand 1997)
• Valuation changes
• Management of reserve portfolio
• Interest-rate policy

17
Exchange-rate regime classificationapproaches

• Inference from policy reserves
• Reliability of reserves data (Thailand 1997)
• Valuation changes
• Management of reserve portfolio
• Interest-rate policy

18
Exchange-rate regime classificationapproaches

• Inference from policy reserves
• Reliability of reserves data (Thailand 1997)
• Valuation changes
• Management of reserve portfolio
• Interest-rate policy

19
Exchange-rate regime classificationapproaches

• Inference from outcome exchange-rate
• Credible floats enjoy incredible stability!?
  'false positive' for a peg
• Peg-change ? high variance statistic
• In other words devaluation ? float!

20
Exchange-rate regime classificationapproaches

• Inference from outcome exchange-rate
• Credible floats enjoy incredible stability!?
  'false positive' for a peg
• Peg-change ? high variance statistic
• In other words devaluation ? float!

21
Exchange-rate regime classificationapproaches

• Inference from outcome exchange-rate
• Credible floats enjoy incredible stability!?
  'false positive' for a peg
• Peg-change ? high variance statistic
• In other words devaluation ? float!

22
Exchange-rate regime classificationapproaches

• Inference from outcome exchange-rate
• Credible floats enjoy incredible stability!?
  'false positive' for a peg
• Peg-change ? high variance statistic
• In other words devaluation ? float!

23
Exchange-rate regime classificationapproaches

• Inference from outcome exchange-rate
• Credible floats enjoy incredible stability!?
  'false positive' for a peg
• Peg-change ? high variance statistic
• In other words devaluation ? float!

24
Exchange-rate regime classification'flexibility'
indices
CR2000 ('Fear of floating') ? s2e / (s2i
s2F)e is change in exchange rate i is
discount rate (level) F is change in
reserves The index is a variance ratio
25
Exchange-rate regime classification'flexibility'
indices
Poirson 2001 ('FLT index')
ratio of change in exchange-rate change in
reserves, scaled by M0
26
Data1930s

• League of Nations, Monthly Bulletin of Statistics
  (various issues)
• League of Nations, Statistical Year-book (various
  issues)
• League of Nations, Review of World Trade 1938
• Economist
• Bank of England archive 'Red books'

27
Data1957-2006

• IMF International Financial Statistics
• National authorities
• OECD

28
Quality check reserve statistics
29
Quality check reserve statistics
30
Quality check reserve statistics
31
Quality check currency in circulation
32
Quality check reserve statistics
33
Quality check reserve statistics
34
Results
35
(No Transcript)
36
Red N60Grey N25
37
Mean value of flexibility index per period
38
Poirson index
39
Poirson 2001 index modern period --------------
-----------------------------------------------
Percentiles Smallest 1 0
0 5 .0028324 0 10
.0197399 0 Obs
146 25 .1354935 0 Sum
of Wgt. 146 50 .349629
Mean 4.524667
Largest Std. Dev. 19.5573 75
.9843689 61.14288 90 5.14071
69.94834 Variance 382.488 95
17.59775 110.321 Skewness
6.966895 99 110.321 184.4557
Kurtosis 57.17034
40
Poirson 2001 index modern period --------------
-----------------------------------------------
Percentiles Smallest 1 0
0 5 .0028324 0 10
.0197399 0 Obs
146 25 .1354935 0 Sum
of Wgt. 146 50 .349629
Mean 4.524667
Largest Std. Dev. 19.5573 75
.9843689 61.14288 90 5.14071
69.94834 Variance 382.488 95
17.59775 110.321 Skewness
6.966895 99 110.321 184.4557
Kurtosis 57.17034
41
Poirson 2001 index 1930s --------------------
-----------------------------------------
Percentiles Smallest 1 .000038
.0000283 5 .0000788 .000038 10
.0002857 .0000471 Obs
115 25 .0025916 .0000607 Sum of
Wgt. 115 50 .0285736
Mean 9.540972
Largest Std. Dev. 39.47082 75
.5425414 129.1733 90 4.833831
177.5537 Variance 1557.946 95
76.9855 183.2627 Skewness
5.087103 99 183.2627 291.6033
Kurtosis 30.55277
42
Poirson 2001 index 1930s --------------------
-----------------------------------------
Percentiles Smallest 1 .000038
.0000283 5 .0000788 .000038 10
.0002857 .0000471 Obs
115 25 .0025916 .0000607 Sum of
Wgt. 115 50 .0285736
Mean 9.540972
Largest Std. Dev. 39.47082 75
.5425414 129.1733 90 4.833831
177.5537 Variance 1557.946 95
76.9855 183.2627 Skewness
5.087103 99 183.2627 291.6033
Kurtosis 30.55277
43
Poirson 2001 index modern period --------------
-----------------------------------------------
Percentiles Smallest 1 0
0 5 .0028324 0 10
.0197399 0 Obs
146 25 .1354935 0 Sum
of Wgt. 146 50 .349629
Mean 4.524667
Largest Std. Dev. 19.5573 75
.9843689 61.14288 90 5.14071
69.94834 Variance 382.488 95
17.59775 110.321 Skewness
6.966895 99 110.321 184.4557
Kurtosis 57.17034
44
Poirson 2001 index 1930s --------------------
-----------------------------------------
Percentiles Smallest 1 .000038
.0000283 5 .0000788 .000038 10
.0002857 .0000471 Obs
115 25 .0025916 .0000607 Sum of
Wgt. 115 50 .0285736
Mean 9.540972
Largest Std. Dev. 39.47082 75
.5425414 129.1733 90 4.833831
177.5537 Variance 1557.946 95
76.9855 183.2627 Skewness
5.087103 99 183.2627 291.6033
Kurtosis 30.55277
45
Poirson 2001 index 1930s --------------------
-----------------------------------------
Percentiles Smallest 1 .000038
.0000283 5 .0000788 .000038 10
.0002857 .0000471 Obs
115 25 .0025916 .0000607 Sum of
Wgt. 115 50 .0285736
Mean 9.540972
Largest Std. Dev. 39.47082 75
.5425414 129.1733 90 4.833831
177.5537 Variance 1557.946 95
76.9855 183.2627 Skewness
5.087103 99 183.2627 291.6033
Kurtosis 30.55277
46
Poirson 2001 index modern period --------------
-----------------------------------------------
Percentiles Smallest 1 0
0 5 .0028324 0 10
.0197399 0 Obs
146 25 .1354935 0 Sum
of Wgt. 146 50 .349629
Mean 4.524667
Largest Std. Dev. 19.5573 75
.9843689 61.14288 90 5.14071
69.94834 Variance 382.488 95
17.59775 110.321 Skewness
6.966895 99 110.321 184.4557
Kurtosis 57.17034
47
CR2000 index'Fear of floating'
48
'Fear of floating' index modern
period -------------------------------------------
------------------ Percentiles
Smallest 1 0 0 5
.0000359 0 10 .0005422
0 Obs 161 25
.1026475 0 Sum of Wgt.
161 50 .4960451 Mean
1.248075 Largest
Std. Dev. 2.886232 75 .9984185
11.25288 90 2.552471 13.38458
Variance 8.330333 95 4.309496
14.2846 Skewness 5.417827 99
14.2846 25.63474 Kurtosis
38.58661
49
'Fear of floating' index modern
period -------------------------------------------
------------------ Percentiles
Smallest 1 0 0 5
.0000359 0 10 .0005422
0 Obs 161 25
.1026475 0 Sum of Wgt.
161 50 .4960451 Mean
1.248075 Largest
Std. Dev. 2.886232 75 .9984185
11.25288 90 2.552471 13.38458
Variance 8.330333 95 4.309496
14.2846 Skewness 5.417827 99
14.2846 25.63474 Kurtosis
38.58661
50
'Fear of floating' index 1930s -------------
------------------------------------------------
Percentiles Smallest 1 0
0 5 1.75e-08 0 10
3.47e-06 0 Obs
137 25 .0007691 1.86e-10 Sum
of Wgt. 137 50 .0250113
Mean 10.45745
Largest Std. Dev. 55.30851 75
.5997505 83.59766 90 5.255409
203.6352 Variance 3059.031 95
44.54796 231.807 Skewness
8.200124 99 231.807 563.8833
Kurtosis 77.28336
51
'Fear of floating' index 1930s -------------
------------------------------------------------
Percentiles Smallest 1 0
0 5 1.75e-08 0 10
3.47e-06 0 Obs
137 25 .0007691 1.86e-10 Sum
of Wgt. 137 50 .0250113
Mean 10.45745
Largest Std. Dev. 55.30851 75
.5997505 83.59766 90 5.255409
203.6352 Variance 3059.031 95
44.54796 231.807 Skewness
8.200124 99 231.807 563.8833
Kurtosis 77.28336
52
'Fear of floating' index modern
period -------------------------------------------
------------------ Percentiles
Smallest 1 0 0 5
.0000359 0 10 .0005422
0 Obs 161 25
.1026475 0 Sum of Wgt.
161 50 .4960451 Mean
1.248075 Largest
Std. Dev. 2.886232 75 .9984185
11.25288 90 2.552471 13.38458
Variance 8.330333 95 4.309496
14.2846 Skewness 5.417827 99
14.2846 25.63474 Kurtosis
38.58661
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
Exchange-rate-only approachesReinhart Rogoff
2002Shambaugh 2004
57
RR2002 approach

• What is the probability that the exchange rate
  moves within a 2 band over any 12-month period?
• (Shambaugh) Allow for one breach during period
  (to address 'false positive' issue)

58
RR2002 approach

• What is the probability that the exchange rate
  moves within a 2 band over any 12-month period?
• (Shambaugh) Allow for one breach during period
  (to address 'false positive' issue)

59
Modern period
60
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
61
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
62
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
63
1930s
64
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
65
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
66
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
67
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
68
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
69
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
70
PR (exchange rate chg gt 2) rolling 12-month
sample using relevant bilateral rate, not
counting peg changes
71
Next Steps

• Third dimension for classificationTime-series
  properties of the exchange rate
• Market-set price (i.e. a 'float') should be an
  AR(1) process ('random walk')
• Regress er on er_lag, use coefficient as index
• Even better

72
Next Steps

• Third dimension for classificationTime-series
  properties of the exchange rate
• Market-set price (i.e. a 'float') should be an
  AR(1) process ('random walk')
• Regress er on er_lag, use coefficient as index
• Even better

73
Next Steps

• Third dimension for classificationTime-series
  properties of the exchange rate
• Market-set price (i.e. a 'float') should be an
  AR(1) process ('random walk')
• Regress er on er_lag, use coefficient as index
• Even better

74
Next Steps

• Third dimension for classificationTime-series
  properties of the exchange rate
• Market-set price (i.e. a 'float') should be an
  AR(1) process ('random walk')
• Regress er on er_lag, use coefficient as index
• Even better

75
Next Steps

• Third dimension for classificationTime-series
  properties of the exchange rate
• Market-set price (i.e. a 'float') should be an
  AR(1) process ('random walk')
• Regress er on er_lag, use coefficient as index
• Even better

76
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

77
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

78
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

79
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

80
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

81
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

82
Next Steps

• Rich econometrics literature to test for presence
  of AR(1) (i.e. unit root)
• Dickey-Fuller test H0 series contains unit root
• KPSS 1992 H0 series is stationary
• Difficult
• Tests severely affected by specification, e.g.
  number of lag
• Rolling regressions produce test statistic with
  little seeming correspondence to regime type

83
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

84
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

85
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

86
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

87
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

88
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

89
Next Steps

• Develop an algorithm for classification for the
  world monetary system as a whole
• Take account of PPPq EP/P
• In face of a shockdo domestic prices adjust,
  or exchange rates?
• 1930s would look like Nineteenth century

90
The Name of the Rose

• "It is remarkable how the worlds short history
  of floating exchange rates has affected popular
  thinking about what is eternally normal and
  proper in the economic system. Floating exchange
  rates have proved a source of tremendous periodic
  instability, yielding repeated currency
  crises."
• Benn Steil, director of international economics
  Council on Foreign Relations

91
The Name of the Rose

• Ferocious debate afoot over the 'international
  financial architecture' and 'global imbalances'
• 1930s cited as prima facie evidence of
  instability of floating systems
• The name we attached to the international
  monetary system in the 1930s does matter
• 1930s should be seen as evidence in the case for
  floating currencies

92
The Name of the Rose

• Ferocious debate afoot over the 'international
  financial architecture' and 'global imbalances'
• 1930s cited as prima facie evidence of
  instability of floating systems
• The name we attached to the international
  monetary system in the 1930s does matter
• 1930s should be seen as evidence in the case for
  floating currencies

93
The Name of the Rose

• Ferocious debate afoot over the 'international
  financial architecture' and 'global imbalances'
• 1930s cited as prima facie evidence of
  instability of floating systems
• The name we attached to the international
  monetary system in the 1930s does matter
• 1930s should be seen as evidence in the case for
  floating currencies

94
The Name of the Rose

• Ferocious debate afoot over the 'international
  financial architecture' and 'global imbalances'
• 1930s cited as prima facie evidence of
  instability of floating systems
• The name we attached to the international
  monetary system in the 1930s does matter
• 1930s should be seen as evidence in the case for
  floating currencies

95
The Name of the Rose

• Ferocious debate afoot over the 'international
  financial architecture' and 'global imbalances'
• 1930s cited as prima facie evidence of
  instability of floating systems
• The name we attached to the international
  monetary system in the 1930s does matter
• 1930s should be seen as evidence in the case for
  floating currencies

96
Thank You