The end of gold's official monetisation in 1971 was the penultimate nail in the coffin of an international monetary system based upon fixed exchange rates. With capitulation in subsequent efforts to turn back the clock, the world was ushered into unprecedented territory. The new monetary system was based on completely ‘fiat’ money and, more important, floating exchange rates among the leading industrialised economies. As Michael Klein and Jay Shambaugh point out, this system has lasted longer than the Bretton Woods system itself and deserves a name of its own (why not ‘Fiat Era’ or ‘Floating Era’?). Whatever its moniker, today's international monetary system is long overdue a thorough empirical treatment and Exchange Rate Regimes in the Modern Era is a masterful start.

Optimum currency area theory, the policy trilemma, and political economy are the frameworks for understanding the genesis of the variety of regime configurations attempted over the past 30-plus years, spanning a continuum from free floating to currency union. The authors' careful accounting of the multiple dimensions of regime type demonstrate intimate familiarity with the subtlety of exchange-rate regime taxonomy. Are the exchange rates of interest nominal or real? Are they effective or bilateral? If the latter, against what base currency?

The book is extremely well thought out and the progression of development is proof that the authors have been grappling with this topic for some time. The empirical sections first seek to understand why regimes are chosen before testing the impact of this choice on nominal and real outcomes such as inflation and growth. This works well because the results from testing for regime determinants assuage econometric concerns in the subsequent testing of regime performance. Time-varying factors are found to have little or no impact on regime choice; hence, tests of regime performance need not be plagued by endogeneity, which would be the case if the dependent variables such as inflation or growth are not only determined by the regime type but also influence the selection of regime type. To anticipate a key conclusion: the authors find little or no link between regime type and long-run growth, a result consistent with the earliest literature on regime performance in the modern age.Footnote ¹

The empirical chapters provide an exhaustive review of the literature on regime choice and performance, coupled with the authors' own econometric results. Their panel dataset for most of these estimations contains 125 countries and annual observations from 1973 to 2004, where the coding for regime type is usually binary (pegged/non-pegged). The reader needs to interpret the results independently. Are non-pegged regimes really so much more volatile than pegged regimes? The initial result says the peg reduces volatility by 5 percentage points compared to non-pegs, where volatility is the average annual standard deviation of the monthly percentage change in the exchange rate for a given country in a given year. If the largest 1 per cent of volatility outcomes and the largest 1 per cent of inflation outcomes are excluded, the difference drops to 3 percentage points. This is a modest difference considering the premium the economy must pay in loss of policy freedom – a trade-off well documented in Shambaugh's extensive work on the trilemma.

If volatility is only modestly reduced by pegging, and if the price is high in terms of independence, it becomes more difficult to remain agnostic on regime choice. The authors admirably do so, noting that no regime is right for all countries at all times. But this reviewer can be more bold. In more than half of the cases where pegs lasted five years or more before collapsing, the collapse produced volatility in the top 5 per cent of the dataset. This is consistent with Friedman's contention that pegs merely store up volatility.Footnote ² Partisanship is nearly irresistible in the fixed-versus-floating debate, and either side can make use of the insights unearthed here; the foregoing peg-collapse statistic is a good example. It comes from a chapter building on the authors' work on survival analysis of exchange-rate regimes, an illuminating contribution to the literature.

This is a book about regime choice and performance much more than regime classification. The latter still needs to be written. For all of the authors' contributions to our understanding of the vexatious issue of classification, the state of the art cries out for improvement. There are likely to be as many dimensions to the problems of classification methodologies as there are dimensions of regime type. One is coherence: different schemes disagree, as documented by the authors (who contend that different research questions demand different classification schemes). Another problem is researcher discretion. Because different schemes feature different numbers of regime category, the researcher's judgement is exercised in compacting categories into a common minimum (e.g. collapsing seven gradations into a binary peg/no-peg category). Subjectivity is also revealed in the various caveats that most classification algorithms employ to remedy nonsensical classifications.

‘False peg’ and ‘false float’ traps flummox these schemes: floating currencies can be incredibly stable (rendering them misclassified as pegs), while one-off devaluations in pegged regimes can produce a variation statistic which renders them misclassified as floats. The authors dismiss the former trap on the basis of Calvo and Reinhart,Footnote ³ who find a less than 3 per cent probability that benchmark floating regimes (Australia, Japan and Germany pre-EMU) actually register low volatility in any 12-month period (i.e. 2.5 per cent or less monthly change in the exchange rate) between when these currencies officially floated until the end of 1999. Yet Canada's regime in the same time period exhibited a 47 per cent probability of such quiescence. And if the threshold is 11 months out of any 12-month period (in the spirit of Shambaugh's own coding algorithm), Canada's rises to 74 per cent and Australia's to 14 per cent; Britain's is 17 per cent since 1992. These are non-trivial occurrences of misidentification, and are probably biased against the biggest economies since their currencies enjoy the highest trading volume. The root problem is perhaps that most schemes, including the ones employed here, are grounded in stylised facts. Ghosh et al. Footnote ⁴ remain correct that most classification algorithms ‘do not correspond closely with theoretic concepts’.

Still, the Shambaugh algorithm is the best of the bunch. The empirical questions tested with this dataset are exhaustive and undoubtedly reveal truths about the modern era. One lament is that the authors make no attempt to join these to a weighting scheme for the panel members (e.g. by GDP or total trade) to indicate a larger pattern in the international monetary system. Surely it matters more what the ascendant giants are doing than the minnows. Agnosticism notwithstanding, some countries' currency regimes are more equal than others.