Long-run evidence on the quantity theory of money

Evidence from low-frequency regressions for 27 countries since the XVIII century suggests that the relationship between broad money growth and inflation has been mostly one-for-one, and largely invariant to changes in the monetary regime. There is little evidence that the relationship had been weaker under commodity standards than it has been under fiat standards. Only for the period since the mid-1980s, which has seen the introduction of monetary regimes in which inflation is directly targeted, the relationship appears to have materially weakened. Crucially, however, the slope relationship between the trends of money growth and inflation produced by time-varying parameters VARs has been near-uniformly one-for-one for all countries and sample periods, including the one following the end of the Great Inflation. This suggests that, although central banks' targeting of inflation has weakened its relationship with money growth, time-series methods can still recover the one-for-one longhorizon relationship between the series. There is no evidence that, since WWII, inflation's low-frequency relationship with credit growth has been stronger than with money growth. The relationship between money growth and nominal interest rates had been non-existent under commodity standards, and it has only appeared under fiat standards.