We examine the implications for monetary policy design of including le
arning-by-doing effects in a macroeconomic model. We show that an infl
ation bias arises because monetary surprises may be exploited to maxim
ise potential output by temporarily raising the rate of human capital
accumulation. Our model also provides an alternative explanation for t
he empirical evidence linking inflation and growth, where the causal l
ink goes from slow growth to high inflation. Unlike traditional credib
ility models, an inflationary bias can persist even when the authoriti
es do not wish to offset labour market distortions through monetary su
rprises which undercut the median voter's income.