Empirical data suggest that new rms tend to grow faster than incumbent firms in terms of their productivity. A sticky-price model with learning-by-doing in new firms fits this data and predicts that for plausible calibrations, the optimal long-run inflation rate is positive and between 0.5% and 1.5% per year. A positive long-run inflation rate helps the fast-growing new firms to align their real price with their idiosyncratic productivity growth. In contrast, the standard sticky-price model without learning-by-doing in new firms predicts an optimal long-run inflation rate near zero. In a two-sector model with learning-by-doing in new firms, the policy tradeoff that arises between new and incumbent firms is considerably more severe than the policy tradeoff that arises between economic sectors.