Using 33 years of intraday Treasury data, we provide the first high-frequency evidence on auction-day price pressure: yields rise in the hours before auction and reverse afterward. This pressure strengthens when dealers face tighter risk-bearing constraints and weakens when investor demand is strong or more elastic. Net order flow explains much of the price pressure, offering direct evidence that secondary market trading transmits dealer constraints into prices. Despite heightened concerns about dealer capacity amid rapidly growing federal debt, price pressure has not increased in recent years in part because investment funds and foreign investors now absorb more auction supply, thus easing dealers’ intermediation burden and reducing price frictions.

About 88% of mutual-fund assets are held by retail investors, who rarely take short positions because of the associated costs and risks. Yet mutual-fund performance is traditionally measured by alpha, which implicitly assumes investors can freely short the benchmark factors. We show that mutual-fund performance for constrained investors is measured by achievable alpha, computed using only factors with strictly positive weights in the shortsale-constrained benchmark portfolio. Empirically, achievable alpha and value-added reveal weaker absolute performance and starkly different rankings. Achievable alphas predict fund flows---especially during market turmoil---and indicate that funds are less scalable than implied by traditional alphas.

We document a striking fact: While gross trading volume has quintupled over the past four decades, net volume -- trading from persistent portfolio reallocations, which excludes transitory round-trip trades -- has remained largely unchanged. We argue that the increase in transitory round-trip trades comes primarily from high-frequency intermediation, which has improved high-frequency liquidity but has not materially changed the market liquidity over the long term. Thus, while markets have become considerably more liquid at short horizons, long-term liquidity measures, such as the quarterly price impact and mean reversion, have remained largely unchanged over time. Using various instruments, we further demonstrate that net volume is the appropriate measure of liquidity at a given horizon and that it substantially outperforms gross-volume-based liquidity measures in explaining both the cross-section and time-series of long-term price impact and expected returns.