We use recently proposed tests to extract jumps and cojumps from three types of assets: stock index futures, bond futures, and exchange rates. We then characterize the dynamics of these discontinuities and informally relate them to US macroeconomic releases before using limited dependent variable models to formally model how news surprises explain (co)jumps. Nonfarm payroll and federal funds target announcements are the most important news across asset classes. Trade balance shocks are important for foreign exchange jumps. We relate the size, frequency and timing of jumps across asset classes to the likely sources of shocks and the relation of asset prices to fundamentals in the respective classes. JUMPS, COJUMPS AND MACRO ANNOUNCEMENTS 895 Andersen et al. (2007c) and Lee and Mykland (2008) assume a continuous time jump-diffusion data-generating process. The log price process evolves as follows: First Step Towards Intraday Jumps Detectionwhere p t) is a log asset price, W t) is a standard Brownian motion, q t) is a counting process, possibly a non-homogeneous Poisson process, independent of W t), and Ä t D p t p t is the jump size. The Brownian motion, W t), jump sizes, Ä t , and the counting process, q t) are independent of each other. In the absence of jumps, the drift t and instantaneous volatility t are such that the underlying data-generating process is an Itô process with continuous sample paths. The drift and diffusion coefficients may not change dramatically over short periods of time.The intuition behind the jump test proposed simultaneously by Andersen et al. (2007c) and Lee and Mykland (2008) is straightforward: In the absence of jumps, instantaneous returns are increments of Brownian motion. Standardized returns that are too large to plausibly come from a standard Brownian motion must reflect jumps. 1 More formally, assume we have T days of b1/c Á M equally spaced intraday returns and denote the ith return of day t by Andersen et al. (2007c) and Lee and Mykland (2008) propose the following test statistic for jumps in r t,i :One must estimate the unobserved volatility, t,i , with a robust-to-jumps estimator. Shephard (2004, 2006a) show that, under weak conditions, realized bipower variation (RBV) converges to integrated volatility under the model described by equation (1): Consequently, Andersen et al. (2007c) and Lee and Mykland (2008) propose estimating 2 t,i as the average of the RBV computed over a local window of K observations preceding period t, i. They both explicitly assume that the spot volatility is approximately constant over that window. There is a clear trade-off in choosing the window size K: K must be large enough to accurately estimate integrated volatility but small enough for variance to be approximately constant. For returns sampled at frequencies of 60, 30, 15, and 5 minutes, Lee and Mykland (2008) recommend using K D 78, 110, 156 and 270 observations, respectively.Under the null of no jumps, the test statistic, J t,i , follows the same distribution as the absolute value of a s...

SUMMARYWe use recently proposed tests to extract jumps and cojumps from three types of assets: stock index futures, bond futures, and exchange rates. We then characterize the dynamics of these discontinuities and informally relate them to US macroeconomic releases before using limited dependent variable models to formally model how news surprises explain (co)jumps. Nonfarm payroll and federal funds target announcements are the most important news across asset classes. Trade balance shocks are important for foreign exchange jumps. We relate the size, frequency and timing of jumps across asset classes to the likely sources of shocks and the relation of asset prices to fundamentals in the respective classes.

We analyze the relationship between interventions and volatility at daily and intra-daily frequencies for the two major exchange rate markets. Using recent econometric methods to estimate realized volatility, we employ bipower variation to decompose this volatility into a continuously varying and jump component. Analysis of the timing and direction of jumps and interventions imply that coordinated interventions tend to cause few, but large jumps.Most coordinated operations explain, statistically, an increase in the persistent (continuous) part of exchange rate volatility. This correlation is even stronger on days with jumps.