How Quickly Do Financial Markets React To News?

It is not news that news moves financial markets. This blog will publish research on how, when, why, and which news moves what financial markets.

Most researchers to data have taken a fixed window over which to measure the announcement effect. An exception is Kim and Sheen (2001) who calculate minute-by-minute returns for the Australian bond futures market. The size of the window is a testable hypothesis so I ran regressions using cumulative returns starting with the close one minute before an announcement up to half an hour after the announcement. As an example, for non-farm payrolls, the ability of news to explain movements in the 5-Year Treasury Bond futures contract can be improved by 6% by changing the definition of the window from the 30 minutes (from 8:25am, 5 minutes before the announcement to 8:55am, as used by Gürkaynak and Wolfers (2006)) traditionally used in the academic literature to a 20-minute window (from 8:25am to 8:45am). The same change for the EUR/USD exchange rate results in a 25% improvement in explanatory power. For the S&P 500 futures contract, a movement from a 30-minute window to a 10-minute window yields a 37% improvement.

For FX markets, in general, I have found that shortening the window even more works well. A minute or two is when the best relationship is found for most foreign exchange rate pairs.

	Interval				R2 improvement
	Length				over 30-minute
	(minutes)	Coefficient	t-Statistic	R2	window
5Year Bond	20	-0.00371	-7.983	0.589563	6%
	10	-0.00359	-7.865	0.583629	5%
	25	-0.00365	-7.629	0.566652	2%
	30	-0.00357	-7.328	0.55454	0%
	15	-0.00339	-7.178	0.538909	-3%
EUR	20	-0.00307	-6.065	0.456968	25%
	15	-0.00307	-6.065	0.456968	25%
	10	-0.00307	-6.065	0.456968	25%
	25	-0.00338	-5.04	0.373143	2%
	30	-0.00338	-4.998	0.366464	0%
S&P 500	10	0.002926	5.656	0.441865	37%
	20	0.002694	5.081	0.375972	17%
	15	0.002426	4.909	0.353967	10%
	25	0.002729	4.66	0.334129	4%
	30	0.002825	4.524	0.321928	0%