A new paper estimated the short-term effects of U.S. economic data surprises on treasury notes and USD exchange rates over the past 20 years. All of 21 commonly followed data releases produced highly significant surprise effects at least for parts of the sample. However, only non-farm payrolls produced a consistently highly significant impact. After short-term interest rates reached the zero lower bound, the importance of surprises to CPI inflation, housing indicators and weekly jobless claims increased noticeably, possibly related to the Fed’s struggle with its dual mandate.

Koch, Christoffer and Julieta Yung (2016), “Macroeconomic News and Asset Prices Before and After the Zero Lower Bound”, Federal Reserve Bank of Dallas, Globalization and Monetary Policy Institute, Working Paper No. 287

The post ties in with the subject of information efficiency, particularly in respect to macro trends (view summary here).
Also on the relation of economic data releases with volatility and large price jumps view post here.

The below are excerpts from the paper. Headings and some other cursive text has been added for context and convenience of reading.

How to measure U.S. economic data surprises?

“We operationalize the concept of macroeconomic ‘news’ as the difference between an announcement and its expected value prior to the actual release of information. The market expectation of the macroeconomic announcement is captured by the median response from a survey of market participants from Bloomberg, L.P., which has to be submitted by midnight of the day leading up to the release. These ‘news’ can be interpreted as updates on the state of the U.S. economy, for which deviations from prior expectations reflect the unanticipated component that alters the information set of market participants.”

“The available data on the 21 U.S. macroeconomic announcements that we analyze span 1996 to 2016 and thus include eight years of the ZLB [interest rates at the zero lower bound]. They comprise real-indicators such as quarterly GDP releases, monthly Nonfarm Payrolls and Industrial Production, as well as ‘flash’ indicators based on household or businesses surveys such as Consumer Confidence or the ISM Manufacturing surveys.”

“We examine the response of 10-year futures prices and currency futures using intra-day data from TickData within a 15-minute symmetric window around the exact daytime of the macroeconomic release. We regress the log change in the price…The surprises are normalized to have a unit standard deviation and responses are in basis points to ease the interpretation of the results.”

Which U.S. data surprises matter most?

“In general, positive news tend to decrease futures prices and thus raise yields. Moreover, domestic interest rates tend to be more sensitive to U.S. macroeconomic news than exchange rates…The literature has previously found that the link between macroeconomic news and bond markets is simpler and stronger than foreign exchange markets.”

Non-farm Payrolls has been the only consistently producer of significant data surprises, insofar as it has been highly significant in respect to its short-term impact on both the 10-year treasury note and the EURUSD exchange rate, both before and after the fed funds rate reached the zero lower bound:

The following indicators have been highly significant in respect to their impact on the 10-year treasury note before and after the fed funds rate reached the zero lower bound:

  • Existing Home Sales and New Home Sales
  • Retail Sales
  • Consumer Confidence (Conference Board) and preliminary Consumer Sentiment (University of Michigan)
  • ISM manufacturing index.

What has changed at the zero-lower bound?

“Intra-day asset price responses to macroeconomic surprises tell us about how markets interpret the macroeconomic releases’ marginal informational content to discern where the economy is currently and where it is heading relative to other economies. Changes in those responses hint at where markets and policy makers perceive risks to the current economic outlook.”

“We…investigate what set of macroeconomic announcements played a more prominent role at the ZLB, both domestically and internationally, by comparing the impact of U.S. macroeconomic news on interest rates and exchange rates before and after the ZLB in small time intervals…Our analysis reveals meaningful changes at the ZLB related to the Fed’s dual mandate.

  • On the inflation side, news on CPI inflation had no significant impact on asset prices prior to 2009. In contrast, during the ZLB period, a hypothetical CPI release that comes in one standard deviation higher than anticipated, lowers 10-year futures prices by 4.7 basis points and appreciates the dollar by 4 to 7 basis points against the Japanese yen, Canadian and Australian dollars…The asset price impact of year-over-year Consumer Price Index (CPI) inflation news rose threefold and took on a novel international role with positive news appreciating the U.S. dollar against three major currencies.
  • Housing releases became more important during the recovery. The strength of the signal from surprises for New and Existing Home Sales flipped in magnitude between the two periods. During the ZLB, all four housing indicators, including Housing Starts and the more leading Permits, raise domestic yields significantly and thus lower Treasury futures prices. New Home Sales, Starts, and Permits are good indicators of the health of the housing market. The housing bust triggered the near collapse of the financial system in the recession and constrained aggregate spending during the recovery, re-focusing investors’ attention to these housing-related indicators.
  • On the employment side, Initial Jobless Claims appear to become more important while standard signals on labor market healing weakened… The weaker impact from Nonfarm Payroll surprises and the simultaneously released unemployment rate may be a result of uncertainties about the impact of secular demographics shifts, e.g. on labor force participation. On the other hand, a high-frequency indicator such as Initial Jobless Claims is less likely to be affected by those secular changes and might thus be more informative about the state of the economy.
  • The effects of two ‘flash’[consumer] survey indicators, the Conference Board Consumer Confidence and the ISM Manufacturing Index, are softer domestically and internationally after2008 and, interestingly, depreciates the dollar against the currencies of the commodity exporters Canada and Australia —a phenomenon also observed for Retail Sales.
  • The informational content of the second and final readings of previous quarter U.S. GDP appear larger in the ZLB era. Positive revision surprises of quarterly GDP contain enough new information to move Treasury futures prices down and ten year rates up, but only during the post-2008 period.
  • The effect of surprises related to manufacturing activity such as Durable Goods Orders, Factory Orders or Industrial Production weakened.”