Predicting GDP growth with stock and bond markets: Do they contain different information?

Abstract

This paper examines the ability of bond and stock markets to predict subsequent GDP growth over a range of horizons for 12 international countries. The results, using linear, probit, time‐ and regime‐varying in‐sample regressions and out‐of‐sample forecasting, confirm the view that both financial markets exhibit predictive power for future output growth. Moreover, there is notable variation within the strength of the predictive relation, for example, predictive power increases during the financial crisis period. Results suggest that while the term structure arguably exhibits stronger predictive power, both series contain distinct predictive information. Notably, predictive power emanating from the stock return series appears stronger over shorter (up to four‐quarter) time horizons, while the term structure series exhibits more consistent predictive power over a range of horizons. Considering different regimes, we observe that the bond market exhibits greater predictive power for a flatter yield curve and lower stock prices relative to fundamentals, while the stock market exhibits greater predictive power for a steeper yield curve and higher relative stock prices. This suggests that the two financial markets exhibit different information content for future output growth. This view is further supported by forecast results whereby a model that includes both financial series outperforms a model that only includes one. Forecast encompassing tests further support the view that stock returns contain additional information over that presented by the term structure alone.

Do financial markets predict macroeconomic performance? US evidence from risk‐based measures

“…They note that while the bond market is typically preferred to the stock market, overall predictive power is enhanced by including both. A similar result is also reported by McMillan (2021) who argues that the stock market provides greater predictive power during ‘normal’ periods, but the bond market is better at capturing recessionary periods. Both papers link with Chinn and Kucko (2015), who note changing predictability arises from unsettled economic conditions.…”

“…For example, Harvey (1989) and Stock and Watson (1990) both argue that term structure provides greater predictive power for subsequent growth compared to the stock return. Kuosmanen and Vataja (2019) and McMillan (2021), while confirming this view also report that improved predictive power is found by incorporating both stock and bond market information. To date, this has not been extended to consider the VRP and default yield (DFY), which may better capture market risk and thus, predictability.…”

Macroeconomic Shocks and Economic Performance in Malaysia: A Sectoral Analysis

“…When stocks in a given sector go up, more often than not that sector will show a rise in sales, earnings, and outlays for plant and equipment." McMillan (2021), using quarterly data from 1973 to 2017 for 12 countries, found that stock prices have predictive power for future GDP in several cases. Liu et al (2007), employing monthly data from 1987 to 2004 for many countries, reported that industry valuations implicit in industry earnings data closely follow industry stock prices in several countries.…”

Forecasting Commodity Prices Using the Term Structure

“…The macroeconomic literature has established that future real economic growth is positively correlated with lagged interest spreads (e.g., Stock and Watson 1989;Estrella and Hardouvelis 1991;Plosser and Rouwenhorst 1994;McMillan 2021a). In parallel, the link between interest rates and commodities has been also investigated and can be classified into two categories.…”