Does ENSO Impact Equity Returns? Evidence via Country and Panel Regression

Abstract

This study employs a panel local projections (LP) model to analyze the influence of El Niño (EN), La Niña (LN) Southern Oscillation (ENSO) on real equity returns based on twenty economies, collectively accounting for 80.2% of global output. We demonstrate that an ENSO shock has a delayed and oscillatory effect on equity returns, where the most significant impact is positive (negative) during an EN (LN) phase. We further analyze the climate state by each of the twenty economies. The literature shows that the effect of an ENSO shock on economic growth can vary across different economies. A main finding is that we show a near uniform result similar to the panel LP model, which we posit is driven in part by strong interconnections of financial markets on a global scale. These findings offer valuable insights for policymakers and investors.

Highlights

• We estimate a panel and economy-specific climate ENSO switching via LP model.

• Twenty economies representing 80.2% of global output are analyzed.

• ENSO shock has a delayed and oscillatory effect on equity returns.

• Most significant impact of an ENSO shock on equity returns is positive (negative) during an EN (LN) phase.

KEYWORDS:

• Equity prices
• linear and nonlinear local projections
• ENSO
• global macroeconometric modeling

JEL Classifications:

• F44
• E44

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 See https://www.weather.gov/mhx/ensowhat.

2 The existing literature estimates the influence of a partial ENSO shock in isolation (e.g., Brunner, 2002 and Cashin et al., 2017) to analyze the partial effects of an EN (a LN) shock by "shutting off" the LN (EN) in their specification, which is problematic considering there is a delayed reaction of ENSO shocks on climate. According to NOAA, episodes of El Nino and La Nina occur 9–12 months. Dufrenot et al. (2021) find there is a delayed reaction on global commodity prices.

3 Our choice for selecting the 20 economies is based on maximizing the number of countries based on data availability.

4 Based on IMF data using 2022 data, see https://www.imf.org/external/datamapper/NGDPD@WEO/EURO/USA/CAN/CHN/CHL/IND/ KOR/BRA/WEOWORLD/COL/GBR/JPN/RUS/SWE.

5 The 20 economies include: Brazil (”BRA”), Switzerland (”CHE”), Chile (”CHL”), Canada (”CAN”), China (”CHN”), Columbia (”COL”), Czech Republic (”CZE”), Euro zone (19 countries; ”EUR”), United Kingdom (”GBR”), Hungary (”HUN”), Ireland (”IRL”), India (”IND”), Israel (”ISR”), Japan (”JPN”), South Korea (”KOR”), Poland (”POL”), Russia (”RUS”), Sweden (”SWE”), Turkey (”TUR”) and the United States (”USA”). Based on IMF data, the twenty economies considered in this paper represent 80.2% of global output, see https://www.imf.org/external/datamapper/NGDPD@WEO/EURO/USA/CAN/CHN/CHL/IND/KOR/BRA/WEOWORLD/COL/GBR/CZE/HUN/ IRL/ISR/JPN/POL/RUS/TUR/SWE/CHE. The Euro zone values are based on the 19 member countries (i.e., Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Portugal, Slovakia, Slovenia, and Spain).

6 For India, manufacturing production index (FRED mnemonic INDPRMNTO01IXOBM) is used as opposed to total production index (FRED mnemonic INDPROINDMISMEI), considering data availability (the correlation between the is 0.9918 for Jan 2000 to Dec 2018). For China, we use total production excluding construction (FRED mnemonic CHNPRINTO01IXPYM). As the production index for China includes missing values, the Kalman smoother using an ARIMA state space representation is used to impute missing values.

7 See https://www.weather.gov/mhx/ensowhat.

8 We include ENSO as a shock variable, considering it is clearly not influenced by equity returns.

9 The HP filter is applied using smoothing parameter λ = 129,600 for data at monthly frequency (Ravn & Uhlig, 2002).

10 According to NOAA, episodes of EN and LN occur 9–12 months, see https://oceanservice.noaa.gov/facts/ninonina.html.

11 The output gap is estimated for each economy using the HP filter, where the smoothing parameter λ = 129,600 for data at monthly frequency (Ravn & Uhlig, 2002).

12 The exception is the lag of the target variable in Z_i,t is removed in the country LP model since the autoregressive term is determined by the BIC.

13 See https://www.bloomberg.com/news/articles/2021-06-04/powell-diverges-from-european-central-bankers-on-climate-policy