Effect of Weather on Cryptocurrency Index: Evidences From Coinbase Index

This study proposes to investigate the dynamic relationships between the three weather factors (temperature, humidity, and wind speed) in New York City of USA and Coinbase Index from Federal Reserve Bank of St. Louis, in the USA. Statistical tools like Descriptive Statistics, Unit Root, Granger Causality Test and Johansen Co-Integration test were employed. This study clearly found that the temperature influenced the investors‟ mood and their investment decision in respect of Cryptocurrency index (Coinbase Index) and also found that there was long run equilibrium between the sample variables during the study period. The results of study provided strong evidence against the Efficient Market Hypothesis (EMH).


Introduction
Digi Cash, the first digital currency, was introduced in the 19 th Century (Chaum, 1981;Phillip A., 2017). Nakamoto (2008), improved concept of peer-to-peer networking and crowd sourcing and it was introduced as new digital cash format called the Bitcoin. Cryptocurrencies have emerged as a growing digital currency, introduced into a new investment segments (Brauneis A. and Mestel R., 2018;Urquhart, 2016). The number of Cryptocurrencies like Bitcoins has been reached more than 1,500 (Kim, 2017). More than 500 Cryptocurrencies did have a market capitalization, worth over 25 million dollars (Tomá s and Ibañez, 2018). Cryptocurrencies are now accepted as legal tender in many developed and developing nations and also accepted by different financial institutions, including banks, hedge funds and even Government bodies (Vidal-Tomá s and Ibañez, 2018). The most popular segment of Cryptocurrencies, in the form of market capitalization, was the Bitcoin (Brauneis A., Mestel R., 2018). The market capitalization of Bitcoin had reached 10.1 to 79.7 billion from October 2016 to October 2017 (M. Brandvold et al., 2015) the price of Bitcoin also has increased from 616 to 4800 US dollars (Shaen Corbet et al 2017). Bitcoin developed as the largest decentralized Cryptocurrency, with the help of block chain technology (Bariviera, 2017). Therefore, many academic researchers have started to conduct intensive research on Bitcoin Katsiampa, 2017;Urquhart, 2016 andKim, 2017). No wonder cryptocurrencies have become the most trending topics in recent economic and financial issues and, they have been discussed by several bodies like European Central Bank, European Banking Authority and Financial Action Task Force, businesses and academic communities (Dwyer, 2015; Bariviera et al., 2017). Besides, were many financial scholars attempted to analyze the suitability of cryptocurrencies as investment assets and found some statistical significance such as market efficiency (Urquhart, (2016); Bariviera, 2017); Nadarajah and Chu, 2017), leptokurtosis , heteroschedasticity, long-memory (Phillip et al., 2018), return-volume relationships (Gkillas and Katsiampa, 2018). According to Figure 1, four different psychological biases, namely, overconfidence, conservatism, herding attitude and availability directly influenced the investors" decision-making process and this has been proved in the developed nations at different periods of time.  (2016) Majority academic research studies, conducted so far, have focused only on time series based on technical aspects of cryptocurrency markets. But there was a lack of comprehensive research, on behavioral aspects of the investors, with different cryptocurrencies in respect of weather factors. To fill this gap, this study examines the dynamic relationship between Cryptocurrency index, namely Coinbase Index and three different weather factors (temperature, humidity, and wind speed) in New York City. This is the first study of this nature that addresses the behavioral aspects of investors covering two variables-weather factors (namely temperature, humidity, and wind speed) and cryptocurrency.
Thus this study contributes to the existing body of literature. The study was structured as follows; in Section 2, the study discusses the data source and the methodology, the Section 3 discusses the empirical findings and concludes with Section 4.

Objectives of the Study
The aim of the study is to find out the cause and effect relationship and long run equilibrium relationship between weather factors (temperature, humidity, and wind speed) in New York City of USA

Sample Selection
In order to analyse the dynamic relationships between weather factors and Coinbase Index, the study used three weather factors (temperature, humidity, and wind speed) in New York City of USA and cryptocurrency data, namely, Coinbase Index from of Federal Reserve Economic Data (FRED) database of Federal Reserve Bank of St. Louis.

Study Period
The present study covered a period of three years from 01.01.2015 to 30.06.2018. The Coinbase index was introduced in USA from January 1, 2015 (https://am.coinbase.com/index)

Data Sources
For the purpose of examining the relationship between weather factors and Coinbase Index, the daily closing values of sample index and weather factors were collected from two different databases. The data relating to Cryptocurrency (Coinbase Index) were collected from Federal Reserve Economic Data (FRED) database (https://fred.stlouisfed.org/series/CBCCIND) of Federal Reserve Bank of St. Louis. The data relating to New York City daily weather factors (temperature, humidity, and wind speed) were obtained from the National Climatic Data Center (https://www.ncdc.noaa.gov/cdo-web/), in Asheville, North Carolina. The missing values in the data on sample variables, for some days, were filled up by taking the average of the two nearest cases. The formula of calculating the natural log of closing prices is given below. R t = 1n (p t /p t-1 ) Where: R t : Return on day"t" P t : Index Closing Value on day"t-1" 1n: Natural log

Tools Used for Analysis
The following tools were used for the purpose of analysis.


Descriptive Statistics (to find out the normality of Cryptocurrency index and Weather Factors)  Unit Root Test (to test stationarity of Cryptocurrency index and Weather Factors).

 ARCH and GARCH models (to examine the impact of Cryptocurrency index and Weather Factors)
 Granger Causality (to examine the cause and effect of Cryptocurrency index and Weather Factors), and  Johansen Co-Integration (to find out long run relationship of Cryptocurrency index and Weather Factors)  the lowest risk value in terms of standard deviation (SD) of 0.307323. The analysis values of skewness, kurtosis and the Jarque-Bera tests also clearly indicated that the distribution of return data for temperature, humidity, and wind speed was normal during the study period. Hence the null hypothesis (NH01) -"There is no normality in the daily return data of Cryptocurrency index and weather factors in New York City over the sample period", was rejected.    In other words, all the weather variables, except temperature, namely, humidity, and wind speed were not statistically significant. It means that these two variable did not influence the return of Cryptocurrency index. Hence, the Null Hypothesis (NH3) -There is no volatility among the Cryptocurrency index and weather factors in New York City, was partially rejected.

Granger Causality for the Returns of Sample Cryptocurrency Index and Weather Factors in New York City of USA
The results of Granger Causality for the returns of Cryptocurrency index (namely Coinbase Index) and three different weather factors (temperature, humidity, and wind speed) in New York City of USA, during the study period from 01.01.2015 to 30.06.2018, are displayed in Table 4. The study generally accepted the null hypothesis when the p-Values for sample variables were above 0.05 and rejected same when the values were less than 0.05, under the Granger Causality Test. The study clearly found bidirectional causal relationship between Coinbase Index and temperature in New York City (i.e. P-Value was at 0.0021 for Coinbase Index and 0.0005 for temperature) during the study period. In other words, no one weather variable, except temperature, did show statistically significant relationship during the study period. Hence the Null Hypothesis (NH03) -"There is no causal relationship between CBCCIND index and temperature in New York City", was partially accepted. Note: Rejection of Null Hypothesis when the Probability value is less than or equal to 0.05

Johansen Co-integration for the Returns of Sample Cryptocurrency Index and Weather Factors in New York City of USA
Tables 5 show the results of Johansen Co-Integration Test for the returns of sample Cryptocurrency index, namely Coinbase Index and three different weather factors (temperature, humidity, and wind speed) in New York City of USA during the study period from 01.01.2015 to 30.06.2018. It is to be noted that the returns data of the sample Cryptocurrency index and weather factors were used to test the Co-Integration among the above samples. It is found from the results of Johansen Co-Integration Test that the Coinbase index returns was integrated with weather factors in New York City of USA. Besides, the p-Value under trace statistics and maximum Eigen values for sample variables, were below the significant levels (below 0.05 level). This indicated that there was long run equilibrium relationship or Co-Integration relationship between the returns of sample Cryptocurrency index, namely Coinbase Index and three different weather factors (temperature, humidity, and wind speed) in New York City of USA during the study period. Hence NH4-There is no long run equilibrium relationship between the Coinbase Index and weather factors in New York City was rejected.

Graphical Exposition
The Dot Plot, drawn for the results of the weather factors (temperature, humidity, and wind speed) in New York City, USA and Coinbase Index from Federal Reserve Bank of St. Louis, over the period of study from 1st January 2015 to 30th June 2018was exhibited In Figures 3 and 4. It is observed from the Figure 3 that temperature would have spread over the whole area of New York City than other two weather variables, namely, humidity, and wind speed. It means that temperature would have strongly influenced the human attitude and their day to day activities in respect of their investment during the study period. It is clearly evident from the Figure 4 that the Coinbase Index (CBCCIND) also gradually moved in the upward direction. This indicated that the performance of Coinbase Index was better and provided higher returns to the investors during the later part of the study period.
The movement of scatter (regression line) of weather factors (temperature, humidity, and wind speed) in New York City, USA and Coinbase Index from Federal Reserve Bank of St. Louis, over the period of study from 1st January 2015 to 30th June 2018 is shown in Figure 5. It is clear that regression lines of all the sample variables moved in the upward direction, showing positive sign and these variables did have a strong linear relationship. This shows the fact that there was interrelationship between weather factors and Coinbase Index but one variable, namely, temperature alone influenced the returns through the study period.

Conclusion
It is evident that weather factors could influence the moods of investors and their behaviors, which would, in turn, help them to take investment decisions in their life (Kathiravan et al., 2017. The present study, which attempted to understand the dynamic relationships between three weather factors (temperature, humidity, and wind speed) in New York City of USA and Coinbase Index from Federal Reserve Bank of St. Louis in United States, found that the temperature in New York City influenced the Cryptocurrency index negatively (from their p-value of 0.0021 and 0.0005 respectively). The study also found long run equilibrium with the Cryptocurrency index and sample weather factors. The findings of the present study confirmed the findings of previous studies of Howarth & Hoffman 1984;Kramer & Runde 1997;Kamstra, et al. 2000;Pardo & Valor 2003;and Tufan & Hamarat 2004, who found that the mood of individual investors and their consequent investment decisions were influenced by different weather factors. In short, there was chain linking between temperature levels and, human mood, their behavior and investment decisions and index returns.