Financial Econometrics Analysis of Coca-Cola HBC AG Share Prices

This study uses financial econometrics to analyze the impact of macroeconomic variables on Coca-Cola HBC AG share prices from January 2013 to January 2018. The study includes descriptive statistics, random walk theory, linear regression modeling, and GARCH modeling to examine the impact of FTSE 100 index, consumer price index, gross debt, 3-months Treasury bill, monthly consumer credit, and spot exchange rates on share prices.

P a g e | 1
681902

P a g e | 2
Executive Summary
Financial econometrics can be envisaged as a pivotal element in analysing any
company’s financial position based on historical data. Coca-Cola HBC AG , the second
largest anchor bottler company in the world is a revenue spinner and various factors affect its
share prices. Macroeconomic parameters like FTSE 100 index, consumer price index (CPI),
gross debt, 3-months Treasury bill, monthly consumer credit (including student loans), spot
exchange rates with respect to US dollar and against Euro have been taken to harness the
impact on share prices. The time period taken is from January 2013 to January 2018. A time
series approach and modelling techniques have been impregnated here.
Through the application of statistical software Eviews, the descriptive statistics of
share prices will be analysed, then random walk hypothesis will be evaluated, linear
regression modelling will be internalized for analysing the impact of various macroeconomic
variables on share prices and GARCH model of time series for analysing volatility and risk
mitigation.

P a g e | 3
I. Introduction
A bird’s eye view
In the world of beverage industry across the globe, the name and brand of ‘Coca-
Cola’ hardly needs any mention. In the current scenario, Coca-Cola HBC AG is the second
largest anchor bottler company with respect to volume sales. In the year 2017, the company
has reached landmark sales of 2.1 billion unit cases and catering to around 600 million people
in 28 countries spanning across three continents (About Us, Coca Cola HBC AG, 2018).
Glimpse of the business
Year 2017 has been of significant progress for the company in terms of core business.
The net sales revenue triggered by 5.9%. The aggregate volume of trade increased by 2.2%
across all industry segments. Momentum has been significant in emerging and developing
countries. In terms of comparable EBIT, there has been a substantial jump of 20%. Robust
growth has been witnessed in the segment of sparkling and still drinks. The free flow of cash
has been € 425.9 million, a significantly higher operating cash flow as compared to the
previous year (Coca-Cola HBC AG, Financial Statement, 2018).
2. Other macroeconomic variables affecting share prices
The main macroeconomic variables affecting the share prices of the company are
FTSE 100 Index, consumer price index (CPI), General government gross debt (Percent of
GDP), Monthly average rate of discount, 3 month Treasury bills, Monthly consumer credit
(including student loans) (seasonally adjusted), Spot exchange rate (Sterling against USD)
(Month average) and Spot exchange rate (Sterling against Euro) (Month average). There are
other macroeconomic variables that affect the share prices of the company. As for instance,
changes in technological factors can be contemplated as one of the significant
macroeconomic factors affecting the share prices of the company. Suppose a new technology
has come in the market and the company has not been able to implement that technology then
that affects the share prices (Kettell, 2002). Other macroeconomic variables that come into
play for affecting share prices are political factors like change in government, wars, strikes in
the country, financial crisis, any international events and so on. Although not
macroeconomic, but micro-economic variables like announcements associated with profits,

P a g e | 4
dividends, orders, management changes, issue of new shares and various other happenings
(Mott, 2008).
3. Descriptive statistics of the company’s monthly returns
When we see the mean, it is 16.73 which is well positioned between the minimum
value i.e. 11.066 and 25.85. There is no mode which reflects that there is no share price
occurring twice which is absolutely perfect as the variations are daily basis and then the
average value is taken month-wise and also represents the volatility in the stock prices. The
standard deviation is 3.73 and the skewness is 0.983378 which is right skewed. The
difference between the mean of the distribution and the observations in the frequency
distribution curve is concentrated more towards the right tail. The distribution is also not
normal in nature as the mean and standard deviation are not 0 and 1 respectively.
Fig.1 Variables after importing the dataset

P a g e | 5
Fig.2 Descriptive Statistics
5. Random walk theory and ADF testing for stationarity
Time series data like stock prices over time are usually non-stationary in nature. Due
to volatile nature of share prices and randomness, it becomes difficult to predict the share
prices over time. In a very simple language, the path of the share prices across time follows a
random walk (Cox, 2017).
Thus, given random walk, the importance of stationarity comes into the being. In
terms of statistics, a stationary time series is a time series where the major statistical
properties like mean, autocorrelation, variance and so on are throughout constant over the
given period of time. From the descriptive statistics, it can be seen that the share prices of
Coca-Cola HBC AG does not follow stationary path and thus there is non-stationarity in the
data. One of the statistical tests to analyze stationarity is ADF (Augmented Dickey Fuller
Test). The test is conducted in order to test from where the non-stationarity in data is coming
from. A length lag is necessary for running the test so there is no serial correlation amongst

P a g e | 6
the residuals. The test is also called unit root test and the null and alternative hypotheses at a
particular level of confidence let’s say at 95%, the same are as follows:
H0 (Null Hypothesis) : There is presence of a unit root
H1 (Alternative Hypothesis): There is no unit root
Fig.3 Line Graph of the Share prices
It can be seen that the share prices have started falling from approximately 2013
quarter IV to 2015 quarter 1 which is the breakpoint in the data series. After that, increase in
the share prices has been witnessed. It can be seen that the data fluctuates from the mean
16.72 if we draw a line from 16.72 which is a constant, we can compare the fluctuations
around the mean. In case of random walk, this constant is zero.

P a g e | 7
Fig.4Setting up the assumptions for ADF test
Fig.5 ADF test result

P a g e | 8
From the above table results, it can be seen that the test has been conducted with share
price as the independent and dependent variable is difference of SP. When we see the model
we find that the p-values for SP(-1) and C are 0.9903 and 0.8481 are greater than 0.05 and
they are not significant.
Fig.6 ADF test result for Log of share price
For stationarity check, we need to see the ADF test statistic which is less than the
critical values at 1%, 5% and 10% level. For stationarity, ADF value must be greater than the
critical values of the t-statistic.

P a g e | 9
Fig. 7 ADF test: 1st difference
At 1st difference, it can be seen that the ADF value is greater than all the critical
values and it can be stated that the data is stationary at 1st difference.
Fig. 8 1st difference of share price

P a g e | 10
Fig. 9 Stationarity in 1st difference of the logarithm of share prices
Thus, it can be stated that transforming the data at 1st difference, non-stationarity is
avoided and thus, the estimation of the model becomes robust after 1st differencing. This way
the random walk hypothesis is also satisfied here.

P a g e | 11
6. Regression model that seeks to examine whether unexpected changes in the independent
variables have any impact on the monthly share returns
The share prices of the company is basically the dependent variable here and the
independent variables are FTSE 100 Index, CPI, General government gross debt (GD),
Monthly average rate of discount-3 month Treasury bill, Monthly consumer credit (including
student loans), Spot exchange rate (Sterling against USD) (Month average) and Spot
exchange rate (Sterling against Euro) (Month average).
The linear regression model can be given as follows:
SP=α + FTSE 100 Index+GD+CPI +MONTHLY ¿+ MONTHLY ¿+ SPOT ¿ S 01+ SPOT ¿ STER+ϵ
The equation is as follows:
Fig. 10 Regression Equation in Eviews

P a g e | 12
Fig. 11 Regression Output
From the above regression output, it can be concluded that CPI and monthly
consumer credit (including student loans) have statistically significant impact on the
dependent variable i.e. share price at 95% level of significance.
7. Analysis of coefficients
Now, analysing the signs of the coefficients, it can be stated that with the increase in
CPI, the share prices of the company increases. As the consumer spends high, it gets directly
reflected in the share prices (CPI Vs. Stock Prices, n.d.). the sign of coefficient of FTSE100
index is negative which implies that with increase in FTSE 100 index, there is a decrease in
the share price of Coca Cola. The FTSE 100 index basically reflects the average change in
the price of the share of top 100 companies trading on the London Stock Exchange
(Anderson, 2010).
The 3-month treasury bill is an instrument of open market operation by the
government wherein the government sells them to reduce money supply and buys them to
increase money supply in the market. The signs of the spot exchange rates with share prices

P a g e | 13
are also valid. With the increase in spot exchange rates with respect to foreign currencies, the
share prices of the company also increase (Farmer, 2010).
8. Modeling the risk levels of the company using GARCH model
For GARCH modeling, first we have to generate a return variable from the SP
variable of the dataset in eviews. The return variable acts as a measure of volatility and risk in
the share prices across the time span taken into consideration.
Fig. 12 Return Variable creation in eviews

P a g e | 14
Fig. 13 Checking the pattern
Now, we square the returns and run a correlogram of the same.
Fig. 14 Squaring the return variable

P a g e | 15
Fig.15 1st difference correlogram

P a g e | 16
Fig.16 GARCH model development

P a g e | 17
Fig.17 GARCH model results
9. Comment on the overall performance of the company
The share prices of the company increased after 1st quarter 2015 triggered by
consumer credit including student loan and consumer price index. The share prices did not
follow a random walk at the first place but after first order differencing and running the ADF
test, stationarity has been attained. The GARCH test also states that after 1st difference of the
returns gives a significant result for predicting the returns that is useful for proper estimation.

P a g e | 18

P a g e | 19
References
About Us, Coca Cola HBC AG, (2018). Available at:
<https://coca-colahellenic.com/en/about-us/> [Accessed on 21st February, 2018]
Anderson, D, R (2010). Statistics for Business and Economics.Cengage Learning EMEA
Cox, D., R. (2017).The Theory of Stochastic Processes.Routledge.
CPI Vs. Stock Prices (n.d.). Available at: <https://finance.zacks.com/cpi-vs-stock-prices-
5166.html> [Accessed on 21st February, 2018]
Coca-Cola HBC AG, Financial Statement, (2018). Available at: <https://coca-
colahellenic.com/media/2943/ye-2017-press-release_eng_final.pdf> [Accessed on 21st
February, 2018]
Farmer, R, E, A (2010). How the Economy Works: Confidence, Crashes and Self-Fulfilling
Prophecies. Oxford University Press
Kettell, B. (2002). Valuation of Internet and Technology Stocks: Implications for Investment
Analysis. Elsevier

P a g e | 20