Advanced Computer Applications Involving Analytics

Assignment 4
Financial Econometrics Assignment

Ardine, Femi

Mahecha Zuñiga, Ana Sofia Alejandra Berenice

Mohammed, Waseefuddin

Muthukur, Prathyusha

Oquendo Castaño, Susana

Rodríguez Bernal, Marcelo

Salih, Aliah

George Brown College

Analytics of Business Decision-Making

Professor: Jay Qi

November 2024

1
The Economist produces “The Big Mac Index” http://www.economist.com/content/big-
mac-indexusing data from various sources. “The Big Mac Index” is computed and
reported every six months by The Economist http://www.economist.com/content/big-
mac-index. It reports both the Raw Index and the Adjusted Index.

The average price of a Big Mac in America in July 2015 was $4.79; in China, it was only
$2.74 at market exchange rates. So, the “raw” Big Mac index says that the yuan was
undervalued by 43% [(=100*(2.74 – 4.79)/4.79)].

The adjusted index uses the “line of best fit” between Big Mac prices and GDP per
person for 48 countries (plus the euro area). The difference between the price predicted
by the [solid] line for each country, given its income per person, and its actual price
gives a supersized measure of currency under- and over-valuation

The Economist posts the raw data that I downloaded (July 2015). Here is the relevant
scatter plot. The simple regression used to create the Adjusted Index uses GDP per
capita in 2014 ($1,000s USD) to predict the price of a Big Mac hamburger in USD at
market exchange rates.

2
Questions:

1. What kind of data are illustrated in the scatter diagram? Experimental?

Observational?

The Scatter Diagram shows an observational data type, as it reflects real-world

information without manipulation, sourced from various countries’ GDP and Big Mac
prices.

This scatter plot shows the relationship between GDP per capita and Big Mac prices.

2. Please use big mac jan 2015.xlsx to replicate the graph above and get the
regression equation.

The regression analysis yielded the following equation:

Predicted Price=Intercept +( Slope×GDP per capita∈thousands)

3
 The regression analysis yielded the following equation:

usd price=2.52251+ 0.03487 × gdp¿ usd

Intercept (2.52251): This represents the estimated Big Mac price when the GDP per
capita is 0. While a GDP of 0 is not realistic, the intercept provides a baseline starting
value for the regression model.

Slope (0.03487): For every additional $1,000 increase in GDP per capita, the Big Mac
price is expected to increase by $0.03487.

3. In Canada a Big Mac is $4.54 USD and GDP per capita is $50,398 USD.
According to the OLS line, what is the predicted price of a Big Mac in Canada?

To calculate the predicted price of a Big Mac based on the regression equation, we use
the following formula:

Predicted Price=2.52251+ ( 0.03487 × 50.398 ) Predicted Price=2.52251+1.757375

Predicted Price=4.29USD
Comparison:

Predicted Price: $4.29 USD. Actual Price: $4.54 USD

The residual is the difference between the actual price and the predicted price:

Residual=Actual Price−Predicted Price Residual=4.54−4.29Residual=0.25 USD

Analysis:

The regression model predicts that the price of a Big Mac in Canada should be $4.29
USD, which is $0.25 USD lower than the actual price of $4.54 USD. This residual
analysis suggests that the model provides a good estimate for Canada, However the
internal inflation play an important variable to predict the best price.

4
 4. In Pakistan a Big Mac is $3.44 USD and GDP per capita is $1,343 USD. What is
the residual for Pakistan?

To find the residual, we first calculate the predicted price using the regression equation:

Predicted Price=2.52251+ ( 0.03487 × 1.343 )

Predicted Price=2.52251+0.04684
Predicted Price=2.57 USD

Comparison:

Predicted Price: $2.57 USD

Actual Price: $3.44 USD

The residual is the difference between the actual price and the predicted price:

Residual=Actual Price−Predicted Price Residual=3.44−2.57Residual=0.87 USD

Analysis:

The residual for Pakistan is $0.87 USD, meaning the actual price of a Big Mac is $0.87
higher than what the model predicts. This suggests that local factors in Pakistan such
as internal conflict cause the Big Mac price to be slightly above the expected value
based on GDP per capita.

5. In Japan a Big Mac is $2.99 USD. What is the Raw Index for Japan?

The Raw Index compares the local price of a Big Mac in one country to the local price in
the United States. It is calculated using the following formula:

US price∈ Japan−Local Price∈the USA

Raw Index= x 100
Local Price∈the USA

Analysis:

5
The Raw Index for Japan is -37.67%, meaning the Big Mac in Japan is 37.67% cheaper
than in the United States. This suggests that the Japanese Yen is fluctuating relative to
the US Dollar based on Big Mac prices.

6. In Japan a Big Mac is $2.99 USD and GDP per capita is $36,332 USD. Given its
GDP per capita, does it seem that the Japanese Yen is undervalued?
To check if the Japanese Yen is undervalued, we compare the actual Big Mac price in
Japan ($2.99 USD) to the predicted price based on its GDP per capita using the
regression model.

Predicted Price=2.52251+ ( 0.03487 × 36.332 )

Predicted Price=2.52251+1.26762

Predicted Price=3.79 USD

Comparison:

Predicted Price: $3.79 USD. Actual Price: $2.99 USD

Analysis:

The actual Big Mac price in Japan is lower than the predicted price. This suggests the
Japanese Yen is undervalued compared to what the model expects based on GDP per
capita. In conclusion Yes, the data shows that the Japanese Yen is undervalued.

7. How should you interpret the R-squared of this OLS regression?

The R-squared value of 0.6226 shows that the model explains a significant part of the
relationship between GDP per capita and Big Mac prices, but some variation remains
unexplained, perhaps because it is complicated to identify all internal variables for each
country.

8. How should you interpret the intercept of this OLS regression?

6
The intercept of 2.52251 is a baseline value that helps define the regression line. It
provides a starting point for understanding how Big Mac prices change as GDP per
capita increases.

9. How should you interpret the slope of this OLS regression?

The slope of 0.03487 shows that for every $1,000 increase in GDP per capita, the
predicted Big Mac price rises by about $0.03 USD. This indicates a positive correlation
between wealth and prices.

10. On average countries with GDP per capita (USD) that is 1 s.d. higher have Big
Mac prices (USD) that are 0.78908 s.d.’s higher.

For every standard deviation increase in GDP per capita (adjusted to increments of
$1,000 USD), the price of a Big Mac in USD increases by an average of 0.789 standard
deviations. Furthermore, with a p-value of less than 0.0001, GDP per capita has a
statistically significant impact on the price of a Big Mac.

7
 11. Get the variance-covariance matrix for the Raw and Adjusted indices for the 49
countries. What is the coefficient of correlation between the Raw Index and the
Adjusted Index?

The correlation between raw index and the adjusted index is 0.5772.
Additionally, since the p-value (Prob > |r|) is less than 0.0001, this indicates a
statistically significant positive correlation between the variance raw index and
the adjusted index.

12. Forecast the price of a Big Mac for a particular country that has a GDP of
$15,000 USD per capita and give the associated margin of error for your
forecast. Answer with a quantitative analysis and 1 sentence.

Margin of error = 1.96 x 1.25412 = 2.4580752

Confidence Interval = ( 3.02728 - 2.4580752 , 3.02728 + 2.4580752)
= ( 0.5692048 , 5.4853552)

The price of a Big Mac in Chile, with a margin of error of 1.25412, can vary between
0.5692 and 5.4853 with 95% confidence

13. How would you check for a violation of the assumption that 𝜖𝑖 is Normally
distributed (i.e. Assumption #4), which is relevant to the calculations in the
previous part? Answer with 1 sentence.
Check for a violation of the assumption by performing a normality test (e.g.
Shapiro-Wilk Test, Kolmogorov-Smirnov Test, and Anderson-Darling Test) and
can generate a Q-Q plot on the residual to check if they adhere to the rule of a
normal distribution.
8
Question 2 (Based on Data big mac jan 2017.xlsx, However, you don’t need to
work data for this part, just answer based on your analytics skills)

The big Mc data is updated in 2017. Part of those data1 is reproduced below, showing
the price of a Big Mac hamburger in China between June 2005 and January 2017. For
each reported time period, it shows the local price, which is in Chinese Yuan ¥. The
regression output, given below too, shows a regression of the local price on the time
period measured by t, which is the number of months since June 2005.

(a) If the variable t were measured in years, not months, since June 2005, what would

took a value of 1.5833 years. Also, would the 𝑅2 be higher, lower or the same?
the equation for the OLS line be? For example, if instead of a value of 19 months, t

Explain your answers. Answer with an OLS equation using standard notation and 2
– 3 sentences.

 OLS equation (monthly)

local_price = 9.818 + 0.0647 * t
 OLS equation (yearly)
local_price = 9.818 + (0.0647 *12) * tyears
= 9.818 + 0.7764 * tyears
The OLD equation changed to yearly will be adjusted from monthly to yearly by
adjusting the t monthly to t years, the number is changed from 0.0647 to 0.7764 which
is a multiplication of 12. The t variable is only being rescaled from monthly to yearly
without much impact to the result of local_price, therefore, the R2 would remain the
same. Both dependent and independent variables would remain the same as well, thus
R2 would not be affected.

(b) Imagine an isolated data-entry mistake: the local price in June 2005 is recorded as
19.50 instead of 10.50. Would this data-entry mistake create an outlier? What
effect, if any, would this mistake have on the number 0.448771056 under
“Regression Statistics” in the Regression output? Explain why 0.448771056 would
go up, down or remain the same, making sure to include its units of measurement
and what that number measures in a practical sense. Answer with 4 – 5 sentences.
The data-entry mistake would likely to cause trouble in the data by creating an outlier,
because 19.50 is a pretty significant deviation from the original entry of 10.50. The
number 0.44877 under “Regression Statistics” would jump up if this data-entry mistake
happened. The reason is the outlier will increase the residual result, therefore, the linear
9
regression model will have a bigger spread. The model will be less fitting due to more
variability in the data relative to the fitted regression model.

10