Industry Analysis (E-Commerce)

1 | Page

Assignment 3

Industry Analysis (E-Commerce)

This paper serves as the second portion of the fundamental analysis – industry analysis. We choose e-commerce retail industry. The most typical company is Amazon.com.

1. Current performance – sales trend

According to the most recent US Census Bureau news released on November 17, 2017[1], the estimate of U.S. retail e-commerce sales for the third quarter of 2017, adjusted for seasonal variation, but not for price changes, was $115.3 billion, an increase of 3.6 percent (±0.7%) from the second quarter of 2017. E-commerce sales in the third quarter of 2017 accounted for 9.1 percent of total sales. The third quarter 2017 e-commerce estimate increased 15.5 percent (±1.1 %) from the third quarter of 2016 while total retail sales increased 4.0 percent (±0.4%) in the same period.

Figure 1

Source of data:

From Figure 1 we can see a basic trend of e-commerce retail sales – increasing percentage of total quarterly retail sales. In the first quarter of 2008, the percentage was around 3.5%. By the third quarter of 2017, this percentage has increased to 9.1%. The trend line equation indicates that the percentage is increasing at about 0.12 percentage point per quarter or 0.48 percentage point per year rate.

Figure 2

Source of data:

Figure 2 shows the e-commerce retail sales in million of dollars since quarter 4 of 1999, not seasonally adjusted. The trend line follows a very good exponential growth pattern. The quarterly growth rate is 3.94%. At this rate, it will take less than 18 quarters or 4.5 years to double.

1. Current performance – profit trend

E-commerce retail industry is not a highly-profitable industry. In fact the net profit margin of this industry is very low. The average net profit margin for the third quarter of 2017 was 1.45%, while the net profit margin for the retail industry as a whole was 2.22%[2]. Figure 3 below shows the net profit margin for both the retail industry and the e-commerce industry from Q4 of 2014 through Q3 of 2017. We can see that for most of the time, the net profit margin for the e-commerce industry is lower than that for the retail industry. The only exceptions were for Q4 of 2015 and Q2 of 2016.

Figure 3

Source of data:

1. Forecast of industry sales

We gathered the following macroeconomic data for the same period (Q4 1999 through Q3 2017):

Disposable personal income (DPI)[3]

Disposable personal income: Per capita (DPIPC)[4]

Personal consumption expenditure (PCEC)[5]

Real GDP per capita (RGDPPC)[6]

Correlation analysis indicates that all of these variables including e-commerce sales are highly correlated. See the correlation matrix presented in Table 1 below:

Table 1 Correlation matrix

E-com sales / DPI / DPIPC / PCEC / RGDPPC
E-com sales / 1
DPI / 0.952 / 1
DPIPC / 0.944 / 0.999 / 1
PCEC / 0.956 / 0.998 / 0.997 / 1
RGDPPC / 0.893 / 0.926 / 0.933 / 0.943 / 1

If we choose more than one predictors from the macroeconomic variables, there may be multicollinearity problem – high correlation between two predictors in a multiple linear regression. To avoid that problem, we choose only one predictor. We pick the one that is most correlated with e-commerce sales, which is Personal consumption expenditure (PCEC). The correlation between e-commerce sales and PCEC is 0.956.

Figure 4

Table 2 Regression analysis

The scatterplot in Figure 4 and the regression analysis in Table 2 present the regression equation for predicting e-commerce sales using PCEC. We can see that the model is highly significant, with F(1,70)=751.155, p=3.72x10-39. The model explains 91.5% of variation in e-commerce sales.

The empirical equation is

E-commerce sales = 14.919*PCEC – 103,999

It shows that for each $1 billion increase in PCEC the e-commerce sales are expected to increase by 14.919 million dollars. Or we can roughly sale, about 1.5% of personal consumption is spent on online purchasing.

Purely for the forecasting purpose, it might be even better to consider a quadratic model shown in Figure 5 below. This model improves the model R2 to 0.9619, that is, 96.2% of the variation in e-commerce sales is explained by the quadratic model. Recall that the linear model explains only 91.5% of the variation in e-commerce sales.

Figure 5

Recall that in Figure 2 we plotted e-commerce sales against time and achieved an R2 as high as 0.96 too. From forecasting point of view, this trend equation is even more useful because if we want to use the macroeconomic variable as the predictor, we have to first forecast the macroeconomic variable first, which can be again a more challenging job. But with the trend model, it is easier because we know the time variable for the future.

To capture the seasonal variation, we created three dummy variables Q1, Q2 and Q3 to represent the quarter of a year:

Q1=1 if quarter 1 0 otherwise

Q2=1 if quarter 2 0 otherwise

Q3=1 if quarter 3 0 otherwise

Plus the time variable (T=1 for Q4 of 1999), we can build a multiple linear regression model as follows:

Table 3 Time-based regression model

The regression equation is

E-commerce sales = -253+ 1387*T – 10300Q1 – 10115Q2 – 10433Q3 (1)

This model explains 92.8% of variation in e-commerce sales. It shows that each quarter e-commerce sales is expected to increase by 1387 million (seasonally adjusted). The slopes for the three dummy variables show the seasonal variation. For example, compared with Quarter 4, Quarter 1 sales are typically 10,300 million lower, Quarter 2 sales are typically 10,115 million lower, and Quarter 3 sales are typically 10,433 million lower. This makes sense because the biggest online sales occur during the end of year holiday seasons such as Thanksgiving Black Friday, Christmas and New Year.

1. Forecast of industry profit

In Figure 3, we see that net profit margin of e-commerce industry varies between 0.44% and 4.76%, but most typically around 1.45% (median value). So for forecasting purposes, we assume that the future net profit margin will be about this figure. Given that, the predictive equation for the net profit of e-commerce industry is

Net profit=1.45%*Sales (2)

where sales are estimated using equation (1).

Applying the models (1-2), we obtain the forecasts for the next four quarters as follows:

Table 4 Forecasts of the next four quarters

Quarter / Forecasted
sales ($million) / Forecasted
net profit ($million)
Q4’2017 / 100,989 / 1,464
Q1’2018 / 92,076 / 1,335
Q2’2018 / 93,648 / 1,358
Q3’2018 / 94,716 / 1,373

1 | Page

[1]

[2]

[3]

[4]

[5]

[6]