FU902: Seminar 01
David Procházka
2023-10-10
1 Install R & RStudio
- First install R, then install RStudio, both can be downloaded Here
2 Get familiar with the R-basics
a. Tutorials
b. YouTube channels
3 Get familiar with the regression modelling in R
4 Learn (the R-basics) by example
a. Download the dataset
All files are available at this shared folder
For this Tutorial, download the file “FU902-dataset-01” to your local computer
Set a working directory using the command setwd(“C:/…/…/…/…”); where you have the downloaded file stored and where outputs will be exported (e.g., setwd(“C:/Users/MyPC/Documents/R/Paper01”)
Upload the download file from the working directory (for how to upload data in other formats, like txt, csv, please refer to the manuals listed above)
library(readxl)
dataset <- as.data.frame(read_excel("FU902-dataset-01.xlsx"))
attach(dataset)
head(dataset)## Firm Year Industry Country Profit Revenue Orders ESG
## 1 1 2017 Services ESP 68761 435676 1.038 58
## 2 1 2018 Services ESP 27267 220696 0.976 39
## 3 1 2019 Services ESP 69729 255782 1.111 85
## 4 1 2020 Services ESP 87363 513659 1.055 63
## 5 1 2021 Services ESP 74637 615406 1.013 50
## 6 1 2022 Services ESP 139750 627756 1.112 85b. Answer the questions
- How many firms are included in the sample?
table(Firm)## Firm
## 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
## 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
## 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
## 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6- How many years does the sample cover?
table(Year)## Year
## 2017 2018 2019 2020 2021 2022
## 51 51 51 51 51 51- It seems that the dataset is a balanced panel (51 firms with data over 6 years). Is it true?
library(plm)
is.pbalanced(dataset)## [1] TRUE- How many countries and how many industries does the sample comprise? What about the cross-country-industry distribution?
table(Country)/6## Country
## CZE ESP NED SWE
## 10 12 18 11table(Industry)/6## Industry
## ICT Manufacturing Mining Services Utility
## 5 13 12 14 7table(Country, Industry)/6## Industry
## Country ICT Manufacturing Mining Services Utility
## CZE 1 3 3 1 2
## ESP 0 3 2 6 1
## NED 2 2 5 6 3
## SWE 2 5 2 1 1- What are the basic descriptive statistics regarding Profit?
min(Profit)## [1] -41478median(Profit)## [1] 44216max(Profit)## [1] 195802mean(Profit)## [1] 53826.28sd(Profit)## [1] 43677.19- What are about the descriptive statistics regarding Profit across countries?
aggregate(Profit ~ Country, FUN = max)## Country Profit
## 1 CZE 161901
## 2 ESP 195802
## 3 NED 177800
## 4 SWE 181660aggregate(Profit ~ Country, FUN = min)## Country Profit
## 1 CZE -20361
## 2 ESP -2351
## 3 NED -17488
## 4 SWE -41478aggregate(Profit ~ Country, FUN = mean)## Country Profit
## 1 CZE 46665.45
## 2 ESP 59288.17
## 3 NED 55115.19
## 4 SWE 52268.59aggregate(Profit ~ Country, FUN = median)## Country Profit
## 1 CZE 34165.5
## 2 ESP 50939.5
## 3 NED 44900.0
## 4 SWE 41557.5aggregate(Profit ~ Country, FUN = sd)## Country Profit
## 1 CZE 40407.52
## 2 ESP 41743.17
## 3 NED 44107.13
## 4 SWE 47728.59- What are about the descriptive statistics regarding Profit across industries?
aggregate(Profit ~ Industry, FUN = max)## Industry Profit
## 1 ICT 159744
## 2 Manufacturing 181660
## 3 Mining 177800
## 4 Services 195802
## 5 Utility 179293aggregate(Profit ~ Industry, FUN = min)## Industry Profit
## 1 ICT -17488
## 2 Manufacturing -41478
## 3 Mining -20361
## 4 Services -14025
## 5 Utility -9245aggregate(Profit ~ Industry, FUN = mean)## Industry Profit
## 1 ICT 46424.43
## 2 Manufacturing 47364.54
## 3 Mining 59115.46
## 4 Services 57867.13
## 5 Utility 53964.83aggregate(Profit ~ Industry, FUN = median)## Industry Profit
## 1 ICT 32678.5
## 2 Manufacturing 36107.5
## 3 Mining 48511.0
## 4 Services 53417.5
## 5 Utility 43756.5aggregate(Profit ~ Industry, FUN = sd)## Industry Profit
## 1 ICT 42346.68
## 2 Manufacturing 42649.50
## 3 Mining 48335.25
## 4 Services 42439.02
## 5 Utility 40160.99- What about the descriptive statistics for all variables and across countries?
library(psych)
describeBy(dataset, Country)##
## Descriptive statistics by group
## group: CZE
## vars n mean sd median trimmed mad min
## Firm 1 60 38.80 12.10 42.00 41.75 5.19 5.00
## Year 2 60 2019.50 1.72 2019.50 2019.50 2.22 2017.00
## Industry* 3 60 3.00 1.28 3.00 3.00 1.48 1.00
## Country* 4 60 1.00 0.00 1.00 1.00 0.00 1.00
## Profit 5 60 46665.45 40407.52 34165.50 42318.92 27075.24 -20361.00
## Revenue 6 60 341549.33 153517.89 316265.00 333537.83 177058.76 118035.00
## Orders 7 60 0.99 0.03 0.99 0.99 0.04 0.92
## ESG 8 60 48.42 16.42 43.50 47.77 12.60 15.00
## max range skew kurtosis se
## Firm 49.00 44.00 -2.06 3.18 1.56
## Year 2022.00 5.00 0.00 -1.33 0.22
## Industry* 5.00 4.00 0.29 -1.04 0.16
## Country* 1.00 0.00 NaN NaN 0.00
## Profit 161901.00 182262.00 1.02 0.71 5216.59
## Revenue 727096.00 609061.00 0.43 -0.79 19819.07
## Orders 1.07 0.15 -0.03 -0.34 0.00
## ESG 85.00 70.00 0.36 -0.34 2.12
## ------------------------------------------------------------
## group: ESP
## vars n mean sd median trimmed mad min
## Firm 1 72 25.67 13.65 29.0 26.28 14.83 1.00
## Year 2 72 2019.50 1.72 2019.5 2019.50 2.22 2017.00
## Industry* 3 72 2.42 0.96 3.0 2.41 0.74 1.00
## Country* 4 72 1.00 0.00 1.0 1.00 0.00 1.00
## Profit 5 72 59288.17 41743.17 50939.5 54950.24 39711.44 -2351.00
## Revenue 6 72 391353.92 171825.62 402012.5 387911.74 230438.29 111562.00
## Orders 7 72 1.01 0.05 1.0 1.00 0.05 0.92
## ESG 8 72 50.86 14.77 49.0 49.95 12.60 17.00
## max range skew kurtosis se
## Firm 45.00 44.00 -0.38 -1.08 1.61
## Year 2022.00 5.00 0.00 -1.32 0.20
## Industry* 4.00 3.00 -0.33 -1.13 0.11
## Country* 1.00 0.00 NaN NaN 0.00
## Profit 195802.00 198153.00 0.98 0.89 4919.48
## Revenue 717552.00 605990.00 0.10 -1.27 20249.84
## Orders 1.21 0.29 0.96 1.10 0.01
## ESG 90.00 73.00 0.51 0.09 1.74
## ------------------------------------------------------------
## group: NED
## vars n mean sd median trimmed mad min
## Firm 1 108 21.50 10.91 21.50 20.86 10.38 3.00
## Year 2 108 2019.50 1.72 2019.50 2019.50 2.22 2017.00
## Industry* 3 108 3.33 1.21 3.50 3.41 0.74 1.00
## Country* 4 108 1.00 0.00 1.00 1.00 0.00 1.00
## Profit 5 108 55115.19 44107.13 44900.00 51342.93 40479.43 -17488.00
## Revenue 6 108 381606.18 171534.37 365409.50 379468.38 217206.09 107043.00
## Orders 7 108 1.02 0.06 1.02 1.02 0.06 0.87
## ESG 8 108 48.66 15.90 47.00 47.86 13.34 16.00
## max range skew kurtosis se
## Firm 48.00 45.00 0.50 0.05 1.05
## Year 2022.00 5.00 0.00 -1.30 0.17
## Industry* 5.00 4.00 -0.46 -0.65 0.12
## Country* 1.00 0.00 NaN NaN 0.00
## Profit 177800.00 195288.00 0.78 -0.05 4244.21
## Revenue 741595.00 634552.00 0.12 -1.23 16505.90
## Orders 1.14 0.26 0.08 -0.57 0.01
## ESG 95.00 79.00 0.46 0.20 1.53
## ------------------------------------------------------------
## group: SWE
## vars n mean sd median trimmed mad min
## Firm 1 66 22.09 16.95 16.0 21.11 17.79 2.00
## Year 2 66 2019.50 1.72 2019.5 2019.50 2.22 2017.00
## Industry* 3 66 2.45 1.17 2.0 2.33 1.48 1.00
## Country* 4 66 1.00 0.00 1.0 1.00 0.00 1.00
## Profit 5 66 52268.59 47728.59 41557.5 47235.26 44079.92 -41478.00
## Revenue 6 66 377650.41 160171.43 374480.5 375495.63 198617.25 122046.00
## Orders 7 66 1.01 0.06 1.0 1.00 0.05 0.89
## ESG 8 66 48.65 16.92 46.0 47.59 11.86 16.00
## max range skew kurtosis se
## Firm 51.00 49.00 0.56 -1.13 2.09
## Year 2022.00 5.00 0.00 -1.32 0.21
## Industry* 5.00 4.00 0.80 -0.19 0.14
## Country* 1.00 0.00 NaN NaN 0.00
## Profit 181660.00 223138.00 0.91 0.44 5874.98
## Revenue 704345.00 582299.00 0.10 -1.27 19715.74
## Orders 1.16 0.27 0.79 0.14 0.01
## ESG 97.00 81.00 0.67 0.40 2.08- How to visualise the cross-country differences in Profit graphically?
plot(as.factor(Country), Profit, main = "Boxplot for Profit", xlab="", ylab="")
- What is the difference between mean Profit for CZE and ESP?
library(dplyr)
CZE <- dataset %>% filter(Country == "CZE")
ESP <- dataset %>% filter(Country == "ESP")
mean(ESP$Profit) - mean(CZE$Profit)## [1] 12622.72- Is the difference between mean Profit for CZE and ESP significant?
A. Check whether data is normally distributed
shapiro.test(CZE$Profit)##
## Shapiro-Wilk normality test
##
## data: CZE$Profit
## W = 0.91879, p-value = 0.0006855shapiro.test(ESP$Profit)##
## Shapiro-Wilk normality test
##
## data: ESP$Profit
## W = 0.93388, p-value = 0.0009448B. Check the homogeneity of variance
- as the CZE and ESP subsamples are of different length (different number of firms), they need to be stacked, before running the variance tests
ProfitCZE <- CZE$Profit
ProfitESP <- ESP$Profit
stacked <- stack(list(ProfitCZE = ProfitCZE, ProfitESP = ProfitESP))- as data is not normally distributed (see the results of Shapiro-Wilk’s test in Step A), Levene’s test or Fligner-Killeen’s test shall be used
library(car)
leveneTest(values~ind,data=stacked,center=mean)## Levene's Test for Homogeneity of Variance (center = mean)
## Df F value Pr(>F)
## group 1 0.1445 0.7045
## 130fligner.test(values~ind, data=stacked)##
## Fligner-Killeen test of homogeneity of variances
##
## data: values by ind
## Fligner-Killeen:med chi-squared = 0.67492, df = 1, p-value = 0.4113C. Run the t-test
- as the subsamples has the same variance (see the results of Levene’s test or Fligner-Killeen’s test in Step B), the parametric t-test with equal variance is run
t.test(ProfitCZE, ProfitESP, var.equal = TRUE)##
## Two Sample t-test
##
## data: ProfitCZE and ProfitESP
## t = -1.7552, df = 130, p-value = 0.08159
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -26850.726 1605.293
## sample estimates:
## mean of x mean of y
## 46665.45 59288.17D. Alternatively, a nonparametric Mann Whitney U Test can be applied
wilcox.test(ProfitCZE, ProfitESP, exact = FALSE)##
## Wilcoxon rank sum test with continuity correction
##
## data: ProfitCZE and ProfitESP
## W = 1747, p-value = 0.05941
## alternative hypothesis: true location shift is not equal to 0please note that the Mann Whitney U test is a non-parametric test, not requiring the normality and variance tests (as it exploits directly the the original data)
please also note that the M-W U test is less efficient than the parametric t-test which may lead to a different conclusions (this is not the case for this dataset; however, note that p-value of the M-W U test is much closer to 0.05 than p-value of the t-test)
- Is there any impact of covid-19 on Profit of Czech firms (compare 2019 and 2020 data)?
# Filter data first
CZEP2019 <- CZE %>% filter(Year == "2019")
CZEP2020 <- CZE %>% filter(Year == "2020")
# Calculate the difference (for illustrative purposes)
mean(CZEP2020$Profit) - mean(CZEP2019$Profit) ## [1] -21779.3# Run the test (here only M-W U test for a paired sample)
wilcox.test(CZEP2019$Profit, CZEP2020$Profit, paired = TRUE)##
## Wilcoxon signed rank exact test
##
## data: CZEP2019$Profit and CZEP2020$Profit
## V = 38, p-value = 0.3223
## alternative hypothesis: true location shift is not equal to 0- Are there cross-industry differences in the impact of covid-19 on Profit in CZE and NED?
library(dplyr)
library(ggplot2)
# A graphical illustration using boxplots for each country and year
NED2019 <- ggplot(dataset %>% filter(Country == "NED", Year == 2019),
aes(Industry, Profit, fill = Industry)) +
geom_bar(stat = "summary", fun.y = "mean") +
labs(title = "NED2019", x = "") +
scale_fill_brewer(palette = "Dark2") +
ylim(0, 100000) +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
NED2020 <- ggplot(dataset %>% filter(Country == "NED", Year == 2020),
aes(Industry, Profit, fill = Industry)) +
geom_bar(stat = "summary", fun.y = "mean") +
labs(title = "NED2020", x = "") +
scale_fill_brewer(palette = "Dark2") +
ylim(0, 100000) +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
CZE2019 <- ggplot(dataset %>% filter(Country == "CZE", Year == 2019),
aes(Industry, Profit, fill = Industry)) +
geom_bar(stat = "summary", fun.y = "mean") +
labs(title = "CZE2019", x = "") +
scale_fill_brewer(palette = "Set2") +
ylim(0, 100000) +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
CZE2020 <- ggplot(dataset %>% filter(Country == "CZE", Year == 2020),
aes(Industry, Profit, fill = Industry)) +
geom_bar(stat = "summary", fun.y = "mean") +
labs(title = "CZE2020", x = "") +
scale_fill_brewer(palette = "Set2") +
ylim(0, 100000) +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
# Plotting all graphs together
library(patchwork)
(NED2019 + NED2020) / (CZE2019 + CZE2020)
- Developments in CZE are in line with expectations. But how do you explain that profits in the NED are higher in 2020 than in 2019 (except for manufacturing)?
Well, this is a good research question.
To answer it, a researcher should:
clearly identify the contribution of examining this question
review the relevant literature and develop a theory explaining the determinants of the heterogeneity of the impacts
formulate variables for the empirical model
to test the theory and compare the results with expectations and other relevant studies
5 Practice (the R-basics)
The following section provides the answers to the questions, but does not provide the code. It is up to you to develop it. Please note that there may be more ways to find the answer.
- What is the minimum for Revenue?
## [1] 107043- What is the standard deviation for Revenue in SWE?
## [1] 160171.4- What is the difference between the minimum and maximum ESG score in NED?
## vars n mean sd median trimmed mad min max range skew kurtosis se
## X1 1 108 48.66 15.9 47 47.86 13.34 16 95 79 0.46 0.2 1.53## [1] 79Visualise the development of ESG in SWE over time graphically
What is the difference between 2020 and 2019 Profit in ESP?
## [1] 20768.58- Is the increase in the 2020 Profit in ESP (compared to 2019) statistically significant?
##
## Wilcoxon signed rank exact test
##
## data: ESPP2019$Profit and ESPP2020$Profit
## V = 21, p-value = 0.1763
## alternative hypothesis: true location shift is not equal to 0What is the relation between Orders and Profit (with a cross-country and cross-industry breakdown)?
How strong is the association between Orders and Profit (separately for each country and industry)?
