In order to analyse the factors that influence CSR we utilise the CSR dataset, first upload it.
library(readxl)
CSR_data <- read_excel("~/Desktop/UPF - MSc Management and BuA/SECOND Term/Benchmarking and Management Control/Benchmarking R FILES/CSR_data.xlsx") It is a set of 6580 observation with 90 variables. By looking at the data we see that companies in the set repeating over a 3 year period 2004-2005-2005. Hence I decided to subset them before working on them.
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionyear04 <- CSR_data %>% filter(year == 2004)
year05 <- CSR_data %>% filter(year == 2005)
year06 <- CSR_data %>% filter(year == 2006)
print(year04) ## # A tibble: 2,071 x 90
## n year name isocountrycode mnemonic ftseindsector ftseindsubsctr
## <dbl> <dbl> <chr> <chr> <chr> <dbl> <dbl>
## 1 3916 2004 TOKY… JP J:TZ@N 75 757
## 2 63 2004 ADVA… JP J:AB@N 95 957
## 3 1247 2004 DTE … US U:DTE 75 753
## 4 2588 2004 MITS… JP J:JS@N 13 135
## 5 1158 2004 DELH… BE B:DEH 53 533
## 6 2959 2004 PARK… US U:PH 27 275
## 7 1952 2004 IBER… ES E:IBL 57 575
## 8 2066 2004 INVE… GB INVP 87 877
## 9 1267 2004 E*TR… US @ETFC 87 877
## 10 3425 2004 SEKI… JP J:SE@N 37 372
## # … with 2,061 more rows, and 83 more variables: economicscore <dbl>,
## # socialscore <dbl>, environmentalscore <dbl>,
## # bookvalueoutsharesfiscal <dbl>, commonsharesoutstanding <dbl>,
## # numberofshares <dbl>, marketvalue <dbl>, marketcapitalization <dbl>,
## # totalassets <dbl>, operatingincome <dbl>, longtermdebt <dbl>,
## # totalintangibleotassetsnet <dbl>, totalliabilities <dbl>,
## # netsalesorrevenues <dbl>, operatingexpensestotal <dbl>,
## # netincomebasic <dbl>, researchdevelopment <dbl>, employees <dbl>,
## # earningspershare <dbl>, netmargin <dbl>, operatingprofitmargin <dbl>,
## # returnonequitytotal <dbl>, returnoninvestedcapital <dbl>,
## # cashflowsales <dbl>, epsbasicyear <dbl>, singlebiggestowner <dbl>,
## # boardsize <dbl>, valueboardstructureindependentbo <dbl>, ceoduality <dbl>,
## # chairmanexceoofthecompany <dbl>, chairmanisexceo <dbl>,
## # ceoboardmember <dbl>, valueboardstructureboardgenderdi <dbl>,
## # ceocompensationlinktototalshareh <dbl>, goldenparachute <dbl>,
## # otherantitakeoverdevices <dbl>, lnta <dbl>, ROA <dbl>,
## # Leverage_liabassets <dbl>, anglo <dbl>, pdi <dbl>, idv <dbl>, mas <dbl>,
## # uai <dbl>, egalitarianism <dbl>, harmony <dbl>, embeddedness <dbl>,
## # namecountry <chr>, strengthofinvestorprotectioninde <dbl>,
## # investmntcoheld <dbl>, strategicholdings <dbl>, tobinq <dbl>,
## # zcountryname <chr>, zcountrycode <chr>, marketcapl <dbl>, gdpgrowthl <dbl>,
## # gdppercapital <dbl>, domesticcreditl <dbl>, healthexpenditurel <dbl>,
## # femaleparticipationl <dbl>, womeninparliamentl <dbl>,
## # board_independence <dbl>, women_on_board <dbl>, soc_env_score <dbl>,
## # soc_env_ec_score <dbl>, y2004 <dbl>, y2005 <dbl>, y2006 <dbl>,
## # stakelaw <dbl>, csrlaw <dbl>, guillen_capron <dbl>, sizeinind <dbl>,
## # lnsizeinind <dbl>, lnmarketcapl <dbl>, lngdppercapital <dbl>,
## # lnhealthexpenditurel <dbl>, countrycode <chr>, co2emissions <dbl>,
## # greenhousegas <dbl>, renew_energy_con <dbl>, gov_exp_education <dbl>,
## # labor_force_part <dbl>, women_parliament <dbl>print(year05)## # A tibble: 2,233 x 90
## n year name isocountrycode mnemonic ftseindsector ftseindsubsctr
## <dbl> <dbl> <chr> <chr> <chr> <dbl> <dbl>
## 1 473 2005 BARR… GB BDEV 37 372
## 2 3237 2005 REST… GB RTN 57 575
## 3 3839 2005 TENE… US U:THC 45 453
## 4 3045 2005 PITN… US U:PBI 95 957
## 5 364 2005 AVAL… US U:AVB 87 873
## 6 1821 2005 HEND… GB HGG 87 877
## 7 27 2005 ACCI… ES E:ANA 23 235
## 8 2225 2005 KING… US <NA> 45 457
## 9 906 2005 CIENA US @CIEN 95 957
## 10 1025 2005 CONV… US U:CVG 27 279
## # … with 2,223 more rows, and 83 more variables: economicscore <dbl>,
## # socialscore <dbl>, environmentalscore <dbl>,
## # bookvalueoutsharesfiscal <dbl>, commonsharesoutstanding <dbl>,
## # numberofshares <dbl>, marketvalue <dbl>, marketcapitalization <dbl>,
## # totalassets <dbl>, operatingincome <dbl>, longtermdebt <dbl>,
## # totalintangibleotassetsnet <dbl>, totalliabilities <dbl>,
## # netsalesorrevenues <dbl>, operatingexpensestotal <dbl>,
## # netincomebasic <dbl>, researchdevelopment <dbl>, employees <dbl>,
## # earningspershare <dbl>, netmargin <dbl>, operatingprofitmargin <dbl>,
## # returnonequitytotal <dbl>, returnoninvestedcapital <dbl>,
## # cashflowsales <dbl>, epsbasicyear <dbl>, singlebiggestowner <dbl>,
## # boardsize <dbl>, valueboardstructureindependentbo <dbl>, ceoduality <dbl>,
## # chairmanexceoofthecompany <dbl>, chairmanisexceo <dbl>,
## # ceoboardmember <dbl>, valueboardstructureboardgenderdi <dbl>,
## # ceocompensationlinktototalshareh <dbl>, goldenparachute <dbl>,
## # otherantitakeoverdevices <dbl>, lnta <dbl>, ROA <dbl>,
## # Leverage_liabassets <dbl>, anglo <dbl>, pdi <dbl>, idv <dbl>, mas <dbl>,
## # uai <dbl>, egalitarianism <dbl>, harmony <dbl>, embeddedness <dbl>,
## # namecountry <chr>, strengthofinvestorprotectioninde <dbl>,
## # investmntcoheld <dbl>, strategicholdings <dbl>, tobinq <dbl>,
## # zcountryname <chr>, zcountrycode <chr>, marketcapl <dbl>, gdpgrowthl <dbl>,
## # gdppercapital <dbl>, domesticcreditl <dbl>, healthexpenditurel <dbl>,
## # femaleparticipationl <dbl>, womeninparliamentl <dbl>,
## # board_independence <dbl>, women_on_board <dbl>, soc_env_score <dbl>,
## # soc_env_ec_score <dbl>, y2004 <dbl>, y2005 <dbl>, y2006 <dbl>,
## # stakelaw <dbl>, csrlaw <dbl>, guillen_capron <dbl>, sizeinind <dbl>,
## # lnsizeinind <dbl>, lnmarketcapl <dbl>, lngdppercapital <dbl>,
## # lnhealthexpenditurel <dbl>, countrycode <chr>, co2emissions <dbl>,
## # greenhousegas <dbl>, renew_energy_con <dbl>, gov_exp_education <dbl>,
## # labor_force_part <dbl>, women_parliament <dbl>print(year06)## # A tibble: 2,276 x 90
## n year name isocountrycode mnemonic ftseindsector ftseindsubsctr
## <dbl> <dbl> <chr> <chr> <chr> <dbl> <dbl>
## 1 705 2006 CAPI… GB CSCG 87 873
## 2 2527 2006 MERC… CH S:SEO 48 482
## 3 3926 2006 TOMR… NO N:TOM 27 275
## 4 3978 2006 TRAV… US U:TRV 85 853
## 5 1089 2006 CSR GB CSR 95 957
## 6 3355 2006 SANO… FR F:SQ@F 45 457
## 7 3315 2006 SACY… ES E:VAL 23 235
## 8 1398 2006 ESSI… FR F:EI 45 453
## 9 163 2006 ALLI… US <NA> 27 279
## 10 1296 2006 EDIS… IT I:EDN 77 775
## # … with 2,266 more rows, and 83 more variables: economicscore <dbl>,
## # socialscore <dbl>, environmentalscore <dbl>,
## # bookvalueoutsharesfiscal <dbl>, commonsharesoutstanding <dbl>,
## # numberofshares <dbl>, marketvalue <dbl>, marketcapitalization <dbl>,
## # totalassets <dbl>, operatingincome <dbl>, longtermdebt <dbl>,
## # totalintangibleotassetsnet <dbl>, totalliabilities <dbl>,
## # netsalesorrevenues <dbl>, operatingexpensestotal <dbl>,
## # netincomebasic <dbl>, researchdevelopment <dbl>, employees <dbl>,
## # earningspershare <dbl>, netmargin <dbl>, operatingprofitmargin <dbl>,
## # returnonequitytotal <dbl>, returnoninvestedcapital <dbl>,
## # cashflowsales <dbl>, epsbasicyear <dbl>, singlebiggestowner <dbl>,
## # boardsize <dbl>, valueboardstructureindependentbo <dbl>, ceoduality <dbl>,
## # chairmanexceoofthecompany <dbl>, chairmanisexceo <dbl>,
## # ceoboardmember <dbl>, valueboardstructureboardgenderdi <dbl>,
## # ceocompensationlinktototalshareh <dbl>, goldenparachute <dbl>,
## # otherantitakeoverdevices <dbl>, lnta <dbl>, ROA <dbl>,
## # Leverage_liabassets <dbl>, anglo <dbl>, pdi <dbl>, idv <dbl>, mas <dbl>,
## # uai <dbl>, egalitarianism <dbl>, harmony <dbl>, embeddedness <dbl>,
## # namecountry <chr>, strengthofinvestorprotectioninde <dbl>,
## # investmntcoheld <dbl>, strategicholdings <dbl>, tobinq <dbl>,
## # zcountryname <chr>, zcountrycode <chr>, marketcapl <dbl>, gdpgrowthl <dbl>,
## # gdppercapital <dbl>, domesticcreditl <dbl>, healthexpenditurel <dbl>,
## # femaleparticipationl <dbl>, womeninparliamentl <dbl>,
## # board_independence <dbl>, women_on_board <dbl>, soc_env_score <dbl>,
## # soc_env_ec_score <dbl>, y2004 <dbl>, y2005 <dbl>, y2006 <dbl>,
## # stakelaw <dbl>, csrlaw <dbl>, guillen_capron <dbl>, sizeinind <dbl>,
## # lnsizeinind <dbl>, lnmarketcapl <dbl>, lngdppercapital <dbl>,
## # lnhealthexpenditurel <dbl>, countrycode <chr>, co2emissions <dbl>,
## # greenhousegas <dbl>, renew_energy_con <dbl>, gov_exp_education <dbl>,
## # labor_force_part <dbl>, women_parliament <dbl>Unfortunately the observation for each year do not match. In order to avoid any mistake due to subsets’ mistake we follow with an approach in which we put in relation variables indipendently from the year subset.
Here we decided to put in relation with our dependent variable “soc_env_score” the follwing indipendent variables: “femaleparticipationl”, “healthexpenditurel”, “co2emission” and "BoardInd.
We rename them for tidiness purposes
CSRlevel <- CSR_data$soc_env_score
FEMpart <- CSR_data$femaleparticipationl
HEALTHexp <- CSR_data$healthexpenditurel
CO2 <- CSR_data$co2emissions
BoardInd <- CSR_data$board_independenceWe run different singluar simpl regressions to see there is any stastical significance and correlation to test if our hypothesis were right.
reg1 <- lm(CSRlevel ~ FEMpart)
summary(reg1)##
## Call:
## lm(formula = CSRlevel ~ FEMpart)
##
## Residuals:
## Min 1Q Median 3Q Max
## -50.956 -26.009 -5.005 26.869 53.342
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 84.26834 3.61948 23.282 <2e-16 ***
## FEMpart -0.65970 0.06659 -9.906 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 28.5 on 6563 degrees of freedom
## (15 observations deleted due to missingness)
## Multiple R-squared: 0.01473, Adjusted R-squared: 0.01458
## F-statistic: 98.14 on 1 and 6563 DF, p-value: < 2.2e-16reg2 <- lm(CSRlevel~ HEALTHexp)
summary(reg2)##
## Call:
## lm(formula = CSRlevel ~ HEALTHexp)
##
## Residuals:
## Min 1Q Median 3Q Max
## -44.777 -26.700 -4.691 27.648 53.445
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 52.5165250 0.5683821 92.397 < 2e-16 ***
## HEALTHexp -0.0013206 0.0001628 -8.113 5.88e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 28.6 on 6418 degrees of freedom
## (160 observations deleted due to missingness)
## Multiple R-squared: 0.01015, Adjusted R-squared: 0.009997
## F-statistic: 65.82 on 1 and 6418 DF, p-value: 5.881e-16reg3 <- lm(CSRlevel ~ CO2)
summary(reg3)##
## Call:
## lm(formula = CSRlevel ~ CO2)
##
## Residuals:
## Min 1Q Median 3Q Max
## -50.501 -24.201 -4.131 25.553 57.687
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 64.63846 0.91616 70.55 <2e-16 ***
## CO2 -1.25392 0.06628 -18.92 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 27.96 on 6563 degrees of freedom
## (15 observations deleted due to missingness)
## Multiple R-squared: 0.05172, Adjusted R-squared: 0.05158
## F-statistic: 358 on 1 and 6563 DF, p-value: < 2.2e-16reg4 <- lm(CSRlevel ~ BoardInd)
summary(reg4)##
## Call:
## lm(formula = CSRlevel ~ BoardInd)
##
## Residuals:
## Min 1Q Median 3Q Max
## -48.904 -26.457 -2.149 27.275 45.572
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 55.77382 1.08426 51.440 <2e-16 ***
## BoardInd -0.04049 0.01686 -2.401 0.0164 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 28.14 on 3141 degrees of freedom
## (3437 observations deleted due to missingness)
## Multiple R-squared: 0.001833, Adjusted R-squared: 0.001515
## F-statistic: 5.767 on 1 and 3141 DF, p-value: 0.01639all these three regression show a strong stastically siginifcant relatio between CSRlevel and singulralry in order FEMpart, HEALTHexp and CO2.
Here we run a multiple regression to see if the assumptions still holds.
MULTreg <- lm (CSRlevel~ FEMpart + HEALTHexp + CO2 + BoardInd )
summary (MULTreg)##
## Call:
## lm(formula = CSRlevel ~ FEMpart + HEALTHexp + CO2 + BoardInd)
##
## Residuals:
## Min 1Q Median 3Q Max
## -61.011 -23.680 -1.852 24.227 60.149
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 77.5484753 5.9763353 12.976 < 2e-16 ***
## FEMpart -0.1889020 0.1238697 -1.525 0.127
## HEALTHexp 0.0009231 0.0002197 4.202 2.72e-05 ***
## CO2 -1.7544745 0.1170239 -14.992 < 2e-16 ***
## BoardInd 0.1481291 0.0204219 7.253 5.10e-13 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 26.85 on 3119 degrees of freedom
## (3456 observations deleted due to missingness)
## Multiple R-squared: 0.09286, Adjusted R-squared: 0.0917
## F-statistic: 79.82 on 4 and 3119 DF, p-value: < 2.2e-16By running the multiple regression the stastical significance of female participation decreases and therefore we could assume that it does not make a bih change in terms of CSR level for a company. On the other hand CO2 emmisions are negatively correlated meaning that the higher the CO2 level the lower is the CSR of a company, which is logial and make sense (the more you pollute the worse you score in terms of Corporate Social Responsability). Health exp is also negatively correlated, Country level health expenditure as percentage of GSD does not affect positively this regression.
Hence we answer the following questions:
1- Does an high female participation affect positively the level of Corporate Social Responsability of an organisation? Not statisically siginificant, and it loses even more significance when put in a multiple regression, hence it is not correlate positively.
2- Does an high health expenditure affect positively the level of Corporate Social Responsability of an organisation? Negatively
3- Does an high CO2 emmission affect positively the level of Corporate Social Responsability of an organisation? This is obviously negative as CSR also relies on how green a company is.
4 - Does an high board indipendence affect positively the level of Corporate Social Responsability of an organisation? in the somple linear regression it has a stastical siginificance of 95% and it affects negatively CSRlevel. When put in the multiple regression it becomes postive and its siginficance increase to 99%. In this sense we can assume that board indipendence has a more CSR oriented view, with no conflict of interests and it is a positive correlatio.