Project Title: “Analysis of Factors Affecting Female Employment Rate”

NAME: Akshay Ratnawat

EMAIL: akshayratnawat@gmail.com

COLLEGE: Jawaharlal Nehru University (SSS)

Introduction

India has experienced rapid economic growth in the recent past alongwith the structural shifts in the economythere has been an increase in educational attainment levels, and rapid urbanization in the last twenty five years. In the same period there has also been a 23% decline in the female labour force participation rate. What’s the relationship between economic growth and women’s economic activity? Is growth enough or does the nature of growth matter in attracting more women to the labour force? (1)

Female employment has become a crucial issue in today’s world. Female when employed not only leads to increased income of the family but it also leads to increase in overall development of the family. It has been found in repeated research that increased female employment leads to better and more allocation of resources towards health and education of children and family. Thus increased female employment has been found to be related to improved life in a country(2).

Further as income of females increase they find themself more secure and have access to large number of services. This access to service makes them more capable to gain the much required freedom and independence especially in countries like India.

Overview

As per various studies and reasearch we see a wierd pattern in the female employment rate in a country. It has been found that when the income of the family rises women are the first one to leave the job market. Thiss can be seen from the scatter plot given below that this relation is not much correlated with each other. As the income perperson rises the female employment rises with a slower rate. This is quite paradoxical as when incoem rises we expect the people to be more open and thus expect more women in the labor force. (Appendix 1)

Similarly we see that another factor which is crucial in deciding the female employment rate is the urbanisation rate. The female employment rate should increase with the increase in the female employment rate in a country. But on the other context we find just the opposite happening. There is more female employment in the countries having low urbanization rate. There might be various reasons for this. (Appendix 2)

Further we also find that there is a positive relationship between the polity score and female employment rate. As the democracy strengthens in a country the female employment also rises in the country. This was very well expected and it has been proved by the data as well.

Data

For this model we took data from the gapminder website.

References

  1. https://iimb.ac.in/research/sites/default/files/WP%20No.%20414_0.pdf
  2. https://www.imf.org/external/pubs/ft/sdn/2013/sdn1310.pdf
table <- read.csv(paste("gapminder.csv", sep = ""))
library(psych)
describe(table)[c(2,3,7,9,10,12,13,16),]
##                    vars   n    mean       sd  median trimmed     mad
## incomeperperson       2 190 8740.97 14262.81 2553.50 5610.69 3285.71
## alcconsumption        3 187    6.69     4.90    5.92    6.30    5.53
## femaleemployrate      7 178   47.55    14.63   47.55   47.57   12.68
## internetuserate       9 192   35.63    27.78   31.81   33.70   32.76
## lifeexpectancy       10 191   69.75     9.71   73.13   70.75    8.31
## polityscore          12 161    3.69     6.31    6.00    4.36    5.93
## relectricperperson   13 136 1173.18  1681.44  597.14  821.25  778.23
## urbanrate            16 203   56.77    23.84   57.94   56.73   28.70
##                       min       max     range  skew kurtosis      se
## incomeperperson    103.78 105147.44 105043.66  3.20    14.07 1034.73
## alcconsumption       0.03     23.01     22.98  0.61    -0.25    0.36
## femaleemployrate    11.30     83.30     72.00  0.02     0.07    1.10
## internetuserate      0.21     95.64     95.43  0.45    -1.08    2.00
## lifeexpectancy      47.79     83.39     35.60 -0.81    -0.47    0.70
## polityscore        -10.00     10.00     20.00 -0.71    -0.97    0.50
## relectricperperson   0.00  11154.76  11154.76  3.11    12.02  144.18
## urbanrate           10.40    100.00     89.60 -0.02    -1.02    1.67

From the above Table we have following observations:

  1. There is a lot od deviation in the incomeper person across various countries. There is a lot of difference between the minimum and maximum income.

  2. The female employment rate has very less deviation in various countries which is very low thus its a problem of worry. After the Christian Lagard comment that using the female resources in the employment countries all over the world can increase the world GDP by 2%, countries should be seen working in this direction.

  3. Internet USe Rate Life Expectancy Rate has been used here as the proxy of development in the countries. The deviation in the in the internet use rate is highly dispersed across countries as this is a recent development in various countries. But the deviation between the life expectancy is not so much as the Internet use rate. In these two we can say that life expectancy has been a first round of development and countries are fast catching up in this. While the internet use rate is the second round of development which we can see fast to be converging in the near future.

  4. Polity Score has been found as a crucial measure of the polity structure in the country. It has been calculated as autocracy score subtracted from the democracy score. Thus it measures the level of free nature and democracy in the country. Thus we can see it as a crucial factor affecting the female employment rate.

  5. The Residential electricity also has large deviation. We can see this also as a crucial factor affecting the employment rate. As in a country the residential electricity consumption can be seen as a proxy of the street lighting and public place lighting as well, as both these in a country are correlated to each other.

  6. Urbanisation has a large impact on the culture, freedom to women etc. which in turn affect the female employment rate in a country. So we can see this also as a crucial factor in determining the female employment rate.

Hypothesis : We want to test the Hypothesis that there is no effect of the urbanisation rate, polity score, internet use rate on the female employment rate in the country.

model1 <- femaleemployrate ~ urbanrate + relectricperperson + lifeexpectancy + internetuserate + polityscore
fit <- lm(femaleemployrate ~ urbanrate + relectricperperson + lifeexpectancy + internetuserate + polityscore, data = table)
summary(fit)
## 
## Call:
## lm(formula = femaleemployrate ~ urbanrate + relectricperperson + 
##     lifeexpectancy + internetuserate + polityscore, data = table)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -30.8591  -7.9717   0.6151   7.5713  26.4308 
## 
## Coefficients:
##                      Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        90.5101367 12.3317894   7.340 3.08e-11 ***
## urbanrate          -0.1786006  0.0722889  -2.471  0.01493 *  
## relectricperperson  0.0002936  0.0008804   0.333  0.73939    
## lifeexpectancy     -0.6044115  0.2006505  -3.012  0.00318 ** 
## internetuserate     0.1966739  0.0772633   2.546  0.01221 *  
## polityscore         0.4005866  0.2020992   1.982  0.04981 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 12.47 on 117 degrees of freedom
##   (90 observations deleted due to missingness)
## Multiple R-squared:  0.1781, Adjusted R-squared:  0.143 
## F-statistic: 5.071 on 5 and 117 DF,  p-value: 0.0003041
library(leaps)
leap1 <- regsubsets(model1, data = table, nbest=1)
# summary(leap1)
plot(leap1, scale="adjr2")

The best fit model excludes the residential electricity per person.

model2 <- femaleemployrate ~ urbanrate +  lifeexpectancy + internetuserate + polityscore
fit1 <- lm(femaleemployrate ~ urbanrate  + lifeexpectancy + internetuserate + polityscore, data = table)
summary(fit1)
## 
## Call:
## lm(formula = femaleemployrate ~ urbanrate + lifeexpectancy + 
##     internetuserate + polityscore, data = table)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -33.752  -9.485   0.533   8.999  31.117 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)     89.68577   10.54253   8.507 1.88e-14 ***
## urbanrate       -0.23702    0.07004  -3.384 0.000916 ***
## lifeexpectancy  -0.52070    0.18209  -2.860 0.004860 ** 
## internetuserate  0.19380    0.06959   2.785 0.006057 ** 
## polityscore      0.22873    0.19037   1.202 0.231472    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 13.55 on 147 degrees of freedom
##   (61 observations deleted due to missingness)
## Multiple R-squared:  0.1842, Adjusted R-squared:  0.162 
## F-statistic: 8.298 on 4 and 147 DF,  p-value: 4.613e-06

Thus from the model we find the following insights :

  1. Urban rate is negatively related to the female employment rate. As the urbanisation rises the female employment rate falls and it is significant.

  2. Life expectancy is also negatively related to the female employment rate. But here once thing to be noted is that from the plot its quite clear that more countries are clustered around the points of high female employment rate and high life expectancy.

  3. The internetuserate taken as proxy of development has found to be positively related with the female employment rate as expected.

4.The polity score has found to positvely related to the female employment rate but it is not significant .

  1. The overall p value of the F test statistic is found to be quite small which shows that the model is a good fit if not the best.

Thus overall we found that there is quite paradoxical situation. The factors which we thought would affect female employment rate has shown to be negatively realted. There might be various reasons for this in various countries. In India we see that as the urbanisation rises the female employment either falls or remains stagnant. This behaviour is so because Indian cities are yet not gender friendly, which obstructs women from leaving homes. Further most of the migration from the rural areas to urban aras is of male members and females get engaged in the agricultural work. Further most of the women in India work in the informal sector which is not taken into account.

Thus female employment is required to be increased in India if we want to improve the overall status of women and development in India. For this we need to increase their involvement in the economy. Policy recommendations for the government from this small model are:

  1. GOvenment must focus on improving the development in the country as with the increase in development the female employment rises.

  2. More democractic institutions and policies also have a positive impact on female employment.

  3. Government must see that as the female employment rises the quality of life also rises. So improving female employment rate can be good on two folds for the country.

boxplot(table$incomeperperson,
        main = "Income Per Person",
        horizontal = TRUE,
        col = " Blue",
        xlab = "2010 Gross Domestic Product per capita in constant 2000 US$.")

As we see from the graph the Income per person has a lot of outliers.

boxplot(table$femaleemployrate,
        main = " Female employment Rate",
        horizontal = TRUE,
        col = "linen",
        xlab= " Female Employees (15+) % of Female population")

boxplot(table$internetuserate,
        main = "Internet Use Rate",
        horizontal = TRUE,
        col = "blueviolet",
        xlab = " Internet Users per 100 persons")

boxplot(table$polityscore,
        main = "Polity Score",
        horizontal = TRUE,
        col = "blueviolet",
        xlab = "2009 Democracy score (Polity)")

library(ggplot2)
## 
## Attaching package: 'ggplot2'
## The following objects are masked from 'package:psych':
## 
##     %+%, alpha
theme_set(theme_bw())
ggplot(table, aes(x=`country`, y=polityscore, label=country)) + 
  geom_point(stat='identity', fill="black", size=4)  +
  geom_segment(aes(y = 0, 
                   x = `country`, 
                   yend = polityscore, 
                   xend = `country`), 
               color = "black") +
  geom_text(color="white", size=1) +
  labs(title="Diverging Polity Score") + 
  ylim(-10, 10) +
  coord_flip()
## Warning: Removed 52 rows containing missing values (geom_point).
## Warning: Removed 52 rows containing missing values (geom_segment).
## Warning: Removed 52 rows containing missing values (geom_text).

boxplot(table$relectricperperson,
        main = "Residential Electricity Per Person",
        horizontal = TRUE,
        col = "mediumseagreen",
        xlab = "2008 residential electricity consumption, per person (kWh)")

boxplot(table$urbanrate,
        main = "Urbanisation Rate",
        horizontal = TRUE,
        col = "seagreen",
        xlab = "2008 urban population (% of total)")

reg <- lm(table$femaleemployrate ~ table$incomeperperson)
plot(table$incomeperperson+1, table$femaleemployrate+1,
     ylab = "Female Employment Rate",
     xlab = "Income Per Person",
     main = "Appendix 1 : Relation Between Female Employment Rate and Income Per person",
     col = "Red")
abline (reg, col = "black")

reg5 <- lm(table$femaleemployrate ~ table$urbanrate)
plot(table$urbanrate, table$femaleemployrate, 
     xlab = "Female Employment Rate",
     ylab = "Urbanisation Rate",
     main = "Appendix 2: Relation Between Female Employment Rate and Urbanisation Rate",
     col = "Blue")
abline(reg5, col ="Red")

reg3 <- lm(table$femaleemployrate ~ table$polityscore)
plot(table$polityscore, table$femaleemployrate, 
     xlab = "Female Employment Rate",
     ylab = "Polity score of Democracy",
     main = "Appendix 3: Relation Between Female Employment Rate and Polity Score",
     col = "green")
abline (reg3, col = "black")

reg2 <- lm(table$femaleemployrate ~ table$lifeexpectancy)
plot(table$lifeexpectancy, table$femaleemployrate, 
     xlab = "Female Employment Rate",
     ylab = "Life Expectancy Rate",
     main = "Relation Between Female Employment Rate and Life Expectancy Rate",
     col = "black")
abline (reg2, col = "Red")

reg1 <- lm(table$femaleemployrate ~ table$internetuserate)
plot(table$internetuserate, table$femaleemployrate, 
     xlab = "Female Employment Rate",
     ylab = "Internet Usage Rate",
     main = "Relation Between Female Employment Rate and Internet Usage Rate",
     col = "Blue")
abline (reg1, col = "red")

reg4 <- lm(table$femaleemployrate ~ table$relectricperperson)
plot( table$relectricperperson+1, table$femaleemployrate+1,
     xlab = "Female Employment Rate",
     ylab = "Residential Electricity Per Person",
     main = "Relation Between Female Employment Rate and Residential Electricity per person",
     col = "Blue")
abline (reg4, col = "red")

library(car)
## 
## Attaching package: 'car'
## The following object is masked from 'package:psych':
## 
##     logit
scatterplotMatrix (formula = ~femaleemployrate + internetuserate + lifeexpectancy + polityscore + relectricperperson + urbanrate, 
                   data = table,
                   main = " Scatterplot Matrix of all the crucial Factors")

library(corrgram)
vars2 <- c("femaleemployrate", "internetuserate", "lifeexpectancy", "polityscore", "relectricperperson", "urbanrate")
corrgram(table[,vars2], order=TRUE,
         main="Factor affecting female employment rate",
         lower.panel=panel.shade, upper.panel=panel.pie,
         diag.panel=panel.minmax, text.panel=panel.txt)