# Research on Factors Affecting HDI in China: Evidence from 31 Provinces

# Introduction

In GSID, we always study and research why some countries are rich while some countries are poor. Development is a worldwide theme, if we want to answer the question above, we need to know the definition of development firstly. As Professor Otsubo said, development is a transformation of the structure as a whole that causes material and mental poverty. And the development focus had shifted from economic development to human development since the 1990s. According to Amartya Sen, it is important that people acquire the capability, or escape from social isolation and have companions, thereby, people are empowered; economic development is a means, the object is human development. In other words, if one country wants to achieve long-term and stable development, the all-around development of human beings will be the dispensable and essential topic.

Some scholars use HDI to evaluate the human development level of one country or region. HDI (Human Development Index) is a statistic composite index of life expectancy, education (mean years of schooling completed and expected years of schooling upon entering the education system), and per capita income indicators, which is used to rank countries into four tiers of human development. In this proposal, it takes 31 provinces as an example to research on the factors affecting HDI in China; establishes a regression model with HDI as the dependent variable, and life expectancy, education and per capita income as the independent variables; utilizes regression analysis to estimate the different impacts of these three factors on HDI and make policy recommendations accordingly.

# Methodology&Data

Methodology and dataset will be introduced in this part. Because HDI consists of life expectancy, education level and per capita income indicators. Therefore, the regression model will be created based on HDI and these three variables, respectively. Besides, the data will be chosen from 1990 to 2019.

## Linear Regression Model

y=β0+β1*x1+ε

y is the national HDI in China, x1 is the log GNI per capita, or life expectancy at birth, or education level , β0 is the constant term; β1 is the coefficient of independent variable x1; ε is the error term.

## Import Data from Global Data Lab

Data Source: https://globaldatalab.org/

The dataset not only includes national data but also contains subnational data from more than 184 countries and regions within the period between 1990 and 2019. There are 37 variables including the four variables (shdi, lifexp, edindex, lgnic), which will be used in the following part.

## Prepare Data

Select the data of China including the four variables, shdi, lifexp, edindex, and lgnic.

Select the national-level data from 1990 to 2019.

## Descriptive Statistics

# Exploratory Data Analysis

## Line Plot: Regional Differences of HDI in China

Based on the line plot, we can get the information that HDI in all regions tended to increase over time. Beijing, Shanghai, Tianjin are the top three provinces with the highest HDI in China from 1990 to 2019. However, Chongqing, which is also one of the four major municipalities, although its HDI began to grow up quickly since 2004, is still in the middle level among these 31 regions.

And, there are obvious development differences among regions, such as Beijing and Tibet, but the development differences are consistent with the economic development differences, which means higher regional output might bring out higher regional HDI. We will prove this hypothesis in the following regression analysis.

## Choropleth Map: International Difference of HDI

As time goes by, lots of countries' HDI increase obviously from 1990 to 2019. From continental perspective, the HDI of North America, Western Europe and Oceania do not change obviously and still keep at higher level since 1990. As we all know, most developed countries locate at these continents.

Among developing countries, the HDI of the northern part of South America, East Asia, Southeast Asia and Middle East, Northern and Southern parts of Africa grow up apparently, whlie the HDI of Central Africa and Southern Asia (like Afghanistan and Pakistan) do not fluctuate and still at lower level. They still have a long way to go, poverty eradication may be the first step.

## Box Plot: National HDI of China as time goes by

According to the box plot, the difference between maximum and minimum becomes smaller and smaller under the condition without the influence of extremums, which means regional differences of HDI in China are generally decreasing. Moreover, the median is getting closer and closer to upper fence, so more than 50% of regions grow more and more quickly on the way to upper level.

However, there are some extreme values, we need to talk about them separately. In terms of minimal values, Tibet is always the minimum, while Guizhou is below the lower fence in 2004, 2007, 2009, 2010 and 2012, Yunnan and Qinghai are below the lower fence in 2013 and 2015, respectively. These three provinces also have the lower level of economic development among all regions.

As for maximum values, Beijing and Shanghai are always the top two provinces, whose HDI are above the upper fence from 1990, while the difference between them becomes obvious since 2010. To be more specific, Beijing's HDI is lower than Shanghai's from 1990 to 2000, it may because of the development policy in Shanghai Pudong New District since 1990. But this trend has been reversed after 2001. Tianjin exceeds the upper fence from 2003, its promotion might be closely related to the economic strategy of Tianjin Binhai New Area.

# Regression Analysis

## The Impact of log GNI per capita on HDI

H0 (Null Hypothesis): there are no correlations between HDI and log GNI per capita.

H1 (Alternative Hypothesis): there has correlations between HDI and log GNI per capita.

Based on the regression output, we can reject the H0 under more than 95% confidence interval.

HDI=0.1131*log GNI per capita-0.3311

## The Impact of Life Expectancy on HDI

H0 (Null Hypothesis): there are no correlations between HDI and life expectancy.

H1 (Alternative Hypothesis): there has correlations between HDI and life expectancy.

Based on the regression output, we can reject the H0 under more than 95% confidence interval.

HDI=0.0333*life expectancy at birth-1.7861

However, there might have strong multicollinearity or other numerical problems due to the large condition number. I do not know the reason clearly, but I guess it may be because that the sample size is small and there are some missing number in the dataset.

## The Impact of Education Index on HDI

H0 (Null Hypothesis): there are no correlations between HDI and education index.

H1 (Alternative Hypothesis): there has correlations between HDI and education index.

Based on the regression output, we can reject the H0 under more than 95% confidence interval.

HDI=0.9911*education index+0.1069

# Concluding Remarks

There is no doubt that the HDI has positive correlations with log GNI per capita, life expectancy and education index. While the educational level can impact HDI more among these independant variables. And if education index increase by one unit, the HDI can grow up by 0.9911 unit.

As I mentioned at the beginning, the development focus has shifted from economic development to human develoment since 1990s. Human development is a vital development topic for both developed countries and developing countries. More importantly, educational development can promote well-rounded human development directly and effectively. In conclusion, education is the most powerful weapon which can be used to change the world.