Determination of annual nominal wages : the case of China 1985-2012

(1)

Bachelor Thesis

Determination of annual nominal wages:

The case of China 1985-2012

Debbie Keijser

Student number: 10525378

Supervisor: Stephanie Chan

Abstract

China has experienced an increase in international trade during the last decades. The low wages that were mostly responsible for China’s comparative advantage, are however increasing. In this paper the possible factors that caused this increase are analysed and regressed against the dependent variable, Chinese nominal wages. The investigated time-period starts in 1985 and ends in 2012. The variables that are investigated are: productivity, education, cost of living, exports, unemployment and gross domestic product. After correcting for non-stationarity, the results of an OLS regression show that ‘export growth’ and ‘unemployment growth’ have a significant effect on the growth rate of Chinese nominal wages.

(2)

Statement of Originality

This document is written by Student Debbie Keijser who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1 Introduction

In 1978, just after the ‘Mao Era’, China started an economic reform to stimulate the Chinese economy. Since then international trade has been the main driver of

economic growth in China. The expansion in international trade was made possible by the cheap labour force that China had available (Li et al, 2012, p.57). At the beginning of the economic reforms in China the annual wage of a Chinese urban worker was 615 renminbi yuan. In this period the Chinese wages were significantly lower than in neighbouring countries. The differences between Chinese wages and the wages in the U.S. were even more significant. In 1978 the Chinese annual wage was equal to 3% of the average wage in the U.S. (Li et al, 2012).

However, Chinese wages are now showing an increasing trend. China has an annual growth rate of real wages of 12% per annum, which means that the real average wages in China have more than tripled over the decade from 2000 to 2010. Now questions arise about the possible end of ‘cheap labour’ in China, which gave China its comparative advantage in the first place (International Labour Organization, 2013, p. 20). That is why the research question of this thesis is: which factors

influence the Chinese nominal wages?

Answers to this question are relevant to several parties, for example to the Chinese government because they can adjust or implement policies to influence the wages. It can also be relevant for companies that import products from China, because the increasing wages will be reflected in the prices of the products. Firms will have to make trade-offs and look at alternative suppliers in countries with lower wages like India or Indonesia (Li et al, 2012, p. 58).

This paper seeks to fill the gap between the economic theory and the separate research papers trying to explain the rise of Chinese wages. The key purposes of this paper are to give a complete and general understanding of the trend the Chinese wages are showing and a detailed analysis of the factors that might influence the wages.

To do this, the thesis is divided in six parts. In the next part previous literature concerning wage determinants will be analysed. In part 3 the variables, which may have an effect on Chinese wages, are discussed. In part 4 the methodology will be described and also the results that have been found are shown. In part 5 the

(5)

2 Literature review

2.1 Economic theory

In this paper economic theory will be used to analyse the research question: which factors influence Chinese nominal wages?

As mentioned in the introduction China is involved in international trade, therefor it is useful to look at the theories concerning wage determination when international trade is allowed for. The simplest model is the ‘Ricardian model’. It explains

comparative advantages of countries looking only at labour. A comparative advantage means that a country can produce a product relatively more efficient. The theory states that the product that a country has a comparative advantage in, will be exported to the country with the disadvantage in the production of that product. When the trade patterns are established, the wages are determined by the price of the product that the employee produces, divided by the amount of hours needed to produce this product (Krugman, Obstfeld & Melitz, 2012, p. 25-29). If the productivity of employees remains the same but demand for Chinese exports increases, what we have seen the past decades, prices will increase. The increase in price will be directly compensated to the workers. Which means that real wages will have to increase.

The ‘Heckscher-Ohlin theory’ is another early attempt to explain wage changes. The theory states that expansion of trade will benefit the production factor that is abundant in that country. This is the case because a country will produce the goods, which use the production factors that are abundant in that country (Krugman, Obstfeld & Melitz, 2012). In China labour is abundant, which suggests they would export goods that are labour-intensive. This is also what we have seen in practice over the past decades. Regarding the wages of these employees, the theory states that their real wages would increase due to increasing international trade (Mitchener & Yan, 2014, p. 131).

Not only international trade theories, but also the theories developed in macro- and microeconomics are used to determine wages. We begin with the theories that explain wages by macroeconomic variables, this refers to variables that hold for all the workers in a sector or country.

Macroeconomics state the importance of bargaining power when wages are set (Blanchard & Johnson, 2013). This can be individual or collective bargaining power. A worker increases its bargaining power if it has a higher education or higher

(6)

productivity, but also other factors can play a role. Collective bargaining is done by labour unions. We will come back to this in part 3, first the general equation for wage determination will be discussed.

In the book of Blanchard and Johnson (2013) the following equation is stated to explain the determination of nominal wages:

𝑊 = 𝑃𝑒

𝐹(𝑢, 𝑧)   (1)

W stands for the nominal wage. 𝑃,is the expected price level. Unemployment rate is stated as ‘𝑢’. Finally, ‘𝑧’ stand for all other variables that may affect wages.

Expected price level is included because workers care about their real wages, which is equal to nominal wages divided by the price level. Real wage indicates the purchasing power of workers. If the price level increases, nominal wages will have to increase in the same portion to keep the purchasing power of workers stable. So a positive relation between the price level and nominal wages is hypothesized (Blanchard & Johnson, 2013, p. 120). In contrast the relation between unemployment and wages is expected to be negative. This is due to a decrease in collective bargaining power if unemployment increases. Unemployment indicates a bigger supply than demand for labour and workers are forced to accept a lower wage (Blanchard & Johnson, 2013, p. 121). The parameter ‘𝑧’ only includes factors that positively influence nominal wages. This can include minimum wages, education and more.

To link unemployment, inflation and wages together the Philips curve will be discussed next. The Philips curve is used to calculate the inflation rate of a country. To do this the following variables are used: expected inflation, cyclical

unemployment and supply shocks. Cyclical unemployment stands for the deviation of the unemployment from the natural rate ‘𝑢-’. The complete Philips curve is the following (Mankiw, 2013, p. 404):

𝜋 = 𝜋_,− 𝛽 𝑢 − 𝑢_- + 𝑣 π = inflation

πe = expected inflation

ß = parameter that measures response of cyclical unemployment to inflation u = unemployment rate

un = natural unemployment rate v = supply shock

(7)

Actual inflation is positively affected by the expected inflation rate and by supply shocks. However, cyclical unemployment negatively influences the inflation rate. Equation (2) shows that other things equal, higher unemployment will lead to lower inflation. Linking this back to nominal wages, higher unemployment causes wages to decrease (Blanchard & Johnson, 2013). Because higher unemployment decreases wages and inflation, inflation and wages must be positively correlated. An increase in inflation will decrease unemployment and increase nominal wages.

Another theory developed in macroeconomics is the ‘Marginal Rate of

Productivity theory’ (MRP theory). The theory states that wages need to reflect the marginal rate of productivity of a worker. Marginal productivity is determined in terms of the supply of works units per worker. The most profitable MRP will be obtained were the marginal costs of a unit supplied equals its marginal revenue product (Mazumdar, 1959). Nowadays it is known that marginal rate of productivity and wages are related but wages are not solely determined by the MRP. This is why new theories were developed to try to explain the determination of the wages more accurate.

This is for example done by Becker. He is the first economists to look at personal characteristics, microeconomic variables, instead of overall factors. Nobel Laureate Gary Becker developed the ‘Human Capital theory’. This theory explains the differences in wages by the (dis)advantages of workers. An important component is the cost of learning a job. The founder claims that, ceteris paribus, incomes vary according to the amount of personal human capital (Mincer, 1958). Education is the main element of human capital and this is further investigated by Mincer in his paper: ‘Schooling, Experience and Earnings’. In this paper he used data to relate income distribution in America to the amount of education and training among employees. The ‘Mincer earnings equation’ was named after him. In this equation he tries to explain the unknown variable, earnings, by looking at personal characteristics of a specific worker. These characteristics can include sex, race, schooling, experience and more. The ‘Mincer equation’ is used to estimate individual earnings and you need personal information to derive this (Mincer, 1974).

A number of studies have been based on this equation to analyse wages. An example is the study conducted by Gaag and Vijverberg on date from Côte d’Ivoire. Their analysis focuses on two variables: education and experience. They conclude

(8)

that both factors have positive and significant effect on wages. Different types of education levels are also regressed on wages to analyse their effect. An outstanding result was that they found a much higher return on higher education than is usually found in countries in that region. Higher education is defined as all the education after elementary school, this includes high school and university education. An important detail is that they only looked at wage earners and that 41% of the Ivorian wage sector is government employment.

_{Different from neighbouring countries, the Ivorian}

government pays all its employees a reward for completing an education which could have influenced the results (Gaag & Vijverberg, 1989).

More recently, the Mincer equation has been used to examine wage

determination and distribution in urban China and Vietnam (Cai & Liu, 2014). Cai and Liu used panel data of Chinese and Vietnamese households to do their analysis. They investigated the differences in wages looking at the sex of the worker.

Afterwards, they did a regression using the ‘Mincer equation’ including around 20 variables. They found out that in general the returns to industry are higher for men than for women. However, education and experience are more important contributors for female wages. The main factor that differentiate the Vietnamese and Chinese wages is education. Despite the lower wages the Vietnamese offer, they are less attractive than Chinese workers because Chinese workers are better skilled, due to better and higher education levels, and are thus more efficient (Cai & Lui, 2014).

2.2 Previous studies

In the rest of this section a couple of previous analyses on national wages will be discussed. First an example will be given from how wages have been analysed in India. After that the previous findings on Chinese wage determination will be discussed. All the papers are based on macroeconomic variables influencing wages, because this is the approach in this thesis.

Gulati, Jain and Satija investigated the rising farm wages in India (2014). They used panel data of 16 major states for the time period 1990-1991 to 2011-2012. They first analysed the rise in nominal and real wages. They concluded that both have increased but real wages were expected to be higher. Next they discussed possible factors affecting the real wages of farmers in India. They looked at the growth in GDP

(9)

(overall, in the construction sector and Agri-GDP) and they concluded that all GDP factors have a positive effect on wages. However, the positive influence of growth in the construction sector GDP is stronger than growth in overall GDP or agri-GDP. This is due to the fact that construction is a main driver of the overall growth of the economy. The growth in the construction sector leads to more demand for employees and the trend shows that workers have moved from the agricultural sector to the construction sector. The fraction of farmers has decreased and to be able to maintain the farmers at their current position, higher wages will be paid. Also the effect of the implementation of MGNREGA has been analysed to see its effect on wages. The MGNREGA is a political policy that ensures more financial stability for unskilled workers. They conclude that the positive effect is significant but a lot less effective than the growth variables (Gulati, Jain, &, Satija, 2014).

Not only in India but also in China there have been a lot of studies analysing wages. In the paper ‘The End of Cheap Chinese labor’ Li et al (2012) analyse the rise in Chinese real wages. They look at nominal wages deflated by China’s producer price index. After that they compare the productivity changes and the the real wages changes to see if labour has become relatively cheaper or more expensive in the period from 1978 to 2010. They calculated the productivity by deducting the labour force growth rate from the real GDP growth rate. Until the 1990’s productivity of workers increased more than their wages. After the 1990’s this changed and wages grew faster than productivity. This means that labour has become relatively more expensive. They conclude that the real wages remained stable during the 1980’s and early 1990’s but grew substantially since then. After this conclusion, potential reasons for rising wages are given. They make three subsections: institutional reforms, the disappearing ‘demographic dividend’, and the slowing of rural-urban migration. They conclude that the main factors that contribute to the increasing wages are: higher returns on education, lower fertility rates and less migration of cheap labour supply (Li, Li, Wu, & Xiong, 2012). This conclusion is in line with the factors that have a positive effect on wages shown in the Mincer equation (Mincer, 1974).

Mitchener and Yan analyse the relative differences between the wages of skilled and unskilled workers in China (2014). They are able to distinguish these two

categories by a classification system which measures the skill intensity of jobs by educational attainment using data from the 1940 U.S. Census. They state that at the

(10)

beginning of the 20th century China opened up to international trade. China started to import skill-intensive goods and export unskilled-intensive goods. An important point in time in this paper is the World War I, they conclude that this war increased Chinese exports drastically. Increased demand of Chinese export also increased demand for unskilled workers. After the war, costs of trade decreased which stimulated the exports more. They conclude that the war had a positive effect on China’s terms of trade, which decreased the skill premium of Chinese skilled workers (Mitchener & Yan, 2014). They mostly analysed the difference in wages between skilled and un-skilled workers, and not the overall trend in average Chinese wages, which will be the main variable in this thesis.

In the next part the possible factors that influence wages will be analysed for the case of China.

3 Variables

In this part the trend of Chinese nominal wages will be discussed. Afterwards possible factors that influence the Chinese wages will be explained and analysed.

3.1 Chinese wages

China’s Statistical Yearbooks publishes data about for example Chinese wages. They give an extensive range of data. Concerning wages, they provide the average wage of an employee in urban units. Urban units are work facilities in the city. The units can be state-owned or private-owned. Overall employment can be divided between employment in urban units and rural units. Rural units refer to mostly work on farms. The wages of rural employees are difficult to calculate because of the size of the country, which makes the collection of information more difficult. But not only the collection of this information is difficult, also the analysis would be hard because of the flexible wages farmers deal with. Farmers have to deal with variable prices they get for their product which makes the estimation of average wages difficult. Having determined that only the average wages of urban employees will be used we can start with the analysis of their nominal wages. The data about nominal wages provided by China’s Statistical Yearbooks, begin in 1985 and ends in 2013.

(11)

0 10000 20000 30000 40000 50000 60000

Nominal wages 1985-2013

Figure 1: Nominal wages 1985-2013

In Figure 1 the nominal average wages of urban workers are displayed. The wages are in Chinese renminbi (yuan), the Chinese official currency, per year. The first thing that this Figure makes clear is the continuous growth of the nominal wages since 1985. There is however a difference in the growth rate during the past decades. We see a slow growth rate until 1997-1998. From 1998 until now we see an increasing growth rate of nominal wages. To illustrate the large increase in nominal wages during the last decades, consider that the nominal wage of an urban worker was 1.148 renminbi yuan per year in 1985. The nominal average wage in 2013 was 51.483 renminbi yuan. This is a significant increase of income. Nominal annual wage will be the dependent variable in this paper.

3.2 Productivity

The ‘Marginal Rate of Productivity theory’ states that the wages need to reflect productivity of workers (Mazumdar, 1959). We have seen that the nominal Chinese wages have increased drastically since the 1990’s, but to investigate if Chinese labour has become relatively more expensive we need to look at the productivity of workers (Li et al, 2012, 63). Li et al calculated the growth of gross labour productivity by deducting the growth rate of the Chinese labour force from the growth of real GDP. These same will be done in this thesis but then with gross GDP because we look at nominal wages instead of real wages.

(12)

-20 -15 -10 -5 0 5 10 15

Productivity growth

Figure 2: Productivity growth in percentages

Figure 2 shows an average productivity growth of around 5% per year. However, the variance is large. We see a sharp decline in 1990, which is due to an increase of 20% in the labour force. GDP growth in this year was 3,9% which is normal for that period (China’s Statistical Yearbooks). There is however no clear explanation for the drastic increase of labour supply.

Looking at the possible relationship between nominal wages and productivity levels in China, the MRP theory states that there should be a positive relation. Higher productivity would lead to a higher wage if the factors are perfectly competitive, because the higher revenue they produce would be directly compensated. Even if the labour market is not perfectly competitive, a higher wage is expected to be paid if productivity increases. This is due to the increase of bargaining power the workers get because of the increase in productivity. However, the bargaining power only becomes credible if trade unions negotiate for the workers about wages. Trade unions come up for the rights of the workers by means of collective bargaining with employees. However, China only has one trade union. This union is called All China Federation of Trade (ACFTU), but Chan concluded that their influence is really small. He states that any concept of collective bargaining as protectors of workers rights barely exists in China (Chan, 2010). If the compensation for higher productivity is only arraigned by trade unions, then a negative relation between Chinese nominal wages and

productivity is expect, due to the absence of strong trade union. The MRP theory and the bargaining power view on this topic contradict each other. The MRP theory

(13)

predicts a positive effect on nominal wages and the bargaining power view predicts the opposite. In part 4 the regression will be analysed and that will clear which force has the most impact on the relation between Chinese nominal wages and productivity.

3.3 Education level

The Chinese education system is run by the Ministry of Education. They provide a nine-year compulsory education program that is completely funded by the

government. This program includes three years of junior secondary education. This indicates that all workers should have a minimal amount of education.

The ‘Mincer equation’ and the ‘Human Capital theory’ state the importance of education when looking at the determinants of wages (Mincer, 1974). This is also confirmed in macroeconomic theory. It can be included in Equation (1) under the parameter ‘𝑧’. In this paper the education level is measured by the number people that enrol in regular secondary school per year.

Figure 3: Enrolment of regular secondary school x10000 students

We see a high value at the beginning of the time period. Around 6,5 million students enrolled in 1978. A decline in enrolment occurred afterwards. However, the number on enrolled students remained almost stable between 1982 and 1994. After that we see again an increase in enrolments. The last decade shows a small drawback of enrolments.

The expected relationship between education and nominal wages is the

following: higher education is usually compensated by a higher wage (Mincer, 1974). This is because education contributes to human capital. More time and money has been put into the development of a person and this development enhances skills,

0 2000 4000 6000 8000 10000

Enrolment of regular secondary school

(14)

-5 0 5 10 15 20 25 30

Inflation

which increases productivity. That is why they expect higher wages in the future. That is why a positive relationship between nominal wages and education is expected to hold in this analysis. The expected positive relation is also confirmed by Equation (1). Per definition, all the variables under variable ‘𝑧’ must have a positive relation with nominal wages.

3.4 Cost of living

The cost of living is measured by the consumer price index (CPI). CPI shows the trend and degree of changes in prices of services and goods purchased by rural and urban households. This measure indicates the impact of changes in prices of goods and services on the actual living expenses (China statistical yearbook, a). The data provided by China’s Statistical Yearbooks that will be used begins in 1978. These are annual CPI values, but stated in the matching percentages. The percentages are the inflation rates.

Figure 4: Inflation rates

Figure 4 shows a variable inflation. However, we see a more constant growth since 2002. In 1998, 1999, 2002 en 2009 we see a negative inflation. This means that products in this year have become cheaper than in the previous year. This is called deflation and increases the purchasing power.

The Philips curve links inflation to wages through unemployment. As stated in part 2, a positive relation is expected between wages and inflation. The intuition

(15)

stable but inflation increases, you are able to buy less products. Products become more expensive but your income remains the same which means your purchasing power decreases. Employees demand a higher nominal wage if the inflation increase to compensate for the purchasing power loss. Thus also a positive relation between nominal wages and inflation is expected.

3.5 Exports

In 2012 the Chinese exports accounted for 24% of the GDP (China’ Statistical Yearbooks, b). Exports have been a main driver of the Chinese economy since the opening up to international trade around 1978. International trade theories show that China will export products that are labour-intensive because of the labour abundance. An increase in international trade would lead to an increase in labour demand which would then drive up the wages of Chinese workers (Mitchener & Yan, 2014). Figure 5: Exports in million yuan

Figure 5 shows an increase in exports since 1985. A drastic increase in growth of exports starts in 1992. This growth increase from then on, only to have a decrease in 2009. This may be due to the financial crisis (Li, Willett & Zhang, 2012). However, we see an immediate increase in exports in 2010. This increase in exports since 1985 can be due to the low prices Chinese exports ask for their products but also because of the exchange rate policy the Chinese government is conducting. The central bank of China, People’s Bank of China, has been depressing the price of the renminbi since around 2001 (Federal Reserve Bank of Cleveland, 2015). High demand for renminbi arising from China’s exports and inward investment, exceeds the supply of the currency resulting from China’s imports and foreign investments. To maintain a

0 20000 40000 60000 80000 100000 120000 140000 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Chinese exports

(16)

0 100 200 300 400 500 600 700 800 900 1000 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Unemployment

stable exchange rate under these circumstances, the People’s Bank of China buys dollars and other currencies in the foreign-exchange market. Because of this policy the exchange rate did not decrease as much as would have been expected due to the increase in demand for this currency which kept Chinese exports interesting for foreign importers.

The hypothesis is that the relationship between nominal wages and export is positive. This is expected because most of the Chinese export is labour-intensive (Mitchener & Yan, 2014). The production factor that is abundant, in this case labour, will benefit from more trade (Krugman, Obstfeld & Melitz, 2012). This indicates that Chinese wages will increase due to an increase in exports.

3.6 Unemployment

Figure 6 shows the registered urban unemployment since 1985. This data has been collected from the Chinese Statistical Yearbooks but the actual unemployment level will be higher this data shows. This is due to the fact that the government can not capture all the unemployment due to unavailability of information. For example, seasonal unemployment and unemployment in rural units are hard to measure. Figure 6: Unemployment in x10000 persons

Figure 6 shows almost constant increase in unemployment level since 1985.

Blanchard and Johnson included unemployment in their equation. The theory states that higher unemployment levels make it harder for employees to bargain about their wages. This is due to the fact that unemployment is caused by a gap between supply and demand of labour. Unemployment states a higher supply than demand. The

(17)

increase in unemployment would indicate a lower bargaining power of the employees thus lower wages. In this time period a negative effect of unemployment is expected. This bargaining power could increase due to the power of trade unions. But as stated before, trade unions barely exist in China and have almost no power (Chan, 2010). So the negative effect of labour abundance is not compensated by the power trade unions have on Chinese wages.

There is another factor that could increase the bargaining power of workers in times of unemployment, the availability of unemployment benefits. Unemployment benefits are social welfare payments that the government can pay unemployed people to provide them with income. If these benefits are high, workers will have a higher bargaining power due to the low difference between a normal annual wage and the benefits they would receive if they are unemployed.

In China there is an unemployment insurance system since 1999 (China Labour Bulletin). The system works as follows: both workers and employers pay respectively around one and two percent of the nominal wage into the unemployment insurance system. When a worker becomes unemployed, he/she receives an

unemployment benefit based on local minimum wages. Because minimum wages are already set low in China, the unemployment benefit cannot be considered a living wage (China Labour Bulletin). Because of the low unemployment benefits, the decrease in bargaining power due to unemployment cannot be compensated by the power of the unemployment insurance system.

Taking all factors into account, a negative effect of unemployment on Chinese nominal wages is expected.

3.7 Gross Domestic Product

The Gross Domestic Product states the production of a country. In Figure 7 shows the GDP development of China since 1985. We see a continuous growth in GDP. The growth rate of GDP also increases over time. The tipping point seems to be 1993, from this year on the growth sped up.

(18)

Figure 7: Gross Domestic Product in 100 million yuan

An increase in GDP is indicative of an increase in overall production of the country. An increase in production can be accomplished if the productivity of the current workers increases or the number of workers increases. Either way this will lead to an increase in nominal wages. As stated under the variable ‘productivity’, an increase in productivity will have to be compensated by an increase in wage. If productivity remains constant the increase in GDP will increase the demand for Chinese workers. Keeping the supply of workers constant, wages will increase. A positive relation between GDP and nominal wages is expected.

In the next part the empirical analysis will be conducted to see if the above stated implications really hold.

4 Methodology

After looking at the variables that could have an impact on Chinese nominal wages we are going to calculate if there is an actual effect. To do this, the variables mentioned in the previous part are going to be regressed on the dependent variable: average Chinese nominal wages. STATA is used to estimate the regression

parameters.

4.1 Model

The following regression has been used to calculate the effects of the variables on nominal wages. The variables that are not calculated in percentages are converted into their natural logarithms. This is done to make the interpretation easier.

0 100000 200000 300000 400000 500000 600000 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Gross Domestic Product

(19)

ln 𝑛𝑜𝑚𝑖𝑛𝑎𝑙 𝑤𝑎𝑔𝑒 =

𝛽_?+ 𝛽_@ ∙ 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑣𝑖𝑡𝑦 + 𝛽_H∙ ln 𝑒𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛 + 𝛽_I∙ 𝑖𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛 + 𝛽K∙ ln (𝐺𝐷𝑃) + 𝛽N∙ ln (𝑢𝑛𝑒𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡) + 𝛽O∙ ln (𝑒𝑥𝑝𝑜𝑟𝑡𝑠) + 𝜀

The following table summarizes the forms of the variables chosen: Table 1: variables without corrections

Nominal wages Absolute annual nominal wages in yuan

Productivity Growth rate of GDP minus growth rate of

Chinese labour force

Education Absolute level of student enrolment in

secondary education for a given year

Inflation Percentage change of CPI per year

GDP Absolute Gross Domestic Product in 100

million yuan

Unemployment Absolute number of registered

unemployment in 10000 persons

Exports Absolute level exports in million yuan

The above regression is estimated using an Ordinary Least Squares (OLS) regression. This means that the estimator of the regression intercept and slope minimizes the sum of squared residuals (Stock & Watson, 2015, p. 822) The sample period starts in 1985 and ends in 2012. Annual data is used for this regression. This time period has been selected because 1985 is the first year that provides data after the ‘Mao era’ and 2012 is the most recent year that all the variables were available.

4.1.1 Results

Four different regression have been conducted. The first one includes all the variables while the second excludes the insignificant variables. Regression 3 excludes ‘ln (GDP)’ and regression 4 excludes the variable ‘productivity’. Regression 3 and 4 are conducted because GDP and productivity actually measure the same thing, production.

The Breusch-Pagan / Cook-Weisberg test for heteroskedasticity, has been conducted to analyse if correction for heteroskedasticity is necessary. The error term

(20)

is heteroskedastic if the variance of the conditional distribution is not constant and depends on other variables (Stock & Watson, 2015, p. 204). The test showed that the error terms are homoskedastic, the results are stated in the Appendix. However, to correct for possible outliers or influential observations the ‘robust’ option has been applied in the regression.

Table 2

Dependent variable: Ln (nominal wages)

Productivity -0,0054*** -0,0025* -0,0070 (0,0018) (0,0014) (0,0074) Ln(education) 0,5568*** 0,3942*** 0,4289 0,4990*** (0,1390) (0,1095) (0,3118) (0,1464) Inflation 0,0011** 0,0009** 0,0027*** 0,0009** (0,0005) (0,0004) (0,0007) (0,0005) Ln (GDP) 0,9120*** 0,8560*** 0,9202*** (0,0806) (0,0275) (0,0935) Ln (unemployment) -0,4441 -0,3497 -0,2647 (0,2716) (0,4628) (0,2241) Ln (exports) 0,0546 0,7586*** 0,0062 (0,0867) (0,1303) (0,0748) Constant -3,8980*** -4,1431*** 0,0994 -4,1828*** (0,6101) (0,6813) (2,3319) (0,6779) R2 _0,9974 _0,9968 _0,9860 _0,9970 Prob > F 0,0000 0,0000 0,0000 0,0000 * Significant at 10% level ** Significant at 5%level *** Significant at 1% level

The Prob>F value measures the probability of obtaining an F Ratio as large as what is observed, given that all parameters except the intercept are zero. Small values indicate that the observed F Ratio is unlikely. Small Prob>F values are considered evidence that there is at least one significant effect in the model. The results show that the

(1) (2) (3) (4)

(21)

Prob>F values of all four regressions are 0,0000. This means that in each regression contains at least one significant effect.

The first regression includes all variables. The results show that ‘productivity’, ‘education’, ‘inflation’ and ‘GDP’ are significant. The second regression excludes the variables that are not significant at any level. The signs of the coefficients remain the same as in the first regression but are all smaller. The significance of the variables remains unchanged except for productivity which becomes less significant.

Regression 3 and 4 are included to see if either ‘productivity’ or ‘GDP’ should be excluded from the first regression. This is expected because both variables come down to the production of the Chinese workers. Looking at the Table 1, ‘productivity’ is calculated by growth rate of GDP minus growth rate of Chinese labour force. GDP is the absolute Gross Domestic Product of China. So, per definition these two

variables are linked. If both variables measure the same thing, production, the same results are expected in regression 3 and 4. However, the results show different outcomes if one of the variables are excluded. In regression 3 - where ‘GDP’ is excluded - the signs of the coefficients are equal to the ones of the first regression. Inflation becomes more significant and exports becomes significant, which it was not in the first regression. Productivity and education are no longer significant variables in this regression. In the fourth regression ‘productivity’ is excluded. The variables that had a significant coefficient in the first regression, also are significant in the fourth regression. Also the signs of the coefficients are identical. Because the results of regression 3 and 4 are different both variables need to be included into the

regression.

4.2 Stationarity

In the previous regressions the ‘robust’ option was included to correct for possible outliers or influential observations. However, there are other complications that may arise. In this paragraph stationarity will be discussed.

The problem with non-stationarity is that if you keep these variables in your regression the outcome can be determined by the fact that they both grow constantly and not because of actual correlation (Stock & Watson, 2015, p.587).

(22)

1. Look at the difference between the value of 𝑡? and 𝑡S@ and use this as a new variable instead. The downside of this is that one data point is lost. 2. Take the natural logarithm of the variable. However, this is not possible if

there are negative data points.

The Dickey-Fuller test has been used to test if the variables are non-stationary. The results are stated in the Appendix. All the variables, expect for education, are converted into stationary variables. Changing education into its growth rate, natural logarithm or into a percentage of total population did not help against the non-stationarity problem. That is why education has been included in the form with the lowest p-value.

The following table summarizes the forms of the variables chosen: Table 3: variables corrected for non-stationarity

Nominal wage growth Nominal wage in year 𝑡 minus nominal wage in year 𝑡S@, divided by nominal wages in 𝑡S@

Productivity Growth rate of GDP minus growth rate of Chinese labour force

Education growth Enrolment in year 𝑡 minus enrolment in year 𝑡S@, divided by enrolment in 𝑡_S@

Inflation Percentage change of CPI per year

GDP growth GDP in year 𝑡 minus GDP in year 𝑡S@, divided by GDP in 𝑡S@

Unemployment growth Unemployment in year 𝑡 minus unemployment in year 𝑡_S@, divided by unemployment in 𝑡S@

Export growth Exports in year 𝑡 minus exports in year 𝑡S@, divided by exports in 𝑡S@

4.2.1 Results

With the variables stated in Table 3, four regressions are conducted. The first one includes all variables and the second regression excludes the insignificant variables from regression 1. Regression 3 excludes ‘GDP growth’ and regression 4 excludes the ‘productivity’. Again the Breusch-Pagan / Cook-Weisberg test for heteroskedasticity, has been conducted to analyse if correction for heteroskedasticity is necessary. The

(23)

results in the Appendix show that only the error term of the second regression is heteroskedastic. However, the ‘robust’ option has been used to compute all regressions.

Table 4

Dependent variable: Nominal wage growth

Productivity 0,0014 0,0050*** (0,0023) (0,0015) Education growth 0,0003 -0,0001 0,0004 (0,0019) (0,0020) (0,0019) Inflation 0,0005 0,0005 0,0005 (0,0003) (0,0003) (0,0003) GDP growth 0,0079 0,0102*** (0,0057) (0,0026) Unemploym. growth 0,0026** 0,0024 0,0019 0,0028** (0,0012) (0,0018) (0,0015) (0,0011) Export growth 0,1261* 0,1239* 0,1457*** 0,1199** (0,0614) (0,0677) (0,0507) (0,0577) Constant 0,0157 0,1082*** 0,0724*** 0,0016 (0,0482) (0,0160) (0,0193) (0,0355) R2 _0,5695 _0,3251 _0,5363 _0,5657 Prob > F 0,0066 0,1019 0,0156 0,0106 * Significant at 10% level ** Significant at 5%level *** Significant at 1% level

In the first regression ‘unemployment growth’ and ‘export growth’ are significant. The Prob>F value is the lowest of all four regressions, which indicates that the

possibility that a regression contains at least one significant effect is the highest in the first regression.

(1) (2) (3) (4)

(24)

In the second regression the export growth stays significant at the same level as in regression 1. However, unemployment growth becomes insignificant at any level. The Prob>F value increases in comparison to the value of the first regression.

Equal the reasoning in part 4.1, regressions 3 and 4 are included to see if either ‘productivity’ or ‘GDP growth’ should be excluded from the first regression. Again we see a diversity in outcomes which indicates that both variables should be included.

4.2.2 Interpretation of results

In the previous part we stated that the final conclusions will be based on the first regression of Table 4. In this part the interpretations of the coefficients will be discussed and this will be compared with the expectations formed part 3.

The coefficients of the first regression can be interpreted as follows:

If productivity increases by 1% point the growth of nominal wages will increase with 0,0014% point. The ‘Marginal Rate of Productivity theory’ states that an increase in productivity should positively influence wages, but because of low collective

bargaining power it is unsure if this will be the case in China. Regression 1 shows an insignificant positive coefficient. This means that Chinese workers get compensated if productivity increases in terms of wage, but this effect is not significant. This result shows that the absence of trade unions in China does not interfere with the positive effect of productivity on wages expected by the MRP theory.

An increase of 1% point in education growth will increase nominal wages growth with 0,003% point. A positive relationship between education level and nominal wages is expected. Education is included in the ‘𝑧’ of equation (1) and indicates that an increase in education increases the personal bargaining power and will have a positive effect on wages. The increase in bargaining power is due to the fact that higher education makes workers more productive or/and capable of doing more difficult tasks. The results show the hypostasized outcome but the outcome is not significant.

If inflation increases by 1% point the growth of nominal wages will increase by 0,0005% point. The relation between inflation and wages is stated in the Philips curve and is expected to be positive. This is due to the fact that workers want to keep a stable or improved purchasing power. The results in Table 4 show the expected outcome but again this result is not significant.

(25)

An increase in GDP growth of 1% point will result in an increase of 0,0079% point of nominal wages growth. The relation between GDP and nominal wages is expected to be positive and the same is suggested by the results. An increase in production increases the demand for workers and without an increase in supply wages will increase.

An increase of 1% point in unemployment growth will lead to an increase of 0,0026% point in nominal wages growth. Unemployment has a negative hypothesized effect on nominal wages. The equation developed by Blanchard and Johnson states that if unemployment increases the bargaining power of workers decreases if this not compensated by actions of trade unions. In part 3 we stated that the power of trade unions is relatively small in China. However, the results show a significant positive coefficient, which contracts the hypothesized outcome.

Finally, an increase of 1% point in exports growth will result in an increase of 0,1261% point in nominal wages growth. The relation between exports and nominal wages is expected to be positive. Higher demand for labour-intensive goods will increase the demand for workers and drive up the wages. This reasoning is very similar to the one of GDP. However, the positive coefficient of exports growth is significant and unlike the coefficient for GDP growth. Because of the importance of export for the Chinese economy, the higher and significant outcome can be explained (Li et al, 2012).

Comparing the results from before and after the correction for non-stationarity indicates the following. In Table 2 the variables ‘productivity’, ‘education’, ‘inflation’ and ‘GDP’ are significant. However, after correcting for non-stationarity these

variables become insignificant. Table 4 shows that unemployment and export growth become significant. This indicates that the relation between Chinese nominal wages and ‘productivity’, ‘education’, ‘inflation’ and ‘GDP’ is mostly driven by the fact that they all grow over time and probably not by actual correlation.

(26)

5 Conclusion

5.1 Summary

In this paper the potential factors that influence Chinese nominal wages are analysed. First, the economic theory behind wage determination has been discussed.

International trade theories hypothesize an increase in wages due to an increase in international trade. Macroeconomics state that an increase in expected price level will increase nominal wages and the same holds for productivity. Unemployment has a negative effect on nominal wages due to a decrease in collective bargaining power. The Philips curve hypothesizes a positive relation between inflation and wages. The microeconomic explanation of determination of nominal wages is dominated by the theories developed by Becker and Mincer. The ‘human capital theory’ and the ‘Mincer equation’ state the positive effect of education and experience on nominal wages.

Secondly, previews studies have been analysed. Case studies of India and China are analysed. The investigation of the wages of farmers in India indicated that

MGNREGA, GDP, construction-GDP and Agri-GDP all have a positive effect on wages. In the paper of Li et al (2012) the rise in Chinese wages are analysed. They concluded that the main factors that contribute to the increasing wages are: higher returns on education, lower fertility rates and less migration of cheap labour supply.

Because of the diversity of factors that can influence wages, in the third part possible determinants of wages are analysed. Productivity, education level, inflation, exports, unemployment and GDP are investigated. A positive effect on nominal wages of education, inflation, exports and GDP is expected. The effect of productivity is ambiguous because of the absence of trade unions and the effect of unemployment is expected to be negative due to the increase in the gap between supply and demand of labour.

In the forth part regressions were conducted to investigate which factors have a significant effect on nominal Chinese wages. The results before the correction for non-stationarity show that productivity, education, inflation and GDP have a

significant effect on Chinese nominal wages. Because the presence of non-stationary variables can influence the validity of the results, all variables are tested and if needed corrected into a stationary variable. The results show that ‘export growth’ and

(27)

wages. Most results were in line with the expectations stated in the economic theory. However, the coefficient for unemployment growth came out positive, which was expected to be negative.

In the next section the limitations and improvements of this paper will be discussed.

5.2 Limitations and improvements

There are areas of limitations and improvements regarding this paper that will be discussed in this part.

The main limitation of this study is the sample size. With only 28 data points it is difficult to exclude coincidence. This is due to the fact that China’s Statistical Yearbooks do not provide monthly data on all the variables that have been analysed. The lack of information can be eliminated by doing own field research. The

downside is the extreme large surface of the country, which will drive up the costs and time it takes to conduct such an investigation. Another option is to widen the sample size by adding data from years before 1985 and even before or during the ‘Mao Era’. This data is not public but this information might be available for Chinese researchers.

Another limitation of this study is the absence of control databases. All the data gathered for the analysis are collected from China’s Statistical Yearbooks. The question arises about the trustworthiness of this information. This is because China is not known as a transparent country when it comes to distributing information, because of all the censure they impose (Amnesty). The transparency problem is not easy to eliminate, but also in this case doing own research might solve this problem.

This paper has however some interesting outcomes which can be further investigated. For example, the outcome of the unemployment growth coefficient, which was different from what theories suggests, can be analysed.

The research can also be extended with more variables as determinates of wages. Investigate the upcoming behavioural economics theories to explain wage changes, is an option. Look at factors like: the size and prestige of the firm, the personal contribution to the firm per worker, strength of the labour unions and the concern of firms to maintain their workers.

(28)

Another interesting feature is to look at other factors that can determine

compensation for workers. These factors can include more free time or better working conditions for the workers. In 2013 the Chinese government came with the idea of allowing Chinese workers more free time (Tourism review, 2013). It might be interesting to investigate if this policy has a significant effect on the amount of free time the Chinese workers have.

At last, it can be relevant to compare the wages from the ‘Mao era’ and the wages after this period. Because of the lack of available data from this period, this was not a factor that could be included in this research. However, if you are able to get access to this information, a dummy variable for ‘Mao Era’ could be included in the regression to see the effect on nominal wages.

(29)

6 References

Amnesty International. Vrijheid van pers en internet. Accessed on 23 Januari 2016:

http://www.amnesty.nl/vrijheid-pers-en-internet

Blanchard, O., & Johnson, D.R., (2013). Macroeconomics, Sixth edition. Boston: Peason.

Cai, L., & Lui, A.Y.C., (2014). Wage determination and distribution in urban China and Vietnam: A comparative analysis. Journal of Comparative Economics, 43, 186-203.

Chan, A., (2010). Globalization, China's free (read bonded) labour market, and the Chinese trade unions. Asia Pacific Business Review, 6:3-4, 260-281.

China Labour Bulletin. China’s social security system. Accessed on 3 February 2016:

http://www.clb.org.hk/en/view-resource-centre-content/110107

China’s Statistical Yearbooks (Homepage).

http://data.stats.gov.cn/english/easyquery.htm?cn=C01

China’s Statistical Yearbooks (a). Accessed on 22 January 2016:

http://data.stats.gov.cn/english/easyquery.htm?cn=A01

China’s Statistical Yearbooks (b). Accessed on 22 January 2016:

http://data.stats.gov.cn/english/easyquery.htm?cn=C01

Du, Y. & Yang, C., (2014). DEMOGRAPHIC TRANSITION AND LABOUR MARKET CHANGES: IMPLICATIONS FOR ECONOMIC

DEVELOPMENT IN CHINA. Journal of Economic Surveys, 28 (4), 617–635 Federal reserve bank of Cleveland, 2015. The Chinese Renminbi and the Fundamental

Trilemma. Accessed on 5 January 2016:

https://www.clevelandfed.org/newsroom-and-events/publications/economic- trends/2015-economic-trends/et-20151001-the-chinese-renminbi-and-the-fundamental-trilemma.aspx

Gaag van der, J., & Vijverberg, W., (1989). Wage determinants in Côte d'Ivoire: Experience, Credentials, and Human Capital.Economic Development and

Cultural Changes, 37(2), 371-381.

Mankiw, N.G., (2013). Macroeconomics, Eighth edition. Basingstoke: Palgrave Macmillan.

Gulati, A., Jain, S., & Satija, N., (2014). Rising Farm wages in India – The ‘Pull’ and ‘Push’ factors. Journal of Land and Rural Studies, 2 (2), 261-286.

(30)

International Labour Organization, (2013). Global wage report 2012-2013: Wages

and equitable growth.

Krugman, P. R., Obstfeld, M., & Melitz, M. J., (2012). International Economics,

Theory & policy, Ninth edition. Boston: Pearson.

Li, H., Li, L., Wu, B., & Xiong, Y., (2012). The End of Cheap Chinese Labor.

Journal of Economic Perspectives, 26 (4), 57-74.

Li, L., Willett, T.D., & Zhang, N., (2012). The Effects of the Global Financial Crisis on China's Financial Market and Macroeconomy. Economics Research International, volume 2012.

Mazumdar, D., (1959). The Marginal productivity theory of wages and Disguised unemployment. The Review of Economic Studies, 26 (3), 190-197.

Meng, X., & Kidd, M. P., (1997). Labor Market Reform and the Changing Structure of Wage Determination in China’s State Sector during the 1980s. Journal of

comparative economics, 25, 403-421.

Mincer, J., (1974). Schooling, Experience, and Earnings. Human Behavior & Social Institutions No. 2. National Bureau of Economic Research, 0-152.

Mincer, J., (1958). Investment in Human Capital and personal income distribution.

Journal of political Economy, 66(4), 281-302.

Mitchener, K. J., & Yan, S., (2014). Globalization, trade and wages: what does history tell us about China? International economic review, 55 (1), 131-168. Morishima, M., (1973). A dual theory of value and growth. Cambridge: Cambridge

University Press.

Stirati, A. (1994). The theory of wages in classical economics. Vermont: Edward Elgar Publishing Company.

Stock, J. H., & Watson, M. W., (2015). Introduction to Econometrics, Updated third edition. Essex: Pearson Education limited.

Tourism review, 15 March 2013. THE CHINESE TO HAVE MORE FREE TIME TO

BOOST TOURISM. Accessed on 22 January 2016:

http://m.tourism-review.com/chinese-workers-gained-more-free-time-to-boost-tourism--news3654

Zhao, Y., (2001). Foreign direct investment and relative wages: The case of China.

(31)

7 Appendix

7.1 The Breusch-Pagan / Cook-Weisberg test for heteroskedasticity H0: constant variance = homostekastic

H1: non-constant variance = heteroskedastic

Breusch-Pagan/Cook-Weisberg test results:

Prob>chi2 Table 2 Regression 1 0,8685 Regression 2 0,7652 Regression 3 0,2365 Regression 4 0,8177 Table 4 Regression 1 0,4939 Regression 2 0,0113* Regression 3 0,3919 Regression 4 0,4976 * Heteroskedastic at 5%

(32)

7.2 Dickey-fuller test for non-stationarity H0: non-stationarity

H1: stationarity

Dickey-fuller test results:

p-value for Z(t)

Ln (nominal wages) 0,8918 Nominal wage growth 0,0234*

Productivity 0,0011*

Education 0,7181

Ln(education) 0,7059

Education growth rate 0,6543 Education % of population 0,8084

Inflation 0,0000*

GDP 1,0000

GDP growth rate 0,0222*

Unemployment 0,7889

Unemployment growth rate 0,0016*

Exports 0,9986

Ln(exports) 0,2835

Export growth rate 0,0000*

Determination of annual nominal wages : the case of China 1985-2012