Income Inequality and An Economy in Transition Zhichao Yin Southwestern University of Finance and Economics June, 2014
Outline An Introduction to China Household Finance Survey Household income and Income Inequality in China Inequality and Insufficient Consumption Household Assets and the Housing Market 2
An Introduction to China Household Finance Survey
Sampling In 2011, divide 2,585 counties/districts of the country in 28 provinces (except Tibet, Xinjiang, and Inner Mongonia) into 10 groups by per capita GDP and (PPS) randomly select eight counties/districts from each group. Total 80 counties. In 2013, revisit the households of 2011. An expansion of the sample size: the symmetric sampling. Ranking all counties in each province by per capita GDP. Draw counties in each province symmetric to the counties in 2011. 182 new counties (total 262) are in the full sample.
Coverage of CHFS
Sampling Four communities (neighborhood, village) were PPS selected. Number of neighborhoods or villages are chosen based on the urban resident ratios. Total 320 communities. For each community, carefully mapping all apartments/houses to be used as the end-sampling frame. Based on local average housing prices, randomly draw between 20 to 50 households from each community. Total sample size: 8,438 households/29,350 individuals.
Household Map for each community: manually drawn
Household Map for each Community: computer assisted
Implementation Help from local branches of the Central Bank and the Agricultural Bank of China. Representatives from local bank branches introduce the project and interview teams to local communities. Representatives from local communities introduce our interviewers to the selected households. Maintaining close relationships with communities is the key. Help from SRC/NORC/Fed/CHALRS/CFPS/CGSS
Implementation Every interviewer received 56 hours of training Implementation arrangements: Households had to refuse to be interviewed six times before being excluded from the survey Working in teams ensured safety and reduced moral hazard Dedicated and innovative interviewers (students) from Southwestern University of Finance and Economics.
Sample size 2011 sample: 8,438 households, 29,324 individuals. More than 600 students from SWUFE participated. 2013 sample: 28,141 households, more than 99,000 individuals. More than 1,600 students from SWUFE participated.
Overview of the Questionnaire Demographic characteristics and labor market Assets and liabilities Non-financial assets Family business Land and real estates Vehicles Other non-financial assets Financial assets Social and Commercial Insurance Expenditure and non-labor income
Overview of the Questionaire Example: Financial assets Checking CD Stocks Bonds Mutual fund Derivatives Financial wealth-management products Non-RMB assets Gold Cash Lending
CHFS Low Refusal Rates Year Overall Urban Rural 2011 11.6% 16.5% 3.2% 2013 10.9% 15.4% 0.9% Refusal Rates of Old/New Sample in 2013 Overall Urban Rural 2013 New Sample 12.6% 17.4% 0.9% Successfully contacted 2011 sample 5.4% 8.2% 0.7% 2013 contact success rate 82.1% 2011 sample response rate 77.7%
Low refusal rate: international comparison Datasets Year Refusal rate Survey of Consumer Finance (SCF, US) 2007 AP Sample List Sample 32.3% 67.3% Consumer Expenditure Survey (CEX, US) 2005 Interview Diary 25.5% 29% Survey of Household Income and Wealth (SHIW, Italy) Eurosystem Household Finance and Consumption Survey (HFCS) 2008 43.9% 2010 Belgium 57.6% Germany 69.7% France 30.0% Portugal 10.3% Finland 11.1%
Monthly Telephone Interviews We obtained households phone numbers from face-to-face interviews. Computer Assisted Telephone Interview (CATI) questionnaire includes: Expectations about interest rate, CPI, housing price, stock index Employment status Financial market participation and earning status Assets: house price, loan, vehicle, debts Income and consumption Indicators of rural area development
Is it representative? Comparing with the NBS data Indicator NBS CHFS Urban Population 0.513 0.497 (by residence) 0.342 0.360 (by Hukou) Household Urban 2.89 3.04 residence Rural 3.98 3.78 Average Age 36.87 38.96 Proportion of Males 0.514 0.505
Sample age structure comparison Population Age Structure 11.0 0-9 9.3 13 10-19 10.4 17.1 17.3 16.2 20-29 30-39 40-49 15.9 15.2 17.2 12.0 50-59 14 7.5 60-69 10.3 4.3 70-79 5.4 1.6 80+ 2.3-20 -15-10 -5 0 5 10 15 20 Census 2010 CHFS 2011
Building a baseline database for China Original motivation is to for academic purpose. Later development suggested different perspectives. Asset information provides a key baseline information. Household income and much of other information provide verification for NBS data. Proven to be useful for business and people s daily life.
Household Income and Income Inequality
Disposable Income and its Growth Urban Rural NBS CHFS CHFS / NBS NBS CHFS CHFS / NBS 2010 19,109 24,687 129% 5,919 9,373 158% 2012 24,565 28,714 117% 7,917 10,473 132% Two year growth rate 28.2% 16.3% 33.7% 11.7% CHFS per capita income is higher than NBS, but its growth rate is lower than NBS.
Household Income Gini Coefficient Overall Urban Rural 2010 0.607 0.577 0.606 2012 0.606 0.572 0.599 NBS 2010: 0.481 2012: 0.474 2013: 0.473 2010 world average:0.44
Per capita income by category in 2012 NBS Urban CHFS CHFS / NBS NBS Rural CHFS CHFS / NBS Salary 17336 14988 86.5% 3448 5296 154% Business 2548 6993 274% 3533 3805 108% Asset income 707 1831 259% 249 72 28.9% Transfer income 6368 6792 107% 687 2198 320% CHFS has higher income in salary, business and transfers than NBS.
Why so much higher than NBS? Comparing distributions Vastly different at low income groups and high income groups. Caused by difference in high income groups
2012 per capita income by quintile for urban residents NBS CHFS CHFS/NBS Overall 24,565 30,604 125% Lowest 20% 10,352 3,457 33.3% 20% 40% 16,761 10,781 64.3% 40% 60% 22,419 17,307 77.2% 60% 80% 29,814 27,742 93.1% 80% 90% 39,605 43,399 110% Highest 10% 63,824 128,910 202% High income group: CHFS >> NBS
2012 per capita income by quintile for rural residents NBS CHFS CHFS/NBS Overall 7,917 11,370 144% Lowest 20% 2,316 988 42.7% 20% 40% 4,808 3,270 68.0% 40% 60% 7,041 6,483 92.1% 60% 80% 10,142 10,589 104.4% Highest 20% 19,009 28,716 151% High income group: CHFS >> NBS
Quantiles of per capita disposable income 2012 2012 Urban 2012 Rural Quantiles CHFS NBS (estimated) CHFS/ NBS CHFS NBS (estimated) CHFS/ NBS 25% 8055 14824 0.54 2462 4330 0.57 50% 17400 21948 0.79 6240 6868 0.91 75% 32660 32332 1.01 12330 10830 1.13 90% 57125 46548 1.23 20900 16308 1.28 95% 86895 57054 1.52 29018 21037 1.38 99% 228600 86819 2.63 63125 35773 1.76 Estimation is carried by fitting a log-normal distribution
Why CHFS Gini is much higher than NBS? Major differences in the rich household sample. Nationwide Urban Rural Excluding the lowest 0.5% sample 0.603 0.574 0.603 Excluding the lowest 5.0% sample 0.586 0.560 0.581 Excluding the highest 0.5% sample 0.564 0.541 0.541
How to understand the high Gini? A consequence of market economy and efficient resource allocation. More developed market economies have higher Ginis than less developed market economies. East: 0.60 Central: 0.56 West: 0.54 Very little change of Gini when excluding households working in monopoly industries and public sector employees. Excluding households work in monopoly industries: 0.60 Excluding households work in public sector: 0.60
Gini of OECD countries based on market incomes 0.60 0.50 0.50 0.47 0.53 0.47 0.47 0.51 0.41 0.46 0.46 0.44 0.46 0.44 0.49 0.40 Gini 0.30 0.20 0.10 0.00 before transfer Gini coefficients of market income from OECD countries are close to 0.5. China has much higher heterogeneity than OECD countries, and very little income transfer programs. Gini in China is expected to be higher than 0.5.
Income Inequality and Insufficient Consumption
Rising Household Saving Rates 2010 2012 Urban Rural Overall Urban Rural Overall Income 24687 9373 16990 28714 10473 20659 CHFS Consumption 16878 7236 12031 19167 7693 14100 Saving Rate 31.6% 22.8% 29.2% 33.3% 26.5% 31.8% Income 19109 5919 12472 24565 7917 16669 NBS Consumption 13472 4382 8898 16674 5908 11568 Saving Rate 29.5% 26.0% 28.7% 32.1% 25.4% 30.6%
Proportion of households have positive saving in that year 2010 2012 Urban 56.8% 64.1% Rural 57.0% 55.6% Overall 56.9% 60.6%
Unequal distribution of saving Saving rates Income group 2010 2012 Highest 5% 73.5% 72.2% Highest 10% 66.5% 45.2% Highest 25% 56.4% 42.9%
Proportions of total saving and financial assets from high income households (2012) Percentage of total Saving Financial assets Highest 20% Highest 10% Highest 5% Highest 20% Highest 10% Highest 5% Urban 75.8% 62.4% 50.3% 54.3% 39.6% 27.4% Rural 76.8% 59.4% 44.4% 50.6% 37.2% 26.5% Overall 77.1% 62.4% 50.6% 60.6% 44.6% 30.6%
Proportion of positive saving by non-housing assets and by income 2010 Non- Housing Asset Lowest 25% Lowest 25% Income groups 25%-50% 50%-75% Highest 25% 19% 53% 75% 89% 25%-50% 14% 53% 82% 90% 50%-75% 4% 45% 74% 85% Highest 25% 1% 26% 53% 84% Poor do not have money to consume. Rich already consume at the optimal.
Proportion of positive saving by non-housing assets and by income 2012 Urban Income category Non- Housing Asset Lowest 25% Lowest 25% 25%-50% 50%-75% Highest 25% 24% 73% 95% 98% 25%-50% 20% 67% 89% 96% 50%-75% 10% 55% 80% 95% Highest 25% 7% 34% 70% 90%
Various measures to encourage Chinese consumption had limited success China s poor don t have money to spend China's rich are already spending what they need, and pocketing most of the rest. Unequal income distributions and liquidity constraints cause insufficient domestic demand. Improving income distribution would promote economic restructuring.
Short Term Solution: Transfer Payments
Increasing the Minimum Wage Does Not Have Much Effect on the Gini Coefficient Overall Urban Rural GINI before adjustment 0.61 0.56 0.60 Strict enforcement of the current minimum wage Minimum wage increases by: 0.58 0.55 0.56 50% 0.58 0.54 0.56 100% 0.58 0.54 0.56
Income Tax Policy Does Not Have Much Effect on Gini Coefficient Household income Salary and wage income Before tax After tax Before tax After tax Overall 0.61 0.61 0.49 0.48 Urban 0.57 0.56 0.47 0.46 Rural 0.60 0.60 0.49 0.48
Brazil Experience 16 0.62 Transfer payment/gdp(%) 14 12 10 8 6 4 8.5 0.61 10.6 0.61 11.7 0.60 12 12.1 0.59 0.58 13 0.57 13.4 0.55 0.61 0.60 0.59 0.58 0.57 0.56 0.55 0.54 Gini 2 0.53 0 1990-1991 1996-1997 1998-1999 2002-2003 2004-2005 2006-2007 2008 0.52 Transfer payment/gdp Gini A large scale of transfer payment can reduce Gini effectively
Transfer payments can effectively reduce Gini Gini 0.60 0.50 0.40 0.30 0.50 0.30 0.47 0.26 0.53 0.34 0.47 0.47 0.27 0.29 0.51 0.34 0.41 0.25 0.46 0.46 0.32 0.33 0.44 0.31 0.46 0.44 0.33 0.32 0.49 0.38 0.20 0.10 0.00 before transfer after transfer
Ratio of social welfare spending to fiscal expenditure : China vs US China U.S. Excluding social security funds 12.3% 36.6% Including social security funds 21.2% 46.7% According to the US Congressional Budget Office: the poorest 20% of households: market income: $ 7,500; after transfer payments: $ 30,000.
Financial Resources to Implement Large-scale Government Transfer Payments In 2012 China's total fiscal revenue was more than 11.7 trillion yuan, an increase of 1.35 trillion RMB (2010-2012 average annual growth of 1.62 trillion) State-owned enterprises realized profits of 1.98 trillion RMB in 2010, more than 2 trillion RMB in 2011. Current state budget deficit is 1.6% of GDP, accounting for only 8% of tax revenues; room for additional budget deficit for redistribution. 70% state-owned profit +50% incremental revenue + additional deficit (GDP 2%) = 3.8 trillion RMB: redistribution amount reaches 36% of government spending, similar to the US level.
3.8 trillion RMB in Transfer Payments Can Dramatically Narrow the Income Gap Subsidy per family (Yuan) Overall City Rural Before transfer 0.61 0.56 0.60 Subsidize all 9,500 0.49 0.48 0.43 Subsidize the bottom 80% 12,800 0.46 0.45 0.38 Subsidize the bottom 60% 15,800 0.42 0.44 0.32
Incentive Compatible Welfare System Conditional Cash Transfer (CCT) project Motivate families health and education investments Example: Free Lunch Program Earned Income Tax Credit system (EITC) Negative tax rate for working class 30% of American households benefit from this system, according to documented negative rates of up to 30% Currently happening within governments
Non-production government transfers in China used for consumption Income group before subsidies % of households have subsidies % of subsidies used for consumption Lowest 20% 30% 77% 20%-40% 20% 56% 40%-60% 17% 29% 60%-80% 13% 20% Highest 20% 11% 9% Overall 18% 48% Transfers to low-income households can effectively raise consumption.
Long Term Solution: Education
Comparison of China and OECD Countries Expenditures on Education per Student (USD) Primary Middle High Higher Education Japan 2009 7,729 8,985 9,527 15,957 S. Korea 2009 6,658 7,536 11,300 9,513 US 2009 11,109 12,247 12,873 29,201 OECD Average 2009 7,719 8,854 9,755 13,728 China 2011 801 1,055 968 1,547 Chinese government spends too little in education.
Income gap narrows as the education increases Gini Coefficient Primary school or below 0.56 Middle or high school 0.56 Junior college 0.52 College or above 0.50
Improving the educational can lower the Gini in the long run Overall Urban Rural 2010 0.61 0.56 0.60 Improve the education to: -Average level of the OECD countries 0.40 0.47 0.33 -Average level of the US 0.42 0.44 0.34
Conclusions The current income inequality in China is substantial. High Gini coefficients are common in economies with high growth rates, and it s also a natural consequence of the efficient allocation of resources. In short-term, China s government has fiscal capacity to reduce the income inequality through the transfer policy. In long-term, China s government can invest more in education and reduce the inequality of opportunity to reduce the income gap.
Household Wealth and Housing
Growth rates of Household Wealth Total assets Housing Overall 19.6% 26.8%
Household asset distribution in 2011 Top 10% s share in total assets:63.9%
Household Asset Gini Coefficients in 2011 0.78 0.76 0.74 0.761 0.737 0.72 0.7 0.697 0.68 0.66 0.64 0.62 Overall Urban Rural
Asset inequality improved substantial over last two years Middle-asset households (30%-70%) s share in total assets: 2010 : 10.7% 2013: 13.5%
Household Asset Gini decreased 0.78 0.76 0.74 0.72 0.7 0.68 0.66 0.64 0.62 0.761 0.737 0.717 0.697 0.681 0.675 Overall Urban Rural 全国城市农村 2011 年 2013 年
Asset components in China and US China US 2010 2012 2010 Housing 62.7% 66.4% 40.6% Agriculture 0.5% 1.3% Land 4.8% 3.5% 17.5% Business 16.4% 11.8% Vehicle 3.1% 4.5% 3.2% Other non-financial 4.0% 2.3% 0.8% asset Financial asset 8.3% 10.1% 37.9% Huge difference in housing and financial assets
70.0% 60.0% Shares of different assets 66.4% 62.7% 50.0% 40.0% 30.0% 20.0% 10.0% 16.8% 13.2% 4.8% 3.5% 10.1% 7.1% 6.8% 8.3% 0.0% 农业工商业 Business Housing 房屋 Land 土地耐用品汽车 Durable/car 金融资产 Financial 2011 年 2013 年
Housing wealth as a percentage of total wealth Price/income ratio 83.8% Beijing 90.0% 25 Shanghai 76.5% 20 60.0% 15 30.0% 10 5 0.0% 0 青海省山东省宁夏回族自 安徽省四川省陕西省广西壮族自 湖北省江西省河北省吉林省河南省内蒙古自治区黑龙江省湖南省山西省重庆市云南省广东省甘肃省浙江省海南省辽宁省天津市福建省贵州省上海市江苏省北京市
The Rich:Top 5% Criteria Quantile Average Wealth Average net wealth Average Income By wealth 5% 2,629,850 6,507,023 6,205,210 271,811 By net wealth 5% 2,534,100 6,489,605 6,225,171 268,726 By income 5% 189,510 3,809,924 3,643,891 452,095
The Rich:Top 1% Criteria Quantile Average Wealth Average net wealth Average Income By wealth 1% 7,393,500 16,300,000 15,400,000 494,322 By net wealth 7,132,000 16,300,000 15,500,000 507,322 1% By income 1% 485,000 8,051,315 7,767,108 1,151,662
Immigration 14.0% 12.0% 11.6% 10.0% 8.0% 6.0% 6.0% 6.8% 8.1% Plan Wait and See 4.0% 2.0% 2.3% 2.8% 0.0% National Top 5% Top 1%
Home ownership in the city Home ownership rate % with only one house % with two or more houses 2011 84.7% 69.2% 15.5% 2013 87.0% 68.1% 18.9% Home ownership: World average: 63% US: 65% Japan: 60%
Housing Demand No house Total Demand Residential demand Has house New migrants House in a different city Living with parents Annual incremental demand New adults Demolition and relocation 68
Sources of incremental demand 2012 no house 14.6% Has house but house in a different city 6.1% Has the demand of living apart 5.6% Total residential demand in % 26.3% Total residential demand in units 56.62 million Upgrading demand in % 13.6% Upgrading demand in units 29.28 million
Housing Supply Total Demand Housing stock Incremental supply Multi-housing Commercial residential housing Production capacity: startups
Demand and Supply Analysis 2012 Current total residential demand Current total supply Difference Annual incremental demand Startups in 2012 85.9 million 47.1 million 38.8 million 5.8 million 13.3 million The housing demand will be satisfied within 5 years given current industry capacity. Only 40% of current capacity is needed to satisfy incremental demand.
% of home buyers housing status at the purchasing year 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% <1998 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 首套房 No house 改善型住房 Upgrading 投资性购房 Investment
Regulation suggestions of housing market Basic characteristics: Oversupply price expectation leads to price appreciation. Regulation suggestions: Less new construction, particularly less affordable housing construction. Push the existing stock into the market Effectively guide price expectation
Provide accurate information to the market to help develop reasonable price expectation: Put more efforts on collecting and releasing information in time Encourage independent research organization to complement government effort. Replace affordable housing construction with rental vouchers. Make use of the stock of multiple houses. Build an effective renting market.
Thank You! www.chfsdata.org Let China understand herself. Let the world know China.