Housing Prices, Graduates and Income inequality in Singapore

Transcription

1 Housing Prices, Graduates and Income inequality in Singapore By Tilak Abeysinghe Department of Economics, NUS Jointly with Wong Yan Hao SEPF, Oct 31, 2014

2 Income inequality Executive compensation (tied to profit) vs Average worker pay (tied to labor market demand supply) No systematic data Labour income inequality

3 Singapore income Gini among resident employed households

4 Household lifietime income profile: Head 30 years old W9 W8 W7 W6 W5 W4 W3 W2 W

5 Household labour income percentile ratios P90/P P90/P

6 Causes of income inequality (Vast literature) Global forces Income inequality Productivity differential Skill-biased technical change (higher wages in globalized firms) Higher CEO compensations relative to worker pay in globalized firms

7 Causes of income inequality Education Income inequality Mixed results in different countries Income inequality housing prices Reverse relationship not studied

8 Objective For policy purposes, need to understand: Country specific forces Singapore labour market structure Housing prices??? Rapid rise of housing price and falling APC Rapid rise of housing price and falling fertility Is income inequality another aspect?

9 Private median housing prices Four episodes: , , , (1981Q1: 102%) (1994Q3: 47%) (2007Q4: 31%) ( 2010Q2: 38%) Ph Singapore $ mn Ph/CPI

10 Unexpected results Income inequality and higher education inverted U Income inequality and globalization inverted U

11 Theory and data issues Housing price escalations Asset value? Re-distribution of income from home buyers for owner occupation to Developers, rental owners, financiers Need total earnings (labor income + non-labour income)

12 Theory and data issues Data available household labour income Our income inequality measure: Household labour income percentile ratio: P90/P10 Possible links to labour incomes Construction sector Average construction workers Professional & managerial staff Finance sector

13 Data construction methods Income data: annual to quarterly Mixed-frequency non-linear regression (Abeysisnghe and Gulasekaran, 2004 Chow-Lin related series method

14 Data construction methods: Economically active graduate population, 000 Spline interpolation

15 Data: Resident GradShare (%) in total employment

16 Data: Globalization: Productivity ratio (ProdRatio), finance & business services/construction

17 Empirical results: Causality between income inequality and housing price Johansen framework: Vector ECM Y t = Г 0 + αβ Y t 1 + Г 1 Y t 1 + Г 2 Y t 2 Cointegrating relation Zero α s indicates weak exogeneity Zero Г s indicate no Granger causality Weak exogeneity + no Granger causality strong exogeneity

18 Table 1. VECM test results based on VAR(4) specification P90/P10 ln(ph/cpi) GradShare ProdRatio Cointigrating vector *** *** 1.067*** - (0.335) (0.016) (0.260) Adjustment coefficients 0.036** *** (0.016) (0.011) (0.018) (0.051) Granger causality tests on lagged coefficients Δ(P90/P10) (0.508) (0.597) (0.919) Δ(ln(Ph/CPI)) (0.126) (0.647) (0.108) Δ(GradShare) (0.517) (0.640) (0.246) Δ(ProdRatio) (0.319) (0.104) (0.221)

19 Empirical results: OLS and DOLS OLS t-statistics may not have a N(0,1) distribution. Dynamic OLS with one explanatory variable To compute correct t-statistics, need an estimate of long run variance. t-statistic ~N(0,1)

20 Table 2: OLS and dynamic OLS estimates OLS (1) OLS (2) Dept var: P90/P10 OLS (3) DOLS (4) DOLS (5) P90/P20 OLS (6) House price ln(ph/cpi) *** 0.342*** 0.255** 0.188* (0.071) (0.082) (0.125) (0.099) (0.090) GradShare 0.560*** 0.496*** 0.494** 0.510*** 0.571*** 0.149*** (0.020) (0.026) (0.026) (0.043) (0.040) (0.028) GradShare squared *** *** *** *** *** (0.001) (0.001) (0.001) (0.002) (0.002) (0.001) ProdRatio *** 1.565* (0.395) (0.809) (0.789) (0.433) ProdRatio squared *** * (0.058) (0.122) (0.115) ( 0.064) Constant 5.766*** 2.761*** *** (0.095) (0.850) (1.160) (2.065) (1.773) (1.272) Sample period Residual 1 ˆρ Residual ADF -4.76** -4.65** -4.59* ADF 10% critical value R square AIC BIC

21 Findings: Resident Grad Share in total work force, strongest effect on labor income inequality

22 Findings: Globalization Prod Ratio also shows an inverted U effect. Since the 1980s the effect has been to reduce labor income inequality. Similar effect from NODX share in total expenditure. Is Singapore in a phase where globalization is contributing to lower labor income inequality? Needs further analysis.

23 Findings: Housing price Small but significant effect: Rapid rise in housing prices add to increasing labor income inequality. The impact on total income (including non-labor income) inequality is likely to be much larger.

24