THE INDIAN HOUSEHOLD SAVINGS LANDSCAPE

Transcription

1 THE INDIAN HOUSEHOLD SAVINGS LANDSCAPE Cristian Badarinza National University of Singapore Vimal Balasubramaniam University of Oxford Tarun Ramadorai University of Oxford, CEPR and NCAER July 2016

2 Savings and Economic Growth Household savings are inextricably linked with economic growth. Savings directly affect economic growth. For example: Mankiw, Romer and Weil, 1992; Romer, 1986;1989. See Deaton (1999) for review. Savings as a consequence, rather than a cause For example, Permanent Income Hypothesis (Campbell, 1987) Precautionary motives: hedge against expected adverse shocks to future income. For example, Chamon and Prasad (2010) on Chinese savings Domestic savings important. Domestic savings and investment rates are highly correlated. (Feldstein and, 1980) For India, the correlation between domestic savings and investment rate = (Seth, 2011) Badarinza et. al. (2016) 2/26

3 Household Wealth: Accumulated Savings Aggregated stock of savings (wealth) embeds many lags of savings decisions, growth rates. Decades of individual saving decisions within a wider macroeconomic context. Influenced by social, and cultural preferences. Asset holdings typically in both physical (housing etc.) and financial assets (stocks, bonds, retirement assets). Optimal mix: Some risk-taking always optimal. Financial assets provide better liquidity and diversification properties than physical assets. Disaggregated view at the household level relevant and important. Badarinza et. al. (2016) 3/26

4 This paper Understand allocation of household wealth in India. Present disaggregated micro-evidence on the landscape and its determinants. Our contribution: 1. Characterization How do Indian households allocate wealth? 2. Comparison Are Indian households similar to those in China and advanced economies? 3. Determinants Do demographic and household characteristics explain differences in wealth allocation patterns? 4. Experience What is the role of macroeconomic experiences (inflation volatility) in household allocation decisions? Badarinza et. al. (2016) 4/26

5 Data All India Debt and Investment Survey, Decennial survey since Assets as on June 2012 for households in India. Multi-stage design, repeated cross-section. Only survey to collect valuation of asset holdings. Detailed asset holdings data, physical and financial. Only data source to capture valuations. Chinese Household Finance Survey, Household Income and Labour Dynamics in Australia, United Kingdom Wealth and Asset Survey, European Household Finance Survey, Survey of Consumer Finances in the United States, Data limitations Badarinza et. al. (2016) 5/26

6 Roadmap 1 Characterizing Aggregate Savings, Disaggregated Wealth 2 International Comparison 3 Determinants of Household Wealth 4 Explaining Heterogeneity in Wealth Allocation 5 Experience and Household asset allocation 6 Policy Implications Badarinza et. al. (2016) 6/26

7 Percent Gross Savings in the Macroeconomy Composition of Gross Household Savings (Flow, per year) Non-financial assets Financial assets Source: Reserve Bank of India > 50% of gross household savings per year in physical assets. Represents flow of capital, each year into the economy. Most recently, physical assets are about 70% of household savings. Macro-data has important limitations (Rangarajan 2009; Kaur 2011). Data limitations Badarinza et. al. (2016) 6/26

8 Characterizing Indian Household Wealth Financial Assets Non-financial Assets Real Estate Durable Gold Total / Housing Goods / Bullion Assets Mean (Rupees) 52,513 14,00,534 58,151 57,870 15,69,066 Notes: N = 107,950. All reported statistics are weighted by the population weights and are in Rupees. Source: All India Debt and Investment Survey, authors calculations Cross-household Variation 10 th 25 th 50 th 75 th 90 th 99 th Financial Assets 0 0 2,200 19,710 95,000 8,51,250 Non-financial assets Real Estate / Housing 0 1,25,600 4,00,000 11,35,000 29,38,350 1,40,00,000 Durable Goods 0 1,000 11,200 41,960 1,07,700 7,57,000 Gold / Bullion 0 3,500 20,000 60,000 1,50,000 5,50,000 Total assets 63,500 1,79,000 4,93,000 13,15,766 33,01,500 1,50,00,000 Notes: N = 107,950. All reported statistics are weighted by the population weights and are in Rupees. Source: All India Debt and Investment Survey, authors calculations Badarinza et. al. (2016) 7/26

9 Roadmap 1 Characterizing Aggregate Savings, Disaggregated Wealth 2 International Comparison 3 Determinants of Household Wealth 4 Explaining Heterogeneity in Wealth Allocation 5 Experience and Household asset allocation 6 Policy Implications Badarinza et. al. (2016) 8/26

10 Percent International Comparison Equally-weighted Average India China USA UK Australia Germany Real estate Durable goods Financial assets Retirement assets Indian and Chinese households 76 and 82 % of wealth in real estate. Real estate accounts for a lower fraction of total assets in the US (43.8%) and Germany (36.7%). Indian and Chinese financial assets include retirement assets. Badarinza et. al. (2016) 8/26

11 Percent International Comparison Value-weighted Average India China US UK Australia Germany Real estate Durable goods Financial assets Retirement assets Representative for the entire population (aggregate composition). India has the highest non-financial assets ratio at 92.19% (China at 87.60%). While non-financial savings is around 70% for Indian households, non-financial assets are around 92%. Badarinza et. al. (2016) 9/26

12 Roadmap 1 Characterizing Aggregate Savings, Disaggregated Wealth 2 International Comparison 3 Determinants of Household Wealth 4 Explaining Heterogeneity in Wealth Allocation 5 Experience and Household asset allocation 6 Policy Implications Badarinza et. al. (2016) 10/26

13 Household Wealth Composition By household characteristics Financial Non-financial Real Estate, Gold, Assets Assets Housing Durables Bullion (1) (2) (3) (4) (5) Education Illiterate / Below Primary school Primary and Middle School Secondary School Diploma Graduate and postgraduate Total Assets < 179, ,000 to 493, ,030 to 1.3mn mn to 1.48 mn > 1.48 mn Age 24 to to to to > Region Type Rural Urban India Badarinza et. al. (2016) 10/26

14 Household Wealth Composition By geographic location Financial Real Estate, Gold, Non Financial Assets Housing Durables Bullion Assets (1) (2) (3) (4) (5) Bihar Rajasthan Uttar Pradesh Madhya Pradesh Orissa Telengana Kerala Gujarat Jammu & Kashmir Tamil Nadu Haryana Maharashtra Andhra Pradesh West Bengal Punjab Assam Karnataka Delhi Himachal Pradesh Arunachal Pradesh Sikkim Chandigarh Badarinza et. al. (2016) 11/26

15 Roadmap 1 Characterizing Aggregate Savings, Disaggregated Wealth 2 International Comparison 3 Determinants of Household Wealth 4 Explaining Heterogeneity in Wealth Allocation 5 Experience and Household asset allocation 6 Policy Implications Badarinza et. al. (2016) 12/26

16 Explaining Heterogeneity in Wealth Allocation Framework Cross-section regression at household level, across all Indian states: f i,k = α + µ k + βx i,k + ɛ i,k f i,k :- Wealth shares of different asset types µ k :- State fixed-effects X i,k :- wealth, age, education, residency, children, daughters Badarinza et. al. (2016) 12/26

17 Explaining Heterogeneity in Wealth Allocation Results Non-financial Real estate Gold ratio ratio ratio Education Illiterate / Below Primary school Primary and Middle School *** *** 0.021*** Secondary School *** *** 0.038*** Diploma *** *** 0.050*** Graduate and postgraduate *** *** 0.042*** Total Assets < 179, ,000 to 493, *** 0.239*** *** 493,030 to 1.3mn 0.070*** 0.317*** *** 1.3mn to 1.48 mn 0.104*** 0.432*** *** > 1.48 mn 0.149*** 0.537*** *** Age 24 to 35 years to 43 years to 50 years to 60 years *** > 61 years 0.015*** 0.031*** * Region Type Rural Urban *** *** 0.069*** Children *** *** 0.024*** *** * Daughters *** 0.011*** *** 0.015*** Constant term 0.877*** 0.593*** 0.191*** State FE Yes Yes Yes No. of obs. 107, , ,950 Adjusted R Badarinza et. al. (2016) 13/26

18 Residual State-level Variation of Asset Ratios BIHAR JHARKHAND WEST BENGAL ORISSA MIZORAM MADHYA PRADESH TRIPURA CHHATTISGARH MEGHALAYA UTTAR PRADESH MANIPUR PUNJAB MAHARASHTRA HARYANA ASSAM RAJASTHAN UTTARAKHAND LAKSHADWEEP JAMMU & KASHMIR GUJARAT TELENGANA NAGALAND KERALA DADRA & NAGAR HAVELI ANDHRA PRADESH KARNATAKA DELHI CHANDIGARH ARUNACHAL PRADESH HIMACHAL PRADESH TAMIL NADU DAMAN & DIU PONDICHERRY GOA SIKKIM ANDAMAN & NICOBAR Real estate assets ratio (state fixed effects) TAMIL NADU PONDICHERRY ANDAMAN & NICOBAR GOA ANDHRA PRADESH KERALA DAMAN & DIU HIMACHAL PRADESH TELENGANA KARNATAKA LAKSHADWEEP GUJARAT SIKKIM UTTARAKHAND DELHI RAJASTHAN JAMMU & KASHMIR MAHARASHTRA HARYANA CHANDIGARH NAGALAND UTTAR PRADESH MADHYA PRADESH ORISSA MANIPUR CHHATTISGARH ASSAM PUNJAB TRIPURA ARUNACHAL PRADESH DADRA & NAGAR HAVELI WEST BENGAL JHARKHAND MEGHALAYA BIHAR MIZORAM Gold holdings ratio (state fixed effects) Badarinza et. al. (2016) 14/26

19 Inflation Uncertainty and Wealth Allocation Inflation and Gold holdings Ratio Gold holdings ratio Estimated coefficient: 7.19*** (std. err.: 2.02) Historical inflation volatility Lack of financial assets / access, demand-side frictions seek alternate vehicles. International investments not that easy (Liberalised Remittance Scheme). Asset space restricted to non-financial, physical assets. Gold, inflation hedge. GDP Controls Badarinza et. al. (2016) 15/26

20 Inflation Uncertainty and Wealth Allocation Inflation and Real-estate assets Ratio Real estate assets ratio Estimated coefficient: -9.17*** (std. err.: 2.08) Historical inflation volatility Real estate, much less liquid than gold, more difficult to liquidate, more risky. Gold: Physical verifiability, Collateral value (property verification, hard.) Liabilities: High inflation volatility Risky to have adjustable rate mortgages. GDP Controls Badarinza et. al. (2016) 16/26

21 Inflation Uncertainty and Wealth Allocation Inflation and Non-financial assets Ratio (mean) inflation_std_dev_0312 GDP Controls Badarinza et. al. (2016) 17/26

22 Roadmap 1 Characterizing Aggregate Savings, Disaggregated Wealth 2 International Comparison 3 Determinants of Household Wealth 4 Explaining Heterogeneity in Wealth Allocation 5 Experience and Household asset allocation 6 Policy Implications Badarinza et. al. (2016) 18/26

23 The role of personal experience Individual experiences risk attitudes (Malmendier and Nagel, 2011) Shocks experienced by different cohorts influence risk attitudes and individual preferences. E.g.: Mark Carney on recent policy and economic uncertainty in the UK: All this uncertainty has contributed to a form of economic post-traumatic stress disorder......long after the original trigger becomes remote, perceptions endure......uncertainty has meant an inchoate sense of economic insecurity for many people despite generalised economic prosperity... Badarinza et. al. (2016) 18/26

24 The Role of Inflation Experiences f i,k = α + µ k + βx i,k + γπ 25 i,k + ε i,k, Non-financial Real estate Gold ratio ratio ratio Inflation experience when young *** State fixed effects Yes Yes Yes Demographic characteristics Yes Yes Yes No. of obs. 78,486 78,486 78,486 Adjusted R Notes: Households with inflation data from 15 Indian States are included in this analysis. Normalised inflation level ( π25 i,k π i,k 25 σ(π i,k 25 ). ) Inflation measured at age 25 of household head. Early-life inflation determines extent of gold holdings. Size of effect is nearly half the contribution of having a daughter in the household. Inflation matters, beyond deep-seated preferences relating to social norms in India. Badarinza et. al. (2016) 19/26

25 Roadmap 1 Characterizing Aggregate Savings, Disaggregated Wealth 2 International Comparison 3 Determinants of Household Wealth 4 Explaining Heterogeneity in Wealth Allocation 5 Experience and Household asset allocation 6 Policy Implications Badarinza et. al. (2016) 20/26

26 Policy Implications Inflation uncertainty 1. Strong inflation target address entrenched inflation expectations. 2. Clarity of instruments, policy objectives and goals reduce uncertainty. Financial access 1. Alleviating dependence on physical savings important. 2. Access to financial savings products not just credit (liabilities). 3. Demand for new instruments will take a long time. Education 1. Strongly correlated with financial assets. 2. May be due to greater access to formal financial system. 3. Policy on financial literacy needs greater engagement. Badarinza et. al. (2016) 20/26

27 Badarinza et. al. (2016) 21/26

28 Inflation Uncertainty and Wealth Allocation Inflation and Real-estate assets Ratio (State GDP Control) Real estate assets ratio Estimated coefficient: -7.86*** (std. err.: 4.18) Historical inflation volatility Back to page Badarinza et. al. (2016) 22/26

29 Inflation Uncertainty and Wealth Allocation Inflation and Gold holdings Ratio (State GDP Control) Gold holdings ratio Estimated coefficient: 6.59*** (std. err.: 2.03) Historical inflation volatility Back to page Badarinza et. al. (2016) 23/26

30 Inflation Uncertainty and Wealth Allocation Inflation and Non-financial assets Ratio (State GDP Control) Non-financial assets ratio Estimated coefficient: (std. err.: 2.04) Historical inflation volatility Back to page Badarinza et. al. (2016) 24/26

31 Data Limitations Macroeconomic Data in India Residual approach generates inaccuracies especially when it pertains to the household sector. Example: Total savings in cash on hand is determined as a proportion of the total currency in circulation and is set at 0.93 since CSO notes, this proportion is likely to undergo change as soon as more data based on survey results of the RBI become available. Rangarajan (2009) notes that while the methodology is sound, there is a need for better data (quality and closing in on gaps). Kaur (2011) also notes that estimating household savings is difficult with the current data availability, and work within these limitations for savings projections. Back to page Badarinza et. al. (2016) 25/26

32 Data Limitations All India Debt and Investment Survey, 2012 Degree of understatement, especially gold unknown. Subramaniam and Jayaraj (2006) document that it is likely that some households on the upper tail of the wealth distribution understate wealth. Drawing lines between micro-enterprises and households very difficult, although the AIDIS attempts to do this carefully. Understatement of real estate and actual cash holdings may be linked to fear of being reported for tax implications. Back to page Badarinza et. al. (2016) 26/26