Consumption Volatility, Liquidity Constraints and Household Welfare

Transcription

1 Volatility, and Household Welfare Olga Gorbachev, University of Delaware, USA Keshav Dogra, Columbia University, USA RES 2011 April 18, 2011

2 GOALS AND CONTRIBUTIONS What impact did increased income uncertainty and financial liberalization have on household welfare ( )? Income

3 GOALS AND CONTRIBUTIONS What impact did increased income uncertainty and financial liberalization have on household welfare ( )? 1. Combine SCF and PSID data to construct panel data on liquidity constraints (using direct measure of constraints) 2. Assess the evolution of income uncertainty for liquidity constrained and unconstrained households, and for different groups 3. Assess household consumption volatility for constrained and unconstrained households. Income

4 Major Findings: 1. Although household indebtedness increased substantially, the proportion of liquidity constrained households did not fall Income

5 Major Findings: 1. Although household indebtedness increased substantially, the proportion of liquidity constrained households did not fall 2. household income AND household consumption increased between 1970 and 2004 for all demographic groups. Liquidity constraints played an important role. Income

6 We directly measure liquidity constraints using SCF: 1. In the past five years, has a particular lender or creditor turned down any request you, or your husband/wife, made for credit, or not given you as much credit as you applied for? 2. Were you later able to obtain the full amount you, or your husband/wife, requested by reapplying to the same institution or by applying elsewhere? 3. Was there any time in the past five years that you, or your husband/wife, thought of applying for credit at a particular place, but changed your mind because you thought you might be turned down? Income

7 Figure: Proportion of liquidity constrained households, in SCF year Income with 95% confidence intervals

8 Why Did The Probability Of Being Constrained NOT Fall? Our findings challenge the common assumption that liquidity constraints relaxed since 1983 due to increased supply of credit. Percentage with some debt increased by 11%, while average debt for those with some debt increased by 134%. But Glaeser, Gottlieb, and Gyourko [2010] find % mortgage applications rejected did not decline, consistent with our findings Income

9 Why Did The Probability Of Being Constrained NOT Fall? Our findings challenge the common assumption that liquidity constraints relaxed since 1983 due to increased supply of credit. Percentage with some debt increased by 11%, while average debt for those with some debt increased by 134%. But Glaeser, Gottlieb, and Gyourko [2010] find % mortgage applications rejected did not decline, consistent with our findings Possible explanation for increase in debt and no change in % constrained: demand for debt increased along with the supply Dramatic fall in interest rates, increase in housing prices, increase in education and health costs, increase in income uncertainty Income

10 Estimating in PSID We estimate the probability of being constrained as a function of demographic and financial variables common to PSID and SCF We allow for coefficients to change over time (thus, extending work by Jappelli, Pischke, and Souleles [1998]) Income

11 Measuring Income We model income process as (similar to Blundell, Pistaferri, and Preston [2008]) ln(y h,a,t ) = Z h,a,tϑ t + P h,a,t +ν h,a,t (1) Income

12 Measuring Income We model income process as (similar to Blundell, Pistaferri, and Preston [2008]) ln(y h,a,t ) = Z h,a,tϑ t + P h,a,t +ν h,a,t (1) Then income shocks can be computed from Income ln(y h,a,t ) = ln(y h,a,t )+ς h,a,t + ν h,a,t (2)

13 Measuring Income We model income process as (similar to Blundell, Pistaferri, and Preston [2008]) ln(y h,a,t ) = Z h,a,tϑ t + P h,a,t +ν h,a,t (1) Then income shocks can be computed from Income ln(y h,a,t ) = ln(y h,a,t )+ς h,a,t + ν h,a,t (2) Income volatility(σ Y h,t+1 )2 is the squared residual and is composed of household specific shocks to permanent ς h,a,t and transitory ν h,a,t income. We do not separately identify permanent and transitory shocks to income, but we will be able to indirectly observe the evolution of these shocks by looking at volatility of consumption.

14 mean Figure: Mean Family Income Shocks Income year year all households married single parent mean year year white black/hispanic edu>12 edu<13

15 To estimate volatility of consumption, we follow methodology outlined in Gorbachev [2011] We use PSID data 1970 to 2004 (limited to food consumption). We use consumption model with nonseparable preferences for food and other nondurables allow for measurement error and precautionary savings, and control for liquidity constraints and labor decisions. construct volatility based on the unpredictable component Robustness checks: separable utility, wealth effects, extension to nondurables based on CEX-PSID data. Income

16 Using rational expectations, and taking 2nd order Taylor-approximation of the Euler equation, we get: lnf h,t+1 = β 0 ln(1+δ h )+β 1 lnθ h,t+1 + β 2 ln(1+r h,t+1 )+β 3 lnp t+1 + β 4 (σ Y h,t+1) 2 +β 5 ln(1+λ h,t+1 )+z h,t+1 Income

17 Using rational expectations, and taking 2nd order Taylor-approximation of the Euler equation, we get: lnf h,t+1 = β 0 ln(1+δ h )+β 1 lnθ h,t+1 + β 2 ln(1+r h,t+1 )+β 3 lnp t+1 + β 4 (σ Y h,t+1) 2 +β 5 ln(1+λ h,t+1 )+z h,t+1 Income Use Arellano and Bover [1995] with orthogonal transformations to get consistent estimator, allowing for heteroskedasticity and intragroup correlation (1) to control for household specific discount factors, and (2) to address endogeneity due to presence of second and higher-order terms in the residual.

18 Table: Evolution of Food Volatility OLS OLS FE, AR(1) Prais Year/ *** *** (0.23) (0.47) (0.01) (0.41) (σh,t Y ) ** (0.01) (0.01) (0.01) (σh,t 2 Y ) ** ** (0.01) (0.01) (0.01) (σh,t 4 Y ) ** ** (0.01) (0.01) (0.01) Pr(denied)t 0.130*** 0.089* 0.122*** (0.04) (0.05) (0.04) mst 0.103*** 0.102*** 0.075*** 0.099*** (0.01) (0.02) (0.01) (0.01) Const ** (0.45) (0.94) (0.01) (0.81) Obs. 40,246 17,644 18,075 22,103 Number of pid 3,405 R Within R Between R F-test that all ui = p-value 1 Rho Durbin-Watson Statistic Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Source: PSID and SCF data as described in the text. Income

19 Conditional on income volatility and probability of being constrained, race, edu, marital status have no effect on consumption volatility Income

20 Conditional on income volatility and probability of being constrained, race, edu, marital status have no effect on consumption volatility But demographics affect consumption volatility through their effect on liquidity constraints. Out of all unconstrained, only 9% were single parents and only 11% nonwhite Almost 50% of single parents are constrained relative to less than 20% of married Around 40% of nonwhite are constrained vs. 20% of white Income

21 Conditional on income volatility and probability of being constrained, race, edu, marital status have no effect on consumption volatility But demographics affect consumption volatility through their effect on liquidity constraints. Income Additionally, single parents and nonwhites are significantly more likely to be asset poor

22 Conditional on income volatility and probability of being constrained, race, edu, marital status have no effect on consumption volatility But demographics affect consumption volatility through their effect on liquidity constraints. Income Additionally, single parents and nonwhites are significantly more likely to be asset poor These inequalities in wealth and access to credit have large impact on inequalities in hh consumption volatility and welfare.

23 s Probability of being denied credit did not fall during the period. Family income volatility increased by at least 30% between 1976 and volatility increased by 23%, particularly for those with increase in income volatility and higher liquidity constraints. We estimate that the welfare gain from lowering volatility of consumption to its 1976 level would be equivalent to increasing annual nondurable consumption by 2%, or by $250 bil. Eliminating liquidity constraints would increase annual nondurable consumption by b/w 1.4 and 2.7%. Income