Presentation on theme: "Household Savings and Wealth Effect: Evidence from Great Britain."— Presentation transcript:
Household Savings and Wealth Effect: Evidence from Great Britain
Theoretical Foundation Relationship among saving, consumption and wealth explained by Life-Cycle Hypothesis. Modigliani and Brumberg (1954) and Permanent Income Hypothesis by Friedman (1957) It states people tend to smooth consumption during life and they keep a constant marginal utility of consumption People tend to smooth consumption. They borrow in young age, save during working age, dissave when they are old.
Theoretical Foundation 2 When a person has an increase in his wealth he has to increase his consumption before or later A short lag between wealth increase on consumption leads to a strong impact of capital gains house value changes on consumption and vice versa.
Theoretical Foundation 3 Relationship between savings, consumption and wealth can be seen as: (Dynan and Maki 2001) Where ΔC is the target consumption, ΔS is the target saving and mpcΔW is the marginal propensity to consume out of wealth changes
Literature Review Several Studies in US and in other countries with macro and micro data US the value of wealth effect is approximately 4% UK studies by Attanasio and al. (2002) Campbell and Cocco (2005) and Attanasio et al (2005).
Literature Review 2 Attanasio et al. (2002) found that shareholders tend to consume more than non-shareholders Campbell and Cocco (2005) and Attanasio et al. (2005) found heterogeneity in the households behaviour for the effect on consumption from the house price changes. Campbell and Cocco (2005) found a larger impact of house price changes on consumption for older households than on middle aged and younger. Attanasio et al. (2005) found the opposite, a larger effect of house price changes on consumption for the younger households then for the middle aged and a very low for the older.
Data The dataset that we use is the Family Resources Survey. It was launched by the Department of Work and Pension. It collects cross sectional data from 1993 on a various topic as employment, pension,housing, savings.
Data 2 The majority of the studies in this field use the Family Expenditure Survey (FES). We choose the FRS for 2 main reason It has a larger dataset: 26,000 interviewed against 6500-7000 It contains information more detailed about savings, capital gains than the FES.
Data 3 We collected data from 1993/94 to 2005/06 We summed for saving the value of money detained in current accounts, stocks, bonds, gilts, NS&I etc. We summed the capital gains obtained in stocks, bonds etc. For housing we collected data from Nationwide on house price. We calculate the house equity value for all the years and then we took their changes
Methodology We ran a cross sectional regression and also we built a pseudo-panel taking cohorts by age Cross Sectional Regression: Where the subscript i refers at the same individual for the various variables Series of dummies on age, ethnic origins, marital status, kids, area, homeownership.
Methodology 2 Basically with this regression we want to value if house equity appreciation, capital gains and the level of incomes has an effect on the level of saving that an individual detain in a certain moment of time. This is not the Marginal propensity to consume out capital gain, house equity. We do not calculate saving changes
Methodology 3 Panel regression. 12 cross sectional cohorts, 11 time series Here we calculate all variables as changes so we measure the MPC out of capital gains, house equity changes
Descriptive Statistics Saving pretty stable for all the years except an increase in the last ones. Capital gains have a larger variation. Particularly in the last years Capital Gains not very tied with stock market performance. It can be explained with diversification of investments.
Descriptive Statistics 2 For house equity changes there was a peak during 1999-00 then a decrease in the gains. Larger volatility on the house gains after 1999-00 For incomes steady growth in the period object of study
Descriptive Statistics 3 (Cohorts) Saving tend to be lower for younger cohorts, it increases in middle ages and generally it decreases for oldest. This is consistent with life-cycle hypothesis The oldest cohorts has a larger amount of savings in 1999-00 and 2000-01. This cannot be explained well by life-cycle. Oldest cohorts have a larger volatility
Descriptive Statistics 4 (cohorts) Capital Gains is larger for middle age cohorts. It tends to increase for young cohorts and then it decreases for older About house equity it is interesting the fact that old cohorts have a low number of homeowners. This can be a signal of high liquidity for the Britain house market Cohorts of oldest have the lowest incomes for retirement reason I suppose.
Cross Sectional Regression Positive and significant impact on savings for incomes, capital gains, house equity changes Average coefficients of 1,547 for incomes, 0,838 for capital gains and 0,137 for house equity appreciation Values always significant for incomes and capital gains, often significant for house equity appreciation.
Cross Sectional Regression (2) Capital Gains has a larger coefficient when stock market has an increase Age dummies often significant The coefficient is often negative for young it is generally but not always more positive for middle aged than for old. The fact that sometime people being old has a more positive impact on saving is not fully explainable by life-cycle
Cross Sectional Regression (3) Presence of kids sometimes statistically significant. It is more positive when the number of kids is low. Often South East England and some other areas are statistically significant Homeownership always negative and often statistically significant. Homeowners tend to have less savings.
Cross Sectional Regression (4) Positive and sometimes significant value of coefficient for single and couples while it is generally negative and sometimes significant for separed/divorced and widowed Sometimes we found a significant impact and often positive for some ethnic groups like Indian Subcontinent and White
Panel Regression No significance for all variables Positive impact on savings changes by house equity changes (0,095) and capital gains (0,138). Capital gains and housing equity appreciation lead to an increase of savings Negative impact of incomes (-0,057). An increase of incomes lead to a decrease of savings.
Panel regression (2) We run a regression dividing the cohorts in 3 groups: young, middle aged and old We found a negative coefficient for incomes for young and middle aged and positive for old This means that people tend to decrease their savings when they have an income increase if they are young and even more if they are middle aged while they increase their savings if they are old.
Panel regression (3) Young and middle aged increase their savings when they have positive capital gains while old people tend to decrease them People tend to increase their savings when they have an increase in their house equity value particularly old people These results has to be considered with cautions because the number of observations is limited and the coefficients are not statistically significants
Next Steps To measure the impact on savings by the various sources of capital gains. Bond, stocks etc. This can be interesting for the assumption of fungibility of the life-cycle hypothesis. This states that all the different sources of incomes has the same marginal propensity to consume.
Next Steps (2) To value the impact of shocks in capital gains and in housing equity appreciation on savings. This is interesting to value because the life-cycle hypothesis states that people behave differently when the shocks is expected or unexpected. In case of expected shock the individual incorporate the change from the moment when the occurrence of the shock is known. When the shock is unexpected the change occurs from the moment in which it happens.