What Can a Life-Cycle Model Tell Us About Household Responses to the Financial Crisis?

Transcription

1 What Can a Life-Cycle Model Tell Us About Household Responses to the Financial Crisis? Sule Alan 1 Thomas Crossley 1 Hamish Low 1 1 University of Cambridge and Institute for Fiscal Studies March 2010

2 Data: Household Saving Rates Source: IFS

3 Data: Household Saving Rates Source: IFS

4 Data: Flight to Cash Net Flows into Equity

5 Key Questions Sharp rise in savings ratio associated with the current recession and stock market crash 1 Why? Behavioural: Risk aversion soars as growth doubts intensify (FT, 2009) Life-cycle behaviour under uncertainty 2 How permanent?

6 Understanding the Rise in Saving Why not dissave and use wealth to smooth shocks and smooth consumption?

7 Understanding the Rise in Saving Why not dissave and use wealth to smooth shocks and smooth consumption? What is a recession? Fall in income (permanent or transitory) Increased uncertainty Possible stock-market crash (wealth destruction)

8 What is a Recession: Greater Risk? Decompose shocks to income into permanent and transitory How does the variance of each evolve over the business cycle? Use changes in the cross-section variance of consumption to identify the permanent shocks See Blundell, Low, Preston (2009)

9 What is a Recession: Greater Risk? Figure: Permanent Variance: UK Source: Blundell, Low, Preston (2009)

10 Understanding Saving: Life-cycle Framework Standard life cycle dynamic portfolio allocation model Idiosyncratic income uncertainty Uncertainty higher in recessions Possibility of recession: 2 state Markov process Rate of return uncertainty Possibility of stock market crash Heterogeneous agent model, explicit aggregation from micro to macro

11 Model max E t " T t j=0 # (C h,t+j ) 1 γ 1 1 γ (1 + δ) j X t+1 = (1 + r e t+1)s t + (1 + r)b t + Y t+1 where S t is stock holding, B t is bond holding.

12 Income Process Y t+1 is stochastic (idiosyncratic) labour income: Permanent income: ln Y t = ln Y P t + u t, u t N(0, σ 2 u) ln Y P t = ln Y P t 1 + d t + η t

13 Income Process Y t+1 is stochastic (idiosyncratic) labour income: Permanent income: ln Y t = ln Y P t + u t, u t N(0, σ 2 u) ln Y P t = ln Y P t 1 + d t + η t Variance of permanent idiosyncratic shocks driven by aggregate shocks: η t N(0, σ 2 n,l ) in boom η t N( θ, σ 2 n,h ) in recession Business cycle is 2x2 Markov matrix (2/3, 1/3)

14 Stock Market Process Excess returns are iid where ε t+1 v N(0, σ 2 ε ) r e t+1 r = µ + ε t+1 Crash gives return of φ Probability of a crash is p H in a recession, p L in a boom, p H > p L.

15 Parameter Values γ = 2.0 δ = 0.08 σ n,l = 0.1 σ n,l = 0.15 coe cient of relative risk aversion discount rate permanent shock in boom permanent shock in recession p L = 0.01 probability of a stock market crash in boom p L = 0.02 probability of a stock market crash in recession φ = 50% size of the stock market crash σ ε = 0.2 standard deviation of stock returns µ = 0.06 mean equity returns r = 0.02 interest rate d = 0.02 dividend yield (non-stochastic)

16 Baseline No realised recession, no realised stock market crash

17 Simulated Scenarios Stock market crash occurs: e ects on di erent cohorts depending on age at crash (30,50,70) Recession (increased uncertainty) occurs and lasts 5 periods: e ects on di erent cohorts depending on age at crash (30,50,70) Show deviations from the baseline for consumption, savings rate, asset holdings

18 Simulations: E ects on Consumption Consumption never recovers following crash Consumption overshoots after recovery from a recession Young cohorts hit more by recession, older cohorts more by crash

19 Simulations: E ects on Savings Rate

20 Simulations: E ects on Asset Holdings

21 Simulations: E ect on Equity Holdings Temporary Flight to Cash

22 Conclusions Data: increase in savings ratio and ight to cash following current recession Recession characterised by increased uncertainty and stock market crash

23 Conclusions Data: increase in savings ratio and ight to cash following current recession Recession characterised by increased uncertainty and stock market crash Why has savings spiked? Simple life-cycle model can generate spike either through crash or recession How permanent is the increased savings? Post-stock market crash: savings rate would remain high Post-recession: run down extra precautionary balances - fall in the savings rate will overshoot

24 Conclusions Ongoing work: relative contributions to saving decline and predict likely persistence of high savings rates 1. FTSE recovered; household less exposed to stock market than model implies

25 Conclusions 2. Redundancy rates spiked, but are now declining