Deflation, Credit Collapse and Great Depressions Enrique G. Mendoza
Main points In economies where agents are highly leveraged, deflation amplifies the real effects of credit crunches Credit frictions induce amplification and asymmetry (i.e., Great Depressions or Sudden Stops ) in response to standard shocks The transmission mechanism features a pure balance sheet effect and Irving Fisher s debt-deflation process The contribution of the Fisherian deflation is substantial
Fisher s debt-deflation mechanism Assume a period of economic expansion and rising relative prices (prices of specific goods and assets relative to unit of denomination of debts, e.g. CPI) Example: Liability dollarization (debts in units of tradables, leveraged on assets/incomes of nontradables) Credit constraints limit debt to a fraction of the value of income in units of tradables Pure balance sheet effect: borrowing constraint lowers tradables consumption and price of nontradables Fisherian deflation: spiral of tightening debt constraints and falling nontradables prices
Modeling debt deflations Households consume tradables and nontradables, T N combining them according to c( ct, ct ) (e.g. if the composite good is Cobb-Douglas c c T, c N ( c T α ) ( c N α = ) ) 1 ( t t ) Utility function is standard (assuming infinite life horizon) T N ( (, )) u c c c t t t 0 (1 ρ) t t = + Budget constraint (with a tax on nontradables purchases) Credit constraint T N N T N N t + (1 + τ t) t t = t + t t t+ 1 + t + t c p c y p y b br T T N N ( ) bt+ 1 κ yt + pt yt Ω
Modeling debt deflations Government budget constraint: N N T N N tpt ct g pt g Tt τ = + + Resource constraints: market clearing conditions obtained by combining budget constraints of households and government T T T c + g = y b + Rb t t t+ 1 t N N N t + = t c g y Assume: βr=1 and y N t = y N
Perfectly smooth equilibrium (PSE) Recall at the optimal savings plans, IMRS=R, and that when βr=1, this implies that, if credit constraints never bind, consumption of tradables is smoothed perfectly as a constant fraction of wealth: T = (1 β) T 0 + 0 0 t T t t = 0 [ ] ( ) c W Rb g with W R y For nontradables, MRS between CT and CN must equal its relative price. Since CT is constant and demand must equal supply of nontradables we get: T 1 1 T N c c T N p τ, where t =Φ + Φ is MRS( c, c ) ( t) N N N N y g y g
Wealth-neutral shocks to tradables income T T Wealth-neutral shocks ( y0, y1 ) such that the date-0 endowment falls and date-1 endowment rises so as to keep PV of tradables income constant T T ( ) y1 y T T T rw0 = ( y y0 ) with y = is permanent income 1+ r 1+ r If credit constraint does not bind, or credit markets are perfect, we preserve PSE. Agents borrow at date 0 and repay later ( ) T T 1 0 0 2 1 1 0 t 0 b b = y y < 0, b b = b b > 0, b = b for t 2
Fisherian deflation equilibrium (FDE) Unanticipated wealth neutral shock y T 0 If credit constraint does not bind we get again PSE If y 0T falls below critical level, credit constraint binds 0 T N N 0 κ 0 T y b p y yˆ = 1 + κ κ has an upper bound, above which b.c. never binds for positive income, and a lower bound, below which c 0T 0 N Tax shocks ( τ p0 ) increase critical level of y 0T, so devaluation shocks can trigger credit constraints for unchanged y T 0
Fisherian deflation equilibrium (FDE) FDE yields an equilibrium with crash or Sudden Stop Date-0 consumption: c y g κ y p y Rb c T T T T N N T 0 = 0 + 0 + 0 + 0 < Date-0 price of nontradables: y Date-0 current account: p T N c0 1 N 0 =Φ ( 1 + τ0) < p N N 0 g b b κ y p y b b T N N 1 0 = 0 + 0 0 > Ω 0
A Graphic Illustration: Perfectly Smooth Equilibrium
Equilibrium with Fisherian deflation
Main features of FDE Asymmetric response to positive v. negative shocks FDE amplifies response to negative shocks beyond pure balance sheet effect Devaluation or price shocks neutral for PSE, not FDE Nonexistence and multiplicity: If shock is too large, economy cannot borrow (no FDE exists) and moves to autarky solution Unique equilibrium guaranteed if SS is steeper than PP around PSE (point A) and Ω = PSE debt Endogenous volatility of pn drives volatility of rer and g y p y causes Sudden Stops
Quantitative experiments Functional forms: 1 ( ) uc () = c σ /(1 σ) T μ N c = ac ( ) + (1 a)( c ) μ 1/ μ Parameter values: β = 0.96, σ = 2 a = 0.342 Mendoza s (02) estimate for Mexico μ = 0.204, 1/(1+ μ) = 0.83 at upper bound of range of existing estimates for Latin America yt/(pnyn) = 1.543, ct/yt = 0.66, cn/yn = 0.71 from Mendoza s (02) estimates for Mexico Initial debt set at 1/3 of GDP Permanent output normalized to 1 (results as shares)
Credit shock: Change κ with 3% income drop
Income Shock: Change y 0 T with κ=0.34
and it gets worse (output and employment effects) Firms that produce NT goods experience a collapse of the relative price of their output and sales Value of marginal product (i.e. demand) for inputs they use (capital, labor, int. goods) falls, and so does output Falling supply of NT goods has two effects: Weakens deflation by reducing supply Worsens the credit crunch as now both prices and output fall Vulnerable agents that were o.k. before are hit by the crisis as they become unemployed and/or hit credit constraint because of falling incomes and prices
How bad can it get in the U.S.?: The Great Depression
Deflation during the U.S. Great Depression
U.S. CPI inflation now and then 20.000 15.000 10.000 5.000 0.000 2.500 2.000 1.500 1.000 0.500 0.000 0.500 1.000 1920 1922 1924 1926 1928 1930 1932 1934 1936 1938 1940 1942 1944 1946 1948 1950 1952 1954 1956 1958 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 5.000 10.000 15.000 Oct./Oct. Dec/Dec Average Oct./Sept. 1.500
Conclusions 1. In economies with highly leveraged agents and credit frictions, endogenous deflation of relative prices induces large economic fluctuations 2. Credit frictions induce amplification and asymmetry in responses of c, ca,, pn,, and rer to standard shocks 3. The contribution of the Fisherian deflation is substantial, and its welfare implications terrible!