Advanced Macroeconomics Chapter 19 EXPLAINING BUSINESS CYCLES ©The McGraw-Hill Companies, 2005 Themes of the chapter • Explaining business cycles by means of the AS-AD model. The Frisch-Slutzky paradigm: Impulse and propagation. The deterministic versus the stochastic AS-AD model. Supply versus demand shocks. Permanent versus temporary shocks. The theory of real business cycles. ©The McGraw-Hill Companies, 2005 Percent US GDP US Inflation 5 4 3 2 1 0 -1 -2 -3 -4 -5 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 The cyclical components of real GDP and inflation in the United States ©The McGraw-Hill Companies, 2005 percent 4 USA 3 2 Germany 1 0 -1 Japan -2 1996 1997 1998 1999 2000 2001 2002 2003 Growth in real per capita GDP in the USA, Germany and Japan ©The McGraw-Hill Companies, 2005 The Frisch-Slutzky Paradigm The impulses (demand and supply shocks) initiating business cycles may be unsystematic The propagation of the impulses may generate systematic fluctuations due to the structure of the economy Basic questions Why do movements in economic activity display persistence? Why do these movements tend to follow a cyclical pattern? ©The McGraw-Hill Companies, 2005 Restating the AS-AD model yt y 1 gt g 2 rt r vt rt it e t 1 (2) it r te1 h t * b yt y t yt y st e t t 1 e t (1) (3) (4) (5) ©The McGraw-Hill Companies, 2005 The AS-AD Model in compact form Inserting (2), (3) and (5) into (1), we get yt y * t zt , vt 1 gt g zt 1 2b 2h , 1 2b (6) which may be rearranged to give the aggregate demand curve: AD: 1 t yt y zt * (7) Substituting (5) into (4), we obtain the short-run aggregate supply curve: SRAS: t t 1 yt y st (8) ©The McGraw-Hill Companies, 2005 LRAS SRAS 0 E0 AD y0 y y A short-run macroeconomic equilibrium with cyclical unemployment ©The McGraw-Hill Companies, 2005 LRAS SRAS0 SRAS1 (e = 0) SRAS2 (e = 1) SRAS3 (e = ) 0 1 2 3 * AD y0 y1 y2 y3 y y The adjustment to long-run equilibrium ©The McGraw-Hill Companies, 2005 How long is the Long Run? Defining yˆt yt y (output gap) ˆt t * (inflation gap) and assuming st = zt = 0, we may rewrite (7) and (8) as 1 AD: ˆt 1 yˆt 1 , 2h 1 2b SRAS: ˆt 1 ˆt yˆt 1 (9) (10) Inserting (9) into (10), we find ©The McGraw-Hill Companies, 2005 How long is the Long Run? 1 1 yˆt 1 yˆt , (11) ˆt 1 ˆt (12) The solutions to these linear first-order difference equations are yˆt yˆ 0 , t 0,1, 2,..... ˆt ˆ0 , t 0,1, 2,..... t t (13) (14) Note that the long-run equilibrium is stable, since 0 < β < 1. ©The McGraw-Hill Companies, 2005 The speed of convergence th number of periods before half the adjustment to long-run equilibrium has been completed 1 1 1 th h yˆt yˆ 0 yˆ 0 t ln ln 2 2 2 th ln 2 0.693 th ln ln (15) According to equation (12) in Chapter 17 we have ©The McGraw-Hill Companies, 2005 Calibrating the model (one time period = one quarter) 1 1 h 2 1 b 2 1 Dr 2 Y0 (1 Dy ) 1 Dy 0.05 0.2 Dy 0.8 Dr Y0 (1 ) 3.6 From this it follows that 0.958 th h b 0.5 ln 2 16 4 years ln ©The McGraw-Hill Companies, 2005 LRAS SRAS1 s1 SRAS2 E1 SRAS0 E2 E AD y1 y2 y y Effects of a temporary negative supply shock ©The McGraw-Hill Companies, 2005 LRAS SRAS0 E SRAS2 E2 E1 AD0 = AD2 AD1 y1 y y2 y Effects of a temporary negative demand shock ©The McGraw-Hill Companies, 2005 y-y Percent π - π* 1 0,8 0,6 0,4 0,2 0 -0,2 -0,4 -0,6 -0,8 -1 1 3 5 7 9 11 13 15 17 Year 19 21 23 25 27 29 The adjustment to a temporary negative supply shock ©The McGraw-Hill Companies, 2005 (s1=1) y-y Percent π - π* 0,2 0 -0,2 -0,4 -0,6 -0,8 -1 1 3 5 7 9 11 13 15 17 Year 19 21 23 25 27 29 The adjustment to a temporary negative demand shock (z1= -1) ©The McGraw-Hill Companies, 2005 Permanent Shocks When analyzing permanent shocks, we must account for the fact that such shocks will change the long-run equilibrium real interest rate (the ’natural’ interest rate). Denoting the initial values of natural output and the natural interest rate by zero subscripts, we may write our AS-AD model as yt y0 vt 2 rt r0 , vt vt 1 gt g t t 1 yt y0 st (21) (22) Consider an initial equilibrium wherevt st 0 and suppose that st permanently changes from zero to some s ≠ 0. The new long-run equilibriu level of output may then be found from (22) by inserting t t 1 , to get yt y , st s ©The McGraw-Hill Companies, 2005 A permanent supply shock The effect of a permanent supply shock on natural output s y y0 (23) The new equilibrium real interest rate is found from (21) by setting s yt y y0 , rt r to get The effect of a permanent supply shock on the equilibrium real interest rate s r r0 2 (24) ©The McGraw-Hill Companies, 2005 A permanent demand shock A permanent demand shock does not affect natural output. Hence the effect on the equilibrium real interest rate may be found from (21) by setting yt y0 , vt v and rt r to get The effect of a permanent demand shock on the equilibrium real interest rate v r r0 2 (25) To keep inflation close to its target rate and to avoid large deviations of output from trend, the central bank must revise the estimates of natural output and of the equilibrium real interest rate entering the Taylor rule when the economy is hit by permanent shocks. The adjustment of the economy to the new long-run equilibrium will depend on how long it takes the central bank to realize the permanency of the shock. Exercise 19.2 invites you to study these issues further. ©The McGraw-Hill Companies, 2005 The Stochastic AS-AD model The deterministic AS-AD model considered above can explain the persistence in macroeconomic time series, but it cannot explain the recurrent cyclical fluctuations observed in the real world. To generate such fluctuations, we now assume: Stochastic demand and supply shocks zt 1 zt xt 1 , xt N (0, ) , 2 x st 1 st ct 1 , ct N (0, c2 ) , 0 1 (27) xt i.i.d . 0 1 (28) ct i.i.d . Our goal is to calibrate a stochastic AS-AD simulation model which can reproduce the stylized business cycle facts summarized in Table 19.1. ©The McGraw-Hill Companies, 2005 Percent 5 US GDP US Inflation 4 3 2 1 0 -1 -2 -3 -4 -5 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 Year Cyclical components of real GDP and inflation in the USA, 1974-98 ©The McGraw-Hill Companies, 2005 Table 19.1: The stochastic AS-AD model and the stylized business cycle facts Autocorrelation in output Standard deviation (%) Correlation between Autocorrelation in inflation output and inflation t-1 t-2 t-3 t-4 t-1 t-2 t-3 t-4 0,08 0,81 0,66 0,47 0,37 0,99 0,96 0,91 0,85 1,90 -1,00 0,92 0,86 0,79 0,73 0,92 0,86 0,79 0,73 1,66 0,30 0,15 0,82 0,68 0,50 0,38 0,47 0,33 0,24 0,32 1,66 0,29 0,10 0,86 0,65 0,41 0,18 0,50 0,29 0,24 0,17 Output Inflation 1,62 0,52 1,67 AS-AD model with static expectations and no supply shocks1 AS-AD model with static expectations and no demand shocks 2 AS-AD model with adaptive expectations and a combination of demand and supply shocks3 The U.S economy, 1955:I-2001:IV4 1 4 Φ = 0, σc = 0, σx = 1, δ = 0.75, ω = 0 2 Φ = 0, σc = 0.75, σx = 0, δ = 0, ω = 0 3 Φ = 0.92, σc = 0.25, σx = 1, δ = 0.75, ω = 0.25 The cyclical components of output and inflation have been estimated via detrending of quarterly data using the HP-filter with λ = 1600. Common parameter values in all AS-AD simulations: γ = 0.05, τ = 0.2, DY = 0.8, η = 3.6, h = b = 0.5 ©The McGraw-Hill Companies, 2005 Percent y-y 5 4 3 2 1 0 -1 -2 -3 -4 -5 1 11 21 31 41 51 61 Quarter 71 81 91 Simulation of the stochastic AS-AD model with static expectations and no supply shocks ©The McGraw-Hill Companies, 2005 14 12 e Expected inflation rate for the current quarter (πt ) Actual inflation rate during the previous quarter (πt-1) 10 8 6 4 2 0 -2 -4 1981-III 1984-III 1987-III 1990-III 1993-III 1996-III 1999-III 2002-III Expected current inflation and lagged actual inflation in the United States ©The McGraw-Hill Companies, 2005 The Stochastic AS-AD model with static expectations Problems The model with demand shocks can reproduce the stylized facts regarding output, but it generates far too much persistence of inflation The model with supply shocks is unable to reproduce the stylized facts of output as well as inflation ● The assumption of static expectations implies greater fluctuations in the expected inflation rate than what we observe in practice (see Figure 19.12) To solve these problems we will now allow for simultaneous demand and supply shocks as well as adaptive expectations. ©The McGraw-Hill Companies, 2005 Adaptive expectations revision of expected inflation rate last period's inflation forecast error te te1 (1 ) ( t 1 te1 ) , 0 1 (29) For 0 we get static expectations. Eq. (29) may be rewritten as te n 1 (1 ) t n (33) n 1 The AS-AD model with adaptive expectations AD: yt y * t zt (34) SRAS: t te yt y st (35) Expectations: te te1 1 t 1 (36) ©The McGraw-Hill Companies, 2005 The AS-AD model with adaptive expectations The model (34) through (36) may be condensed to: yˆt 1 ayˆt ( zt 1 zt ) st 1 st (41) ˆt 1 aˆt zt 1 zt st 1 st (42) 1 a 1 1 1 1 1 The third row in Table 19.1 shows that this model reproduces the U.S. business cycle reasonably well, given its simplicity. ©The McGraw-Hill Companies, 2005 Pe r ce nt 5 4 3 2 1 0 -1 -2 -3 -4 y-y 0 97 93 89 85 81 77 73 69 65 61 57 53 49 45 41 37 33 29 25 21 17 13 9 Quar te r s 5 1 -5 Percent 5 US GDP US Inflation 4 3 2 1 0 -1 -2 -3 -4 -5 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 Year The AS-AD model with adaptive expectations (top diagram) versus the actual U.S. business cycle (bottom diagram) ©The McGraw-Hill Companies, 2005 The theory of real business cycles Our AS-AD model of the business cycle emphasizes the role of expectational errors and sluggish wage and price adjustment, and the microfoundation for the SRAS curve implies that business fluctuations are associated with fluctuations in involuntary unemployment. The model also assigns an important role to demand shocks. A very different theory is: Real business cycle theory (basic version) ● The business cycle is mainly driven by fluctuations in the rate of productivity growth ● The employment fluctuations observed during business cycles reflect voluntary movements along individual labour supply curves (intertemporal substitution in labour supply, no involuntary unemployment) ● Economic growth and business cycles can and should be explained within unified model framework. To explain business cycles, there is no need to postulate nominal and/or real rigidities. ©The McGraw-Hill Companies, 2005 A simple RBC model: technology Production function: Yt K At Lt , 0 1 t 1 Actual productivity: lnAt gt st , st 1 st ct 1 , 0 <1, ct N 0 , (44) (45) 2 c Trend productivity: lnAt gt (46) Capital accumulation: Kt 1 Kt 1 St 1 (47) For simplicity, we will assume that δ = 1. ©The McGraw-Hill Companies, 2005 A simple RBC model: economic behaviour Saving: St s Yt , 0 s 1 wt s Labour supply: Lt , 0 wt Kt Yt Profit maximization: wt 1 Lt At Lt Trend real wage: wt 1 cAt , (49) c k Labour market clearing: Lt L s t * (48) (50) (51) (52) ©The McGraw-Hill Companies, 2005 A simple RBC model As shown on pp. 585-86 in the text, the model (44) through (52) may be reduced to 1 ˆyt ˆyt 1 st , 1 1 1 1 1 1 ˆ ˆy L t t Propagation mechanism in the model: A positive productivity shock (59) (60) raises current income which in turn raises saving and capital accumulation. This leads to a higher capital stock in the next period, which in turn raises next period’s income and saving, and so on. In this way a temporary productivity shock generates persistence in output and employment. Indeed, we see from Table 19.3 that the calibrated version of the model generates too much persistence compared to the ©The McGraw-Hill Companies, 2005 persistence observed in the U.S. data. Standard deviation (%) Standard deviation of Autocorrelation Autocorrelation output relative to standard in output in hours worked Output Hours worked deviation of hours worked t-1 t-2 t-3 t-1 t-2 t-3 RBC model1 3.42 2.84 0.83 0.75 0.50 0.23 0.75 0.50 0.23 The U.S economy2 3.47 2.88 0.83 0.76 0.38 0.08 0.73 0.29 0.06 α = 0.33, η = 0.83, ω = 0.1, σc = 0.015 1 2 Annual data for the business sector. Note: The cyclical components of output and employment have been estimated via linear OLS detrending of annual data. Source: Economic Outlook Database, OECD. Table 19.3: The RBC model versus the U.S. economy ©The McGraw-Hill Companies, 2005 Some problems with basic RBC theory • The virtue of the basic RBC model is that it is simple and fully integrates the theory of business cycles with the theory of economic growth. However, critics object to the theory by raising the following questions: • Is technological progress really so uneven as postulated in the RBC model? • Is it really plausible that recessions are periods of technological regress? (Alternative hypothesis: the observed fluctuations in productivity reflect fluctuations in capacity utilization caused by demand shocks). ©The McGraw-Hill Companies, 2005 Some problems with basic RBC theory • Are the observed fluctuations in employment really a reflection of intertemporal substitution in labour supply? To reproduce the observed volatility of employment relative to the volatility of output, the wage elasticity of labour supply in our simple RBC model (ε) has to be set at 4.9 which is much higher than the elasticity estimated by labour economists. More generally: Is all recorded unemployment really voluntary? • The RBC model predicts that the real wage is procyclical. This is in line with U.S. data, but it is not consistent with European data. ©The McGraw-Hill Companies, 2005 The lasting contribution of real business cycle theory In response to these criticisms, real business cycle theorists have recently tried to make their models more realistic by allowing for various frictions and rigidities, including (in some cases) nominal rigidities. At the methodological level RBC theorists have made a lasting contribution by pointing out that supply shocks may play an important part in the explanation of business cycles, and by insisting that a satisfactory theory of the business cycle should consist of a dynamic stochastic general equilibrium model which is able to reproduce the most important stylized facts of the business cycle reasonably well. ©The McGraw-Hill Companies, 2005