Alternative theories of the business cycle Lecture 14, ECON 4310 Tord Krogh October 19, 2012 Tord Krogh () ECON 4310 October 19, 2012 1 / 44
So far So far: Only looked at one business cycle model (the RBC model) Tord Krogh () ECON 4310 October 19, 2012 2 / 44
So far II Our linearized RBC model reads: ĉ t = E tĉ t+1 βr σ Etˆr t+1 σĉ t = ŵ t ŷ t = c c ĉt + (1 y y )î t ˆk t+1 = (1 δ)ˆk t + δî t ŷ t = z t + αˆk t + (1 α)ˆn t ŵ t = z t + α(ˆk t ˆn t) ˆr t = r + δ ( ) r z t (1 α)(ˆk t ˆn t) z t = ρz t 1 + ε t Tord Krogh () ECON 4310 October 19, 2012 3 / 44
So far III Main message: Can explain (some of) the business cycle pattern we observe in the data using a model driven purely by technology shocks! Tord Krogh () ECON 4310 October 19, 2012 4 / 44
Monetary policy in an RBC model But we hear a lot about monetary policy and inflation. Why not a part of our model? Let us introduce that. ρ t+1 : The nominal interest rate π t: Inflation Tord Krogh () ECON 4310 October 19, 2012 5 / 44
Monetary policy in an RBC model II What interest rate is the nominal interest rate? Assume that it is the interest rate on a risk-free bond. To maintain equilibrium in the market for savings, we need to assume that E tˆr t+1 = ρ t+1 E tπ t+1 i.e. we need the expected return from investing in capital to equal the expected real return from investing in the risk-free bond. Tord Krogh () ECON 4310 October 19, 2012 6 / 44
Monetary policy in an RBC model III How to model monetary policy? The normal way to add it is to assume that the interest rate follows a Taylor rule. Linearized around steady state: ρ t+1 = φ ππ t + φ y y t + η t We assume φ π > 1 to obey the Taylor principle (Taylor, 1993). η t is a monetary policy shock. Tord Krogh () ECON 4310 October 19, 2012 7 / 44
Monetary policy in an RBC model IV The RBC model appended with monetary policy is then: ĉ t = E tĉ t+1 βr σ Etˆr t+1 (1) σĉ t = ŵ t (2) ŷ t = c c ĉt + (1 )î y y t (3) ˆk t+1 = (1 δ)ˆk t + δî t (4) ŷ t = z t + αˆk t + (1 α)ˆn t (5) ŵ t = z t + α(ˆk t ˆn t) (6) ˆr t = r + δ ( ) r z t (1 α)(ˆk t ˆn t) (7) z t = ρz t 1 + ε t (8) E tˆr t+1 = ρ t+1 E tπ t+1 (9) ρ t+1 = φ ππ t + φ y y t + η t (10) Tord Krogh () ECON 4310 October 19, 2012 8 / 44
Monetary policy in an RBC model V We can now analyze how monetary policy shocks affect the economy. Can solve the model and plot impulse-response functions. OR: We can think for a moment, because it is easy to see what effects it will have.. Tord Krogh () ECON 4310 October 19, 2012 9 / 44
Monetary policy in an RBC model VI None! Reason? The classical dichotomy holds. Tord Krogh () ECON 4310 October 19, 2012 10 / 44
The Classical Dichotomy This is a concept due to Patinkin (1956). When the classical dichotomy holds you will have: The real economy determining real variables The monetary system determining nominal variables Tord Krogh () ECON 4310 October 19, 2012 11 / 44
The Classical Dichotomy II How to understand it in our RBC model? We see that, although we introduce a new interest rate + inflation, these variables do not enter in any of the old equations. Only linkage between a real and a nominal variable is in the new equations we have added. Effect? Monetary policy controls inflation, but nothing more. The equilibrium inflation rate will always adjust to make the real return on bonds equal to the real return on capital. Other than that, inflation plays no role in the model. Tord Krogh () ECON 4310 October 19, 2012 12 / 44
Rest of today s lecture Will give you a more general overview of business cycle models 1 A (very!) brief history of macroeonomics 2 The traditional (American) Keynesian model 3 New Keynesian approach 4 Some comments on what else we may use Tord Krogh () ECON 4310 October 19, 2012 13 / 44
Brief history Timeline Keynes (1936) is often credited for giving birth to macroeconomics as a distinct field of research (Frisch is said to have been the first to use the term macroeonomics ) Hicks (1937) presented the IS-LM interpretation of Keynes theory It is interesting to note that Hicks (1980) admits he later became dissatsified with the IS-LM representation of Keynes. Especially the lack of references to expectations (animal spirits). Post-war period: IS-LM type macro dominated textbooks. Large-scale econometric models were developed in an attempt to model the economy. Neoclassical synthesis. Tord Krogh () ECON 4310 October 19, 2012 14 / 44
Brief history Timeline II IS-LM and AS-AD models still dominate the syllabus usually taught in introductory and intermediate macro courses. Tord Krogh () ECON 4310 October 19, 2012 15 / 44
Brief history Timeline III 1970s and 80s: Rational expectations revolution and introduction of neoclassical RBC models 1980s and 90s: Attempts to formulate New Keynesian models (a microeconomic fundament for Keynesian ideas) 2000s: New Neoclassical Synthesis (Goodfriend and King, 1997): The New Keynesian framework ended up being RBC models appended with various frictions. Common label: DSGE models. Tord Krogh () ECON 4310 October 19, 2012 16 / 44
Brief history Timeline IV The majority of top journal research in macroeconomics today is done using DSGE-type models. Tord Krogh () ECON 4310 October 19, 2012 17 / 44
Trad. Keynes Traditional Keynesian models The IS-LM and AS-AD model are known to you from earlier courses. Let us have a quick review. Consider a model for the closed economy where we have: Output Y Consumption C Investment I Nominal interest rate i Expected inflation π e Real interest rate r = i π e Government spending G Taxes T Money M Price level P Real money M/P Tord Krogh () ECON 4310 October 19, 2012 18 / 44
Trad. Keynes Traditional Keynesian models II Aggregate variables are assumed to be determined as: C = C(Y T ) I = I (r) while output is demand-determined from the national-accounts relationship: Y = C + I + G Collect all equations together: Y = C(Y T ) + I (r) + G Tord Krogh () ECON 4310 October 19, 2012 19 / 44
Trad. Keynes Traditional Keynesian models III Gives us the IS curve: Tord Krogh () ECON 4310 October 19, 2012 20 / 44
Trad. Keynes Traditional Keynesian models IV Then we assume that money demand is M P = m(y, r) Monetary policy can involve either setting M or setting i. Assume that they choose the latter, and that they follow the rule i = ρ(y, π) For this interest rate, money supply will adjust to whatever is being demanded. Tord Krogh () ECON 4310 October 19, 2012 21 / 44
Trad. Keynes Traditional Keynesian models V Drawing the IS curve toghether with the policy rule, we get the IS-MP diagram: Tord Krogh () ECON 4310 October 19, 2012 22 / 44
Trad. Keynes Traditional Keynesian models VI The AS-AD model is completed by adding an assumption for how inflation is determined (the AS curve): π = π(y ) Combine this function, together with the AD-curve, which is based on the IS and MP curves together: Y = C(Y T ) + I (ρ(y, π) π e ) + G Tord Krogh () ECON 4310 October 19, 2012 23 / 44
Trad. Keynes Traditional Keynesian models V Tord Krogh () ECON 4310 October 19, 2012 24 / 44
Trad. Keynes Traditional Keynesian models VI The AS-AD model can be used to analyze how output, consumpion, investment, the interest rate and inflation responds to changes in the exogenous variables (inflation expectations, government expenditure). Can also add exogenous shocks to the model, simulate it, and compare the implied business cycle statistics with the data (just like we did for the RBC model). Note also that there is indeed a literature called post-keynesian macroeconomics. Tord Krogh () ECON 4310 October 19, 2012 25 / 44
Trad. Keynes Main critisism Given what we have learned in the course so far, there are at least 3 things we would be critical of in the AS-AD model: Not properly microfounded. Example: C = C(Y T ). Expectations are treated as an exogenous variable Few references to the supply-side, and no integration within a growth framework Tord Krogh () ECON 4310 October 19, 2012 26 / 44
Trad. Keynes Main critisism II It is along these dimensions that RBC models improve on the AS-AD model. RBC models are: microfounded. Example: C is the solution to a representative agent s optimization problem using rational expectations. building on neoclassical growth model, nesting business cycles and growth in the same model. Tord Krogh () ECON 4310 October 19, 2012 27 / 44
Trad. Keynes Main critisism III At the same time, RBC models are distinctly different from Keynesian models with respect to what is driving business cycles, and what is the optimal stabilization policy. Business cycles in RBC models reflect only the economy s optimal response to exogenous shocks. Recessions in a Keynesian model can be driven by insufficient demand, and makes way to expansionary fiscal policy to stabilize the economy Prices in RBC models are completely flexible, making monetary policy irrelevant Keynesian models treat the price level as a slow-moving variable, allowing monetary policy to have an effect Tord Krogh () ECON 4310 October 19, 2012 28 / 44
New Keynesian approach New Keynesian approach This brings us to the New Keynesian approach. Because what should you do if you Agree with the critisism of Keynesian models But believe that Keynesian effects can be important? New Keynesian economics can be seen as an attempt to take the RBC-critisism seriously. Broadly speaking it was a project to find microeconomic fundament for the Keynesian features that are most important. Tord Krogh () ECON 4310 October 19, 2012 29 / 44
New Keynesian approach New Keynesian approach II Result from this project is, according to its proponents, that they have managed to put Keynesian business cycle analysis into a modern framework. (Still some Keynesian that think they have failed more on that later today.) Tord Krogh () ECON 4310 October 19, 2012 30 / 44
New Keynesian approach New Keynesian approach III How did they do it? To understand that, we need to re-visit the classical dichotomy. As Mankiw (1989) puts it: The macroeconomist must either destroy the classical dichotomy or learn to live with it. Tord Krogh () ECON 4310 October 19, 2012 31 / 44
New Keynesian approach New Keynesian approach IV How to destroy the classical dichotomy? In the RBC model we saw that adding a monetary sector only meant that the nominal interest rate determined inflation, but inflation had no other effects. It is therefore necessary to make (some) real variables depend on the inflation to destroy the classical dichotomy One way to do so is by making prices less flexible ( sticky ), in constrast to RBC models where prices are fully flexible Tord Krogh () ECON 4310 October 19, 2012 32 / 44
New Keynesian approach New Keynesian approach V But this begs the question: How can we make prices sticky? The path New Keynesian theory has ended up taking involves 1 Start out with a standard RBC model 2 Modify it to allow for price setters (assuming monopolistic competition) 3 Add frictions to the price-setting decision, such that all prices aren t adjusted every period Will give the inflation rate real effects Tord Krogh () ECON 4310 October 19, 2012 33 / 44
New Keynesian approach Modifying the RBC model to destroy the dichotomy Let us do the necessary changes. Recall, in the standard RBC model we assume that there is a representative firm acting as a price-taking profit maximizer. With Cobb-Douglas production function the firm solves: ( ) max kt α k,n n1 α t r tk t w tn t every period. From this problem we get two equilibrium conditions: ( ) α 1 kt r t = α n t ( ) α kt w t = (1 α) n t Tord Krogh () ECON 4310 October 19, 2012 34 / 44
New Keynesian approach Modifying the RBC model to destroy the dichotomy II How can we introduce monopolistic competition instead? It turns out that the most elegant way is to assume: That there is a continuum of firms all producing a unique good, giving them monopoly power (setting prices as a constant mark-up over the marginal cost) Given their prices, they choose labor and capital to minimize costs All the goods are then aggregated into one final good which is what is spent on consumption and investment This alone does not have terribly important effects if the price level remains flexible. The new model would replace the factor-price equations with mc t = α α (1 α) 1 α r α t w 1 α t mc t = µ k t n t = αwt (1 α)r t The first is just the definition of (real) marginal costs of a cost-minimizing firm with Cobb-Douglas production. The second is the pricing rule: All firms are setting prices as a constant mark-up over nominal marginal costs. Requires the real marginal cost to stay constant. Last is the cost-minimization condition. Tord Krogh () ECON 4310 October 19, 2012 35 / 44
New Keynesian approach Modifying the RBC model to destroy the dichotomy III The crucial step is when we introduce price stickiness. The standard way to do that in New Keynesian models is due to Calvo (1983). Every period, each firm will only get the chance to change its price with probability 1 θ. This makes θ a measure of price stickiness The details are messy. But one can show that the optimal pricing rule for a firm will result in the following (linearized) equation for price determination: π t = βe tπ t+1 + (1 θ)(1 βθ) mc t θ The is the New Keynesian Phillips curve (see Clarida, Gali and Gertler, 1999). It replaces the mc t = µ condition for a flex-price economy with monopolistic pricing. Tord Krogh () ECON 4310 October 19, 2012 36 / 44
New Keynesian approach Modifying the RBC model to destroy the dichotomy IV The New Keynesian Phillips curve embodies that Price setting is forward looking Inflation responds more strongly to changes in the marginal cost when prices are very flexible Gali and Gertler (1999) argue that the New Phillips curve has strong empirical support. Rudd and Whelan (2005) and Bårdsen et al (2004) do not agree. Notice the similarity with the old Phillips curve. Tord Krogh () ECON 4310 October 19, 2012 37 / 44
New Keynesian approach Modifying the RBC model to destroy the dichotomy V Why will the classical dichotomy fail in this economy? Consider a change in the nominal interest rate. In a flex-price model this would just change inflation, with no other effects. In a New Keynesian model, this change in inflation can only happen in equilibrium if the real marginal cost is changed! Hence we have introduced a link between a nominal (inflation) and a real (the marginal cost) variable! Tord Krogh () ECON 4310 October 19, 2012 38 / 44
New Keynesian approach Modifying the RBC model to destroy the dichotomy VI OK, let us summarize how to modify an RBC model to get a simple New Keynesian model instead. 1 Start out with RBC model 2 Then replace the equations for r t and w t with one equation defining the real marginal cost, mc t, one equation requiering mc t = µ, and a cost-min. condition. 3 Finally introduce price stickiness by replacing the constant marginal cost assumption with the New Keynesian Phillips curve Tord Krogh () ECON 4310 October 19, 2012 39 / 44
New Keynesian approach New Keynesian business cycles In New Keynesian models, business cycles arise because of: Technology shocks (still) But also monetary policy shocks!... and also cost shocks to inflation and demand shocks In many respects, NK models have been very successful in bringing the main elements of the AS-AD model into a microfounded model. Tord Krogh () ECON 4310 October 19, 2012 40 / 44
New Keynesian approach Other extensions Besides trying to destroy the classical dichotomy, there are many other extensions one can make to an RBC model to make it more realistic: Indivisible labor Money Endogenizing productivity (endogenous growth models) Open economy issues see Obstfeld and Rogoff (1996) Financial frictions (financial accelerator) see Bernanke and Gertler (1988, 1989) and Kiyotaki and Moore (1997) Heterogeneous agents Tord Krogh () ECON 4310 October 19, 2012 41 / 44
Other alternatives? Are DSGE models all we need? Hartley, Hoover and Salyer (1997) give an interesting critique of RBC models, which highlights three important questions a macroeconomic modeler must face (equally important for any DSGE model-user): Should models be microfounded, or are aggregate relationships too complex to microfound? Is equilibrium an innocuous assumption, or are disequilibrium-dynamics essential to understand business cycles? What role should statistics and econometrics play in evaluating models? Tord Krogh () ECON 4310 October 19, 2012 42 / 44
Other alternatives? Are DSGE models all we need? II Hartley et al (1997) interpret the RBC methodology as saying: Models should be microfounded (with rational expectations) Equilibrium is an innocuous/realistic assumption Statistics and econometrics will play a limited role: Aesthetic R 2 This is also the case for New Keynesian models, at least the first two points. Tord Krogh () ECON 4310 October 19, 2012 43 / 44
Other alternatives? Are DSGE models all we need? III Maybe there is a future to Genuine macro-based models without a representative agent, maybe based on emergent properties from agent-based models? Disequilibrium models? Multiple-equilibrium models with animal spirits? Tord Krogh () ECON 4310 October 19, 2012 44 / 44