Economics · Model

The Canonical New Keynesian Model

In the canonical RBC model, money plays no role: prices adjust at once and monetary policy cannot move output. The New Keynesian model adds monopolistic competition and prices that adjust only gradually, so that the central bank, by setting the nominal interest rate, moves the real interest rate and with it demand, output and inflation (Clarida, Galí and Gertler, 1999; Woodford, 2003; Galí, 2015). This page sets out its canonical version, without capital, solves it in your browser for the parameters you choose, and uses it to read the US history of output, inflation and the federal funds rate since 1954 as the outcome of four shocks: to technology, to the supply of labour, to demand and to monetary policy.

The Model

Households

The household’s problem

A representative household consumes \(c_t\) and works \(h_t\) hours. Its utility has a constant relative risk aversion \(\sigma\) in consumption, so that \(1/\sigma\) is the intertemporal elasticity of substitution, and a disutility of work with a Frisch elasticity \(1/\eta\); \(\theta_t\) is the weight on the disutility of labour, which can change over time and so acts as a labour-supply shock, and \(\zeta_t\) shifts the discount factor, so that it changes how much the household values the present relative to the future: a demand shock. The household owns the firms and receives their profits as dividends, \(D_t\). Its only financial asset is a one-period nominal bond: a bond bought at the price \(Q_t\) in quarter \(t\) pays one unit of money in \(t+1\), so the gross nominal interest rate is \(R_t = 1/Q_t\). In real terms, with \(b_{t+1}\) the real value of the bonds bought in \(t\) and \(1+\pi_t = P_t/P_{t-1}\) gross inflation:

\[ \max_{\{c_t,\, h_t,\, b_{t+1}\}_{t=0}^{\infty}}\; \mathbb{E}_0 \sum_{t=0}^{\infty} \beta^t \zeta_t \left[ \frac{c_t^{1-\sigma}}{1-\sigma} - \theta_t\, \frac{h_t^{1+\eta}}{1+\eta} \right] \] \[ \text{subject to} \quad c_t + Q_t\, b_{t+1} = w_t h_t + \frac{b_t}{1+\pi_t} + D_t , \] \[ b_0 \ \text{given} . \]

With \(\beta^t \zeta_t \lambda_t\) the multiplier on the budget constraint, the first-order conditions give \(\lambda_t = c_t^{-\sigma}\), labour supply, where the marginal rate of substitution between leisure and consumption equals the real wage, and the Euler equation for bonds:

\[ \theta_t\, h_t^{\eta} = w_t\, c_t^{-\sigma} , \qquad c_t^{-\sigma} = \beta\, \mathbb{E}_t \left[ \frac{\zeta_{t+1}}{\zeta_t}\, c_{t+1}^{-\sigma}\, \frac{R_t}{1+\pi_{t+1}} \right] , \]

where \(R_t/(1+\pi_{t+1})\) is the gross real interest rate (the Fisher relation). Because prices are sticky, inflation does not adjust one for one when the central bank changes \(R_t\): monetary policy moves the real interest rate, and therefore consumption. The two preference shifters follow \(\ln \theta_t = (1-\rho_\theta) \ln \theta + \rho_\theta \ln \theta_{t-1} + \varepsilon^\theta_t\) and \(\ln \zeta_t = \rho_\zeta \ln \zeta_{t-1} + \varepsilon^\zeta_t\). The discount-factor shock is not in the notes this page follows; it is added, as in Galí (2015, Chapter 3), because without it the technology and labour-preference shocks could not be told apart in the data (see Backing out the shocks). It leaves labour supply and the steady state unchanged.

Final-Good Firms

Aggregating the intermediate goods

Competitive firms assemble a unit mass of differentiated intermediate goods, \(y_t(j)\), \(j \in [0,1]\), into the final good with a constant elasticity of substitution \(\varepsilon > 1\) (Dixit and Stiglitz, 1977):

\[ Y_t = \left[ \int_0^1 y_t(j)^{\frac{\varepsilon-1}{\varepsilon}}\, dj \right]^{\frac{\varepsilon}{\varepsilon-1}} . \]

Maximising profit, \(P_t Y_t - \int_0^1 P_t(j)\, y_t(j)\, dj\), gives the demand for each good and, with zero profits, the price of the final good:

\[ y_t(j) = \left( \frac{P_t(j)}{P_t} \right)^{-\varepsilon} Y_t , \qquad P_t = \left[ \int_0^1 P_t(j)^{1-\varepsilon}\, dj \right]^{\frac{1}{1-\varepsilon}} . \]

Each intermediate good has imperfect substitutes, so its producer faces a downward-sloping demand curve and sets its price: this is what makes sticky prices possible.

Intermediate-Good Firms

Costs, and two ways of making prices sticky

Firm \(j\) produces with labour only, with a technology common to all firms:

\[ y_t(j) = A_t\, h_t(j) , \qquad \ln A_t = (1-\rho_A) \ln A + \rho_A \ln A_{t-1} + \varepsilon^A_t . \]

It hires labour in a competitive market, so it first minimises the cost of producing \(y_t(j)\) (Mas-Colell, Whinston and Green, 1995): the multiplier on the technology constraint is the real marginal cost, the same for every firm, and real profit before any cost of changing prices is

\[ \mu_t = \frac{w_t}{A_t} , \qquad \Pi_t(j) = \left( \frac{P_t(j)}{P_t} - \mu_t \right) y_t(j) . \]

Rotemberg (1982) pricing. Every firm can change its price every quarter, at a quadratic cost governed by \(\omega \geq 0\). It chooses its price to maximise the value of its dividends to the household, subject to the demand for its good:

\[ \max_{\{P_t(j)\}}\; \mathbb{E}_0 \sum_{t=0}^{\infty} \beta^t\, \frac{\lambda_t}{\lambda_0} \left[ \left( \frac{P_t(j)}{P_t} - \mu_t \right) y_t(j) - \frac{\omega}{2} \left( \frac{P_t(j)}{P_{t-1}(j)} - 1 \right)^{2} \frac{P_t(j)}{P_t}\, y_t(j) \right] . \]

All firms face the same problem and set the same price, \(P_t(j) = P_t\), which gives a non-linear Phillips curve:

\[ \pi_t (1+\pi_t) + \frac{1-\varepsilon}{2}\, \pi_t^2 = \frac{1-\varepsilon+\varepsilon \mu_t}{\omega} + \beta\, \mathbb{E}_t \left[ \frac{\lambda_{t+1}}{\lambda_t}\, \frac{Y_{t+1}}{Y_t}\, \pi_{t+1} (1+\pi_{t+1}) \right] . \]

Calvo (1983) pricing. Each quarter, a firm can reset its price with probability \(1-\xi\), whatever its history, and otherwise keeps it. A firm that resets chooses the price \(P^*_t\) that maximises the expected value of its profits while that price is in force, which happens \(s\) quarters later with probability \(\xi^s\):

\[ P^*_t = \frac{\varepsilon}{\varepsilon-1}\; \frac{\mathbb{E}_t \sum_{s=0}^{\infty} (\xi\beta)^s\, \lambda_{t+s}\, \mu_{t+s}\, P_{t+s}^{\varepsilon}\, Y_{t+s}}{\mathbb{E}_t \sum_{s=0}^{\infty} (\xi\beta)^s\, \lambda_{t+s}\, P_{t+s}^{\varepsilon-1}\, Y_{t+s}} , \qquad P_t^{1-\varepsilon} = (1-\xi)\, P_t^{*\,1-\varepsilon} + \xi\, P_{t-1}^{1-\varepsilon} . \]

The reset price is a markup \(\varepsilon/(\varepsilon-1)\) over a weighted average of the nominal marginal costs the firm expects while its price is in force. Without price rigidity, \(\omega = 0\) or \(\xi = 0\), both give \(P_t(j)/P_t = 1\) and a constant real marginal cost, \(\mu_t = (\varepsilon-1)/\varepsilon\).

Monetary Policy

A Taylor rule

Unlike the RBC model, the model needs a rule for the nominal interest rate: without one, neither nominal nor real variables are determined. The central bank follows a Taylor (1993) rule, written in deviations from the steady state:

\[ \hat{R}_t = \phi_\pi\, \pi_t + \phi_y\, \tilde{y}_t + \nu_t , \qquad \nu_t \sim \text{i.i.d.}\ \mathcal{N}(0, \sigma^2_\nu) , \]

where \(\tilde{y}_t\) is the output gap, the deviation of output from the level it would reach with flexible prices, and \(\nu_t\) is a monetary policy shock. The rule raises the interest rate when inflation or the output gap rises. Because the bond is traded between identical households and is in zero net supply, \(R_t\) is a shadow rate: households adjust their consumption and hours until the rate set by the central bank clears the bond market at \(b_t = 0\).

General Equilibrium

Definition of competitive equilibrium

Given the initial bond holdings, \(b_0\), the initial prices of the intermediate goods, the initial levels of technology and of the weight on the disutility of labour, and the stochastic processes of the four shocks, a competitive equilibrium consists of state-contingent sequences of household choices (consumption, hours and bonds), firm choices (outputs, hours and prices) and prices (the real wage, the price level and the interest rate), in which every variable dated \(t\) depends only on the history of shocks up to \(t\), such that:

  1. Household optimisation. Given prices, households maximise lifetime expected utility subject to their sequence of budget constraints.
  2. Firm optimisation. Final-good firms maximise profit; intermediate-good firms minimise costs and set their prices to maximise the value of their dividends, subject to the demand for their good and to the Rotemberg cost or the Calvo lottery.
  3. Monetary policy. The interest rate follows the Taylor rule.
  4. Market clearing. In every period and for every history of shocks, the bond market clears, \(b_t = 0\), the labour market clears, \(h_t = \int_0^1 h_t(j)\, dj\), and the goods market clears: \(c_t = Y_t\) with Calvo pricing and \(c_t = Y_t - \frac{\omega}{2} \pi_t^2 Y_t\) with Rotemberg pricing, whose price-adjustment costs use up goods.

The log-linear model

Calibration and Solution

The model is solved in your browser, to a first-order approximation around its zero-inflation steady state, for the parameters below. Change any of them and every table and chart on the page is recomputed; the default calibration is that of the notes on which this page is based, with a markup of 10 percent (\(\varepsilon = 11\)) and prices that last about ten months on average (\(\xi = 0.7\)), in line with the evidence on US consumer prices (Nakamura and Steinsson, 2008). The persistences and sizes of the shocks are estimated from US data on Extracting Shocks for the New Keynesian Model: the persistences are those that the shocks backed out from the data reproduce, with the structural parameters of the notes. The two ways of making prices sticky give the same log-linear model when their Phillips curves have the same slope.

Price setting

Rates are quarterly in the model, so \(\phi_y = 0.125\) is Taylor’s coefficient of 0.5 on the output gap with annual rates. The parameter that the selected price setting does not use is greyed out. The reset restores the calibration of the notes for the structural parameters and the estimated calibration for the shocks.

Loading data…

Implied values

Steady state

Without shocks, inflation is zero, every firm charges the same price, a markup \(\varepsilon/(\varepsilon-1)\) over marginal cost, and the real interest rate is \(1/\beta\). The levels depend on steady-state technology, \(A = 1.2\), and on the steady-state weight on the disutility of labour, \(\theta = 1\), which do not affect the log-linear dynamics.

Solution

Output and inflation are linear in the four states, \(z_t = (\hat{a}_t, \hat{\theta}_t, \hat{\zeta}_t, \nu_t)\): \(\hat{y}_t = g_y z_t\) and \(\pi_t = g_\pi z_t\), and the rule gives the interest rate, \(\hat{R}_t = g_R z_t\) (see Solving the model). Each coefficient is the response, in percent or quarterly percentage points, to a 1 percent deviation of the state.

Impulse Responses

One shock of one standard deviation hits the economy in its steady state in quarter 1, with no further shocks: a rise of \(\sigma_A\), \(\sigma_\theta\) or \(\sigma_\zeta\) percent in technology, in the weight on the disutility of labour or in the discount factor, which then decays at the rate \(\rho_A\), \(\rho_\theta\) or \(\rho_\zeta\), or a monetary policy shock that raises the interest rate by \(\sigma_\nu\) percentage points at an annual rate for one quarter, before the rule responds. Set a standard deviation to 1 for a 1 percent shock. Inflation and interest rates are in percentage points at an annual rate (four times the model’s quarterly rates).

Shock
Technology Percent deviation from steady state

Loading chart…

Output Percent deviation from steady state

Loading chart…

Output gap Percent of flexible-price output

Loading chart…

Inflation Percentage points, annual rate

Loading chart…

Nominal interest rate Percentage points, annual rate

Loading chart…

Real interest rate Percentage points, annual rate

Loading chart…

Hours worked Percent deviation from steady state

Loading chart…

Real wage Percent deviation from steady state

Loading chart…

Real marginal cost Percent deviation from steady state

Loading chart…

With the default calibration, a rise in technology raises output, but by less than it raises flexible-price output: with sticky prices, prices fall too slowly and the rule lowers the interest rate too little, so the output gap and inflation fall and hours decline (Galí, 2015, Chapter 3). A higher weight on the disutility of labour is a negative supply shock: output falls, by less than flexible-price output, so the gap is positive and inflation rises. A rise in the discount factor raises demand without changing flexible-price output: output, the gap, inflation and the interest rate all rise, the textbook demand shock. A monetary tightening raises the nominal and the real interest rate, and lowers output, hours, real marginal cost and inflation; with no persistence in the shock, everything returns to the steady state the next quarter.

The Model and the US Data

The model has four shocks, and four US series pin them down exactly each quarter: output, labour productivity, inflation and the interest rate. In the model, labour productivity is technology, \(Y_t/h_t = A_t\), so technology is measured directly; given technology, output, inflation and the interest rate determine the other three shocks. The model then tells the US history since 1954 as the sum of four stories: what technology, labour-supply, demand and monetary policy shocks did to the economy (see Backing out the shocks, and, step by step, with the properties of the shocks and tests of the model’s assumptions, Extracting Shocks for the New Keynesian Model). Output and labour productivity are HP cyclical components. The model has no inflation trend and a constant steady-state interest rate, so inflation and the federal funds rate are either taken in deviation from their averages since 1954, which keeps the slow swings of the Great Inflation and of the decline in interest rates since the 1980s, or HP filtered like output, which removes them.

Inflation and interest rate
Download data

The decomposition uses the parameters chosen above; the CSV file records them.

Historical decomposition

Each chart shows a US series, in orange, and the contribution of each shock to it; the four contributions add up to the data.

Output

Loading chart…

Inflation

Loading chart…

Nominal interest rate

Loading chart…

How much each shock explains

For each variable, the correlation of each shock’s contribution with the data, and the variance of that contribution relative to the variance of the data.

The shocks backed out from the data

The states that reproduce the data exactly, and the output gap they imply. Nothing forces them to behave as the model assumes: comparing their persistence with \(\rho_A\), \(\rho_\theta\) and \(\rho_\zeta\), and their size with plausible movements in preferences, is a test of the model.

Technology

Loading chart…

Weight on the disutility of labour

Loading chart…

Discount factor (demand)

Loading chart…

Monetary policy shock

Loading chart…

Output gap

Loading chart…

Two episodes need care. From 2009 to 2015 and in 2020–21 the federal funds rate was at its lower bound, which the linear rule ignores: the monetary shocks backed out for those quarters measure how far the rule would have taken the rate below zero. And the model has no capital, no wage rigidity and no other shocks, so whatever the four shocks cannot capture, such as oil prices, fiscal policy or financial crises, is attributed to them.

Definitions, Methods and Solutions

The log-linear model

Around the zero-inflation steady state, with hats for log deviations, \(\pi_t\) the net inflation rate and \(\hat{R}_t\) the deviation of the nominal interest rate, both quarterly, the goods market, technology, labour supply and marginal cost give

\[ \hat{c}_t = \hat{y}_t , \qquad \hat{h}_t = \hat{y}_t - \hat{a}_t , \qquad \hat{w}_t = \hat{\theta}_t + \eta \hat{h}_t + \sigma \hat{c}_t , \qquad \hat{\mu}_t = \hat{w}_t - \hat{a}_t , \]

(the Rotemberg cost \(\frac{\omega}{2}\pi_t^2 Y_t\) is of second order). Combining them, real marginal cost is \(\hat{\mu}_t = \hat{\theta}_t + (\eta+\sigma)\hat{y}_t - (1+\eta)\hat{a}_t\). The Euler equation gives the dynamic IS curve, and both pricing models give the New Keynesian Phillips curve, with different slopes \(\kappa\):

\[ \hat{y}_t = \mathbb{E}_t \hat{y}_{t+1} - \frac{1}{\sigma} \left( \hat{R}_t - \mathbb{E}_t \pi_{t+1} - (1-\rho_\zeta)\, \hat{\zeta}_t \right) , \qquad \pi_t = \kappa\, \hat{\mu}_t + \beta\, \mathbb{E}_t \pi_{t+1} , \]
\[ \kappa^{\text{Calvo}} = \frac{(1-\xi)(1-\xi\beta)}{\xi} , \qquad \kappa^{\text{Rotemberg}} = \frac{\varepsilon-1}{\omega} . \]

With flexible prices, marginal cost is constant, \(\hat{\mu}^f_t = 0\), which gives flexible-price output and the output gap; marginal cost is proportional to the gap:

\[ \hat{y}^f_t = \frac{1+\eta}{\eta+\sigma}\, \hat{a}_t - \frac{1}{\eta+\sigma}\, \hat{\theta}_t , \qquad \tilde{y}_t = \hat{y}_t - \hat{y}^f_t , \qquad \hat{\mu}_t = (\eta+\sigma)\, \tilde{y}_t . \]

The rule, \(\hat{R}_t = \phi_\pi \pi_t + \phi_y \tilde{y}_t + \nu_t\), and the processes \(\hat{a}_t = \rho_A \hat{a}_{t-1} + \varepsilon^A_t\), \(\hat{\theta}_t = \rho_\theta \hat{\theta}_{t-1} + \varepsilon^\theta_t\) and \(\hat{\zeta}_t = \rho_\zeta \hat{\zeta}_{t-1} + \varepsilon^\zeta_t\) complete the model. In terms of the gap, the IS curve is \(\tilde{y}_t = \mathbb{E}_t \tilde{y}_{t+1} - \frac{1}{\sigma}(\hat{R}_t - \mathbb{E}_t \pi_{t+1} - \hat{r}^f_t)\), with \(\hat{r}^f_t = \sigma\, \mathbb{E}_t[\hat{y}^f_{t+1} - \hat{y}^f_t] + (1-\rho_\zeta)\, \hat{\zeta}_t\) the natural real interest rate.

Solving the model

The model is solved by undetermined coefficients (Uhlig, 1999). Guess \(\hat{y}_t = g_y z_t\) and \(\pi_t = g_\pi z_t\). For a state \(z^k_t\) with persistence \(\rho_k\) (\(\rho_\nu = 0\)), \(\mathbb{E}_t z^k_{t+1} = \rho_k z^k_t\), and substituting the guess and the rule into the IS and Phillips curves gives two linear equations in its two coefficients:

\[ \begin{bmatrix} \sigma(\rho_k - 1) - \phi_y & \rho_k - \phi_\pi \\ -\kappa(\eta+\sigma) & 1 - \beta\rho_k \end{bmatrix} \begin{bmatrix} g^k_y \\ g^k_\pi \end{bmatrix} = \begin{bmatrix} n_k - \phi_y f_k - i_k \\ -c_k \end{bmatrix} , \]

where \(f_k\) is the coefficient of the state in flexible-price output (\((1+\eta)/(\eta+\sigma)\) for technology, \(-1/(\eta+\sigma)\) for the labour weight, 0 for the discount factor and the policy shock), \(n_k\) its coefficient in the rule (1 for the policy shock, 0 otherwise), \(i_k\) its coefficient in the IS curve (\(1-\rho_\zeta\) for the discount factor, 0 otherwise) and \(c_k\) its direct effect on inflation through marginal cost (\(\kappa(1+\eta)\) for technology, \(-\kappa\) for the labour weight, 0 otherwise). The solution is unique and stable when the rule satisfies the condition of Bullard and Mitra (2002), \(\kappa(\eta+\sigma)(\phi_\pi - 1) + (1-\beta)\phi_y > 0\): since \(1-\beta\) is small, the interest rate must rise by more than inflation, so that the real interest rate rises when inflation does (the Taylor principle). For the technology, labour-preference and policy shocks, the coefficients coincide, to twelve decimal places, with those of the iterative solver in the MATLAB code accompanying the notes.

Backing out the shocks

Technology and the weight on the disutility of labour enter the IS curve, the Phillips curve and the rule only through \((1+\eta)\hat{a}_t - \hat{\theta}_t\), in marginal cost and in flexible-price output. With the same persistence, they therefore move output, inflation and the interest rate in exactly the same proportions, and these three series cannot tell them apart; with different persistences, the split would rest on that difference alone, and would be very imprecise. They differ in their effect on hours, \(\hat{h}_t = \hat{y}_t - \hat{a}_t\), so labour productivity, \(\hat{y}_t - \hat{h}_t = \hat{a}_t\), measures technology. The demand shock then adds a fourth source of movements in output, inflation and the interest rate, and the model is exactly identified.

Stacking the observables \(x_t = (\hat{y}_t, \hat{y}_t - \hat{h}_t, \pi_t, \hat{R}_t)'\), the solution is \(x_t = M z_t\), where the columns of the \(4 \times 4\) matrix \(M\) are the coefficients \((g_y, 1_{k=A}, g_\pi, g_R)'\) of each state. When \(M\) is invertible, the states are \(z_t = M^{-1} x_t\), quarter by quarter, and the contribution of shock \(k\) to variable \(x\) is \(g^k_x z^k_t\); the four contributions add up to the data exactly. The innovations follow as \(\varepsilon^k_t = z^k_t - \rho_k z^k_{t-1}\). This is the exactly identified case of the historical decompositions of estimated models, which use the Kalman filter when there are fewer observables than shocks (Smets and Wouters, 2007); here no shock is left over to absorb measurement error, so every movement in the data is attributed to one of the four shocks. The standard deviations of the shocks are not needed for the decomposition; they only set the size of the shocks in the impulse responses.

The data are converted to model units: output and labour productivity in percent of trend divided by 100; inflation and the federal funds rate in percent at an annual rate, demeaned or HP filtered, divided by 400 to give quarterly rates.

Data
  • Output: real GDP (GDPC1, U.S. Bureau of Economic Analysis) divided by the population (B230RC0Q173SBEA), HP cyclical component (λ = 1600) of \(100 \times \ln\), as on the Business Cycles page.
  • Inflation: annualised quarterly change in the GDP price index (GDPDEF), \(400 \times \Delta \ln P_t\), the price of the goods produced in the economy, the counterpart of the model’s final good.
  • Interest rate: the effective federal funds rate (FEDFUNDS, Board of Governors of the Federal Reserve System), averaged over each quarter, the US counterpart of the one-period nominal rate set by the central bank.
  • Labour productivity: output per hour in the nonfarm business sector (OPHNFB, U.S. Bureau of Labor Statistics), HP cyclical component (λ = 1600) of \(100 \times \ln\), the counterpart of the model’s technology, \(A_t = Y_t/h_t\). It covers the nonfarm business sector, about three quarters of GDP, and is the standard measure of labour productivity.

The sample starts in 1954 Q3, the first full quarter of the federal funds rate. Inflation and the federal funds rate are demeaned over the whole sample, or HP filtered (λ = 1600) on their full history.

Why “canonical” New Keynesian model?

The model on this page is the benchmark of the New Keynesian literature: households, monopolistically competitive firms that set prices with a friction (Calvo or Rotemberg), and a central bank following a Taylor rule, with no capital. “Canonical” names it as the standard version from which richer models start, as The Canonical RBC Model does for the real business cycle model (King and Rebelo, 1999). Galí (2015, chapter 3) calls it “the basic New Keynesian model”; Clarida, Galí and Gertler (1999) and Woodford (2003) set out the same framework.

“Classical” would be misleading. In this literature the classical monetary model is the flexible-price economy (Galí, 2015, chapter 2), in which money is neutral and monetary policy has no effect on output or employment: the opposite of the nominal rigidities that define a New Keynesian model.

References
  • Bullard, J. and K. Mitra (2002). “Learning about Monetary Policy Rules.” Journal of Monetary Economics 49(6): 1105–1129.
  • Calvo, G. A. (1983). “Staggered Prices in a Utility-Maximizing Framework.” Journal of Monetary Economics 12(3): 383–398.
  • Clarida, R., J. Galí and M. Gertler (1999). “The Science of Monetary Policy: A New Keynesian Perspective.” Journal of Economic Literature 37(4): 1661–1707.
  • Dixit, A. K. and J. E. Stiglitz (1977). “Monopolistic Competition and Optimum Product Diversity.” American Economic Review 67(3): 297–308.
  • Galí, J. (2015). Monetary Policy, Inflation, and the Business Cycle: An Introduction to the New Keynesian Framework and Its Applications, 2nd edition. Princeton University Press.
  • Hodrick, R. J. and E. C. Prescott (1997). “Postwar U.S. Business Cycles: An Empirical Investigation.” Journal of Money, Credit and Banking 29(1): 1–16.
  • King, R. G. and S. T. Rebelo (1999). “Resuscitating Real Business Cycles.” In J. B. Taylor and M. Woodford (eds.), Handbook of Macroeconomics, Volume 1B, Chapter 14: 927–1007. Elsevier.
  • Mas-Colell, A., M. D. Whinston and J. R. Green (1995). Microeconomic Theory. Oxford University Press.
  • Nakamura, E. and J. Steinsson (2008). “Five Facts about Prices: A Reevaluation of Menu Cost Models.” Quarterly Journal of Economics 123(4): 1415–1464.
  • Rotemberg, J. J. (1982). “Sticky Prices in the United States.” Journal of Political Economy 90(6): 1187–1211.
  • Smets, F. and R. Wouters (2007). “Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach.” American Economic Review 97(3): 586–606.
  • Taylor, J. B. (1993). “Discretion versus Policy Rules in Practice.” Carnegie-Rochester Conference Series on Public Policy 39: 195–214.
  • Uhlig, H. (1999). “A Toolkit for Analysing Nonlinear Dynamic Stochastic Models Easily.” In R. Marimon and A. Scott (eds.), Computational Methods for the Study of Dynamic Economies, Chapter 3: 30–61. Oxford University Press.
  • Woodford, M. (2003). Interest and Prices: Foundations of a Theory of Monetary Policy. Princeton University Press.

Page created on 16/11/2026.

© Arthur Galichère.

  • Errors: Every effort has been made to ensure the model and derivations are accurate; any remaining errors are mine. If you believe you have found a mistake, or have a comment or suggestion, please let me know.
  • Reuse and citation: The data are from the U.S. Bureau of Economic Analysis, the U.S. Bureau of Labor Statistics and the Board of Governors of the Federal Reserve System, via FRED. The model solution, simulations, charts and text are the author’s. When reusing these figures, please cite the original data providers and this website: Arthur Galichère, The Canonical New Keynesian Model.