The B.E. Journal of Macroeconomics Topics Volume 11, Issue 1

2011

Article 3

The Cost Channel, Indeterminacy, and Price-Level versus Inflation Stabilization Sebastian Schmidt∗

∗

Goethe University Frankfurt, [email protected]

Recommended Citation Sebastian Schmidt (2011) “The Cost Channel, Indeterminacy, and Price-Level versus Inflation Stabilization,” The B.E. Journal of Macroeconomics: Vol. 11: Iss. 1 (Topics), Article 3. Available at: http://www.bepress.com/bejm/vol11/iss1/art3 c Copyright 2011 The Berkeley Electronic Press. All rights reserved.

The Cost Channel, Indeterminacy, and Price-Level versus Inflation Stabilization∗ Sebastian Schmidt

Abstract In a New Keynesian model with a cost channel of monetary transmission, discretionary inflation targeting may become susceptible to a multiplicity of equilibria and results in a larger stabilization bias than in the standard model. Both observations can be attributed to a reduction in the effectiveness of the traditional demand channel of monetary transmission. Based on this insight, we evaluate the desirability of an alternative monetary policy regime, price-level targeting. We find that the latter is almost immune to equilibrium indeterminacy and remains successful in stabilizing output and inflation. Due to the weak demand channel, the ability of the central bank to manage private sector expectations takes center stage. This is accomplished by price-level targeting through the introduction of history dependence, but not by inflation targeting. KEYWORDS: price-level targeting, inflation targeting, New Keynesian model, cost channel, indeterminacy

∗

I am grateful to the editor, two anonymous referees, Elena Afanasyeva, Thomas Laubach, Volker Wieland and Maik Wolters for useful comments.

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

1 Introduction Recent empirical evidence in support of supply-side effects of monetary policy transmission (e.g. Barth and Ramey, 2001; Chowdhury et al, 2006; Tillmann, 2008) has led economists to introduce a cost channel into monetary business cycle models to study implications for optimal monetary policy (e.g. Ravenna and Walsh, 2006; Tillmann, 2009) and for equilibrium determinacy (e.g. Brueckner and Schabert, 2003; Gabriel et al, 2008; Surico, 2008; Aksoy et al, 2009; Llosa and Tuesta, 2009; Christiano et al, 2010). A common finding of the latter is that inflation targeting usually suffers from shrinking regions of determinacy. Due to the cost channel, changes in the nominal interest rate have a direct effect on firms’ variable production costs, thereby weakening the effectiveness of the traditional demand channel of monetary transmission. In modern macroeconomics it is widely recognized that a reasonable monetary policy strategy should in any case prevent the existence of such multiplicity of equilibria.1 A natural question to ask is therefore how the performance of alternatives to inflation targeting is altered by the presence of a cost channel of monetary transmission. In this paper, we investigate the stabilization properties of price-level targeting relative to inflation targeting in a New Keynesian model with a cost channel. Our main results are twofold. First, determinacy regions under price-level targeting remain nearly unaffected by the introduction of the cost channel whereas determinacy regions shrink under discretionary inflation targeting. Second, price-level targeting continues to be potent in stabilizing inflation and the output gap, whereas the stabilization trade-off deteriorates under inflation targeting. In the standard New Keynesian model as analyzed by Clarida et al. (1999), the equilibrium responses to asymmetric disturbances obtained under discretionary optimization differ from those under optimal commitment policy. This so-called stabilization bias arising under discretionary policy can be mitigated (and under special conditions resolved) by delegating a price-level target to the central bank (Vestin, 2006). For reasonable calibrations of this model, however, the gains from price-level targeting relative to inflation targeting are rather small (e.g. Kryvtsov et al, 2008). We find that explicit consideration of the cost channel suggests to read these previous findings with caution, in that the difference in the stabilization performance of the two monetary policy regimes turns out to be larger than in the model without any supply-side effects of monetary policy. Intuitively, with the demand channel of monetary transmission being restrained, the strength of the expectations channel of monetary transmission takes 1 Indeed,

some economists have argued that the failure to do so may have been an important source of the Great Inflation experienced by the US economy in the 1970s, see e.g. Clarida et al. (2000).

Published by The Berkeley Electronic Press, 2011

1

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

center stage. Under price-level targeting, the lagged price level becomes an endogenous state variable. This makes monetary policy history dependent, allowing the central bank to influence private sector expectations. Specifically, price-level stationarity implies that an unexpected surge in contemporaneous inflation fosters the central bank to engineer a below-trend inflation rate in the following periods, so that the price level returns to its target path. As we show, this ability to induce reversals in the price level has desirable impacts on macroeconomic stability and the performance of monetary policy. The remainder of the paper is organized as follows. Section 2 describes a simple microfounded sticky-price model of the business cycle augmented with a cost channel and introduces the general monetary policy framework. Section 3 determines optimal discretionary policy under inflation targeting and price-level targeting, and contrasts their implications for equilibrium determinacy. Section 4 compares the stabilization performance of the two policy regimes. Finally, section 5 concludes.

2 The Model The economy is represented by a prototypical microfounded New Keynesian model of the business cycle augmented with a cost channel. Most of the modeling devices are standard. Households consume a basket of differentiated goods and supply labor to firms in a perfectly competitive market. Due to monopolistic competition, the goods-producing firms face a downward-sloping demand curve and it is assumed that prices are set in a staggered fashion a` la Calvo. We depart from the standard model by assuming that part of the firms have to pay their wage bill prior to the sales receipts which compels them to borrow the required funds from a financial intermediary. As a consequence, a cost channel of monetary transmission arises, that is, changes in the interest rate have a direct effect on marginal costs and therefore on firms’ pricing decisions.

2.1 Structural equations Following Ravenna and Walsh (2006) and Christiano et al. (2010), private sector behavior is summarized by two log-linearized equations

πt = κ˜ xt + β Et πt+1 + δ κ it + ut 1 xt = Et xt+1 − (it − Et πt+1 ) . σ

http://www.bepress.com/bejm/vol11/iss1/art3

(1) (2)

2

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

All variables are expressed in percentage deviations from the non-stochastic steady state. Equation (1) is a generalization of the standard New Keynesian Phillips curve (NKPC). Inflation, πt , depends on expected future inflation, the output gap, xt , and the one-period nominal interest rate, it . The appearance of the latter in the NKPC is specific to the introduction of the cost channel into the model. The output gap represents the gap between the level of output under sticky prices and output under flexible prices.2 As shown by Ravenna and Walsh (2006), the latter is no longer independent of monetary policy once we introduce the cost channel. In order to obtain a well-defined measure of the gap, we adopt the assumption of an interestrate peg in the flexible-price economy from Ravenna and Walsh. The cost-push shock, ut , follows an AR(1) process with a mean-zero, constant-variance innovation εt ut = ρ ut−1 + εt , (3) where ρ ∈ [0, 1). The coefficients are functions of the structural parameters of θβ) the model. In particular, κ˜ = κ (σ + η ) and κ = (1−θ )(1− , where σ > 0 is the θ inverse of the intertemporal elasticity of substitution, η > 0 denotes the inverse of the elasticity of labor supply, β ∈ (0, 1) is the subjective discount factor and θ ∈ (0, 1) denotes the Calvo parameter representing the probability that a firm is not allowed to reoptimize its price in a given period. The cost-channel coefficient δ deserves particular attention. Empirical estimates of the parameter vary quite substantially depending on the estimation method and error bands can be large.3 Some estimates of δ also exceed the value of one (Chowdhury et al, 2006; Ravenna and Walsh, 2006), pointing to the presence of financial frictions. One way to allow for values of δ > 1 in the model is to add financial intermediation costs which increase in the amount of borrowing, see Tillmann (2009). In this case, δ is a function of two objects, the fraction of firms constrained by the cost channel and η the magnitude of the financial friction. For the remainder, we assume δ ∈ [0, σ + σ ), η where the upper bound on δ , σ + σ > 1, ensures that the demand channel of monetary transmission outweighs the supply-side channel. Equation (2) is an intertemporal IS curve, revealing that the output gap depends on its own expected future realization and the ex-ante real interest rate. Finally, for the purpose of this paper it is useful to explicitly define the price level pt ≡ pt−1 + πt .

(4)

2 In

the presence of the cost channel, the steady state level of output differs from the one in the standard model in that it depends on the level of the nominal interest rate in the steady state. Inflation is zero in the steady state. Details of the derivations can be found in Ravenna and Walsh (2006) and Christiano et al. (2010). 3 Furthermore, estimates generally vary over time (Tillmann, 2008) and across countries (Chowdhury et al, 2006).

Published by The Berkeley Electronic Press, 2011

3

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

2.2 Society and policy objectives Throughout, we distinguish between society’s preferences and those of the monetary authority. Society’s preferences are represented by the following objective function ∞  2  1 2 Lt = Et { ∑ β j πt+ λ x + (5) t+ j }, j 2 j=0 where λ > 0 denotes the relative weight on output gap stabilization. For a specific choice of λ , equation (5) represents an accurate linear-quadratic approximation to household welfare, see Ravenna and Walsh (2006).4 In this paper, however, we will consider alternative values for λ , possibly differing from the particular value implied by welfare considerations. Monetary policy is delegated to a central banker, whose objective function is defined in terms of specific target variables ∞ 1 j ′ LCB t = Et { ∑ β yt+ j Qyt+ j }. 2 j=0

(6)

Here yt is the vector of target variables and Q is a diagonal positive semidefinite matrix containing the central banker’s weights on the individual target variables. Under discretion, the central banker minimizes (6) period-by-period subject to the model equations.

2.3 Calibration To illustrate results, we calibrate the model as summarized in Table 1. The period length is one quarter. Most parameter values refer to the calibration in Ravenna and Walsh (2006). In addition, the AR-coefficient of the cost-push shock is set to ρ = 0.5 and the variance of the innovation, εt , is normalized to one. 4 In

principle, the adequate output gap measure in a loss function based on household welfare, the deviation of output from the efficient level of output, differs from the output gap in equations (1) and (2). This follows from the presence of monopolistic competition and from the dependence of flexible-price steady state output on the steady state level of the nominal interest rate. To focus on stabilization policy, we follow Ravenna and Walsh (2006) in assuming that steady state distortions are eliminated by appropriate fiscal subsidies.

http://www.bepress.com/bejm/vol11/iss1/art3

4

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

Table 1: Calibration Parameter β 0.99 σ 1.50 η 1.00 κ 0.0858 ρ 0.50

3 Discretionary monetary policy and equilibrium determinacy This section characterizes monetary policy under the discretionary inflation and price-level targeting regime, respectively, and examines conditions for equilibrium determinacy. Here, we focus on specific targeting rules, following the terminology of Svensson (2003), i.e. optimality conditions implied by the monetary authority’s objective function.

3.1 Inflation targeting Under the inflation-targeting regime, yt = [πt , xt ]′ in (6) and q1,1 = 1, q2,2 = λ IT , where qk,m is the element from the kth row and mth column of matrix Q and λ IT > 0 denotes the central banker’s relative weight on output gap stabilization which may differ from society’s λ . The first-order conditions of the minimization problem result in the following (specific) targeting rule

πt = −

λ IT xt . κ˜ − δ κσ

(7)

Note, that for δ = 0, (7) reduces to the optimality condition of the standard New η Keynesian model as analyzed by Clarida et al. (1999). The assumption δ < σ + σ implies that inflation above target always requires the monetary authority to contract demand below potential. To examine the implications of the cost channel for equilibrium determinacy, combine (1), (2) and targeting rule (7), to get

πt =

λ IT (β + δ κ ) − (κ˜ − δ κσ ) δ κσ λ IT π E ut . + t t+1 λ IT + (κ˜ − δ κσ )2 λ IT + (κ˜ − δ κσ )2

Published by The Berkeley Electronic Press, 2011

(8)

5

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

Equilibrium determinacy requires the expected-inflation coefficient in (8) to be less than one in absolute value.5 Proposition 1 The model composed of (1), (2), (3) and targeting rule (7) exhibits a unique rational expectations equilibrium {πt , xt , it }, if and only if either of the following two cases is satisfied. Case I:

β +δκ > 1

(9)

λ IT >

(10)

λ IT

(11)

κ˜ − δ κσ (2δ κσ − κ˜ ) 1+β +δκ κ˜ − δ κσ < κ˜ . β +δκ −1

Case II: Equation (10) and the inequality opposite to (9) hold. Proof See Appendix A.1. When the cost channel is weak, such that the inequality opposite to (9) holds, (10) constitutes the sole condition for determinacy on λ IT . This condition is binding iff δ > σ2+ση . Otherwise a determinate rational expectations equilibrium exists for all λ IT > 0.6 Clearly, this result always applies to the standard model without the cost channel, δ = 0. If, on the other hand, the cost-channel coefficient δ is large enough such that (9) holds, (10) and (11) are the relevant conditions for determinacy. In this case, the set of parameter values for λ IT is constraint from above and, for δ > σ2+ση , also from below. Table 2 provides the determinacy conditions under the baseline calibration for alternative values of δ . The second column contains the results for inflation targeting. We observe that a cost-channel coefficient of δ = 0.5 or δ = 0.75 does not impose particular tight constraints on the choice of λ IT as the upper bounds are rather high. Increasing δ further changes the picture. For instance, if δ = 1, values of λ IT below 0.0018 or above 0.2428 will induce indeterminacy. Values of δ above one reduce the upper bound of the determinacy region even further, while the lower bound first increases and then falls. In general, Table 2 reveals that in a cost-channel economy neither the choice of an inflation nutter nor of a central banker putting too much weight on output-gap stabilization seems to be desirable if one wants to ensure a unique bounded rational expectations equilibrium. 5 Llosa and Tuesta (2009) also consider determinacy conditions for inflation-targeting rules in the

presence of the cost channel but restrict the analysis to the case when δ = 1. 6 Throughout the paper, we refer to local determinacy of equilibrium.

http://www.bepress.com/bejm/vol11/iss1/art3

6

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

Table 2: Determinacy conditions for the baseline calibration

δ Inflation targeting Price-level targeting IT 0 0<λ 0 < λ PT 0.5 0 < λ IT < 0.9789 0 < λ PT IT 0 < λ < 0.4656 0 < λ PT 0.75 0.0018 < λ IT < 0.2428 0.0002 < λ PT 1 IT 0.0003 < λ PT 1.25 0.0027 < λ < 0.1183 0.0017 < λ IT < 0.0388 0.0001 < λ PT 1.5

3.2 Price-level targeting Under the price-level-targeting regime, yt = [pt , xt ]′ in (6) and q1,1 = 1, q2,2 = λ PT , where λ PT > 0 denotes the central banker’s relative weight on output gap stabilization. The value function has the form  1 V (pt−1 , ut ) = min{ pt2 + λ PT xt2 + β Et V (pt , ut+1 )}, (12) 2 where the minimization is subject to pt =

β +δκ κ˜ − δ κσ 1 pt−1 + Et p (pt , ut+1 ) + xt 1+β +δκ 1+β +δκ 1+β +δκ 1 δ κσ Et x (pt , ut+1 ) + ut , + 1+β +δκ 1+β +δκ

(13)

and the law of motion of the cost-push shock (3).7 The functions p (pt , ut+1 ) and x (pt , ut+1 ) denote the price level and the output gap that the central bank expects to be realized in period t + 1 in equilibrium, contingent on the realization of the exogenous state variable ut+1 . Specifically, we restrict attention to the minimumstate-variable (MSV) solution p (pt−1 , ut ) = apt−1 + but x (pt−1 , ut ) = cpt−1 + dut .

(14) (15)

The first-order condition of the optimization problem results in the following targeting rule κ˜ − δ κσ β λ PT PT pt + λ xt = Et xt+1 , (16) ω ω 7 The

former equation results from combining (1) and (2) rewritten in terms of the price level.

Published by The Berkeley Electronic Press, 2011

7

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

where ω = 1 + β (1 − a) + δ κ (1 − a − σ c). The details of the derivation are deferred to Appendix A.2. Note, that ω depends on the parameters a and c of the MSV solution. In the case of price-level targeting, there is no simple closed-form solution for these parameters and we employ a bisection method to obtain a, from which one can then determine the other parameters of the solution functions. Substituting (4) into (1) and (2), and using targeting rule (16) the model can be written as Azt = BEt zt+1 +Cut ut = ρ ut−1 + εt , where zt = [pt , xt , pt−1 ]′ and   1 0 0  ˜ 0 λ PT A =  κ −ωδ κσ , 1 κ˜ −δ κσ 1 − 1+β +δ κ − 1+β +δ κ



 B=

(17)

0 0

β +δ κ 1+β +δ κ

0

βλ ω δ κσ 1+β +δ κ

 1 0  . (18) 0

The form of C is omitted since it is not required for what follows. Define Ω ≡ B−1 A. Since we have two free endogenous variables, xt and pt , and one predetermined endogenous variable, pt−1 , the dynamic system (17) has a unique bounded solution if and only if Ω exhibits two eigenvalues outside the unit circle and one eigenvalue inside the unit circle. Proposition 2 The model composed of (1), (2), (3), (4) and targeting rule (16) exhibits a unique rational expectations equilibrium {πt , xt , pt , it }, if and only if

λ PT >

2δ κσ − κ˜ 1 (κ˜ − δ κσ ) . 2 (ω + β ) (1 + β + δ κ )

(19)

Proof See Appendix A.3. The parameter ω is strictly positive.8 Hence, if δ ≤ σ2+ση , condition (19) is trivially satisfied and any λ PT > 0 will induce a unique rational expectations equilibrium. If δ > σ2+ση , condition (19) constitutes a positive lower bound on the set of desirable λ PT . The third column of Table 2 provides the determinacy conditions under the baseline calibration. Low and intermediate values of δ do not impose any constraint on λ PT . In the case of δ = 1, constraint (19) becomes binding but the implied lower 8 It



holds lim a λ PT = 0 and lim a λ PT = min{1, β +1δ κ }. It follows max{1, β + δ κ } ≤ λ PT →0

λ PT →∞

δ κσ PT . ω ≤ 1 + β + δ κ + κ˜ − δ κσ , where ω is decreasing in λ

http://www.bepress.com/bejm/vol11/iss1/art3

8

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

bound on λ PT is close to zero. The same holds true for the two considered values of δ > 1. Note however, that λ PT and λ IT are distinct objects, so that a direct comparison of their absolute values contains only limited information. In order to assess the tightness of the lower bounds under price-level and inflation targeting we may employ the relationship between the variance of the price level and the variance of ρa inflation under price-level targeting.9 It holds Var (pt ) = 2(1−1+ ρ )(1−a) Var (πt ). Substituting this expression into the loss function of the price-level-targeting central bank, it follows that the lower bound under price-level targeting, λ PT , is tighter ρ )(1−a) than the lower bound under inflation targeting, λ IT , iff λ PT 2(1−1+ > λ IT . Unρa der the benchmark calibration, this is not the case. In fact, it turns out that the transformed lower bound corresponds to a slightly smaller value than the associated λ PT . For instance, in the case of δ = 1.25 the transformed lower bound equals 0.0002 whereas λ PT = 0.0003.

4 Stabilization performance Abstracting from equilibrium determinacy, the question remains to what extent the presence of the cost channel alters the stabilization bias resulting from discretionary optimization. A convenient way to depict the central banker’s stabilization trade-off is to reformulate society’s objective function in terms of the variance of the output gap and inflation. In particular, for β → 1, the society’s loss function evaluated under the unconditional expectations operator, E, becomes equivalent to Var (πt ) + λ Var (xt ), a weighted average of the unconditional inflation variance and output gap variance. Figure 1 plots variance frontiers for δ = 0 (no cost channel) and δ = 1, characterizing the combinations of output-gap variance and inflation variance attainable under price-level and inflation targeting, respectively. Different points on the frontiers refer to different values of the preference parameter λ on the interval [0.1, 1.2]. The central bank’s weight on output gap variation in its loss function, λ j , j ∈ {IT, PT }, is determined numerically in order to minimize society’s loss function for a given value of λ . Variance combinations to the south-west of a particular frontier are unattainable in the respective regime whereas combinations to the north-east are inefficient. Beginning with the standard model without the cost channel, the figure shows that price-level targeting leads to an improved trade-off between inflation and output-gap stabilization. This continues to hold true when we move to the model with the cost channel (δ = 1). However, the gap between the frontiers representing the two policy regimes widens as the stabilization trade-off 9 Appendix A.2 contains the expressions for the variances of the price level, inflation and the output gap under price-level targeting.

Published by The Berkeley Electronic Press, 2011

9

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

Figure 1: Variance frontiers 16 Price−level targeting (δ =0) Inflation targeting (δ =0) Price−level targeting (δ =1) Inflation targeting (δ =1)

14 12

Var(x)

10 8 6 4 2 0

0

1

2

3 Var(π)

4

5

Notes: Variance frontiers are shown for the baseline calibration. Society’s preference parameter λ is varied on the interval of [0.1, 1.2] and the central banker’s preference parameter λ j , j ∈ {IT, PT } is determined numerically in order to minimize society’s loss function for a given value of λ .

under inflation targeting deteriorates. In the case of purely transitory exogenous shocks, we can establish this result also analytically. Proposition 3 For ρ = 0, the trade-off faced by discretionary inflation targeting between inflation and output gap variability is always less favorable for δ = 1 than for δ = 0. Proof See Appendix A.4. In contrast, under price-level targeting the frontier shifts only slightly, and leads to an improved trade-off in the case of a sufficiently high preference parameter λ . Intuitively, in the presence of the cost channel a change in the one-period nominal interest rate has opposite effects on inflation via the demand-side and the supplyside channel of monetary transmission. Following the demand-side channel, a (sufficiently large) rise in the nominal rate raises the real interest rate. The induced http://www.bepress.com/bejm/vol11/iss1/art3

10

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

reduction in aggregate demand leads to a fall in labor demand. Thus, equilibrium wages decrease, thereby lowering firms’ marginal costs and attenuating potential inflationary pressures. However, due to the supply-side channel, an increase in the nominal rate raises firms’ borrowing costs, implying a direct positive effect on marginal costs and thus on firms’ pricing decisions. The performance of pricelevel targeting is less affected by the introduction of the cost channel because it employs the dependence of next period’s policy actions on current economic conditions. Price-level stationarity implies that any contemporaneous above-trend inflation episode has to be followed by enough below-trend inflation in order to bring the price level back to its target path. But lower expected future inflation dampens current inflation, which reduces the trade-off for the monetary authority in the contemporaneous period.

5 Conclusion This paper examines the implications of the cost channel of monetary transmission for the desirability of two alternative monetary policy regimes, inflation targeting and price-level targeting. Employing a standard New Keynesian model of the business cycle with a cost channel, we confirm that discretionary inflation targeting can suffer from strongly shrinking determinacy regions. In contrast, price-level targeting remains almost insusceptible to equilibrium indeterminacy, independently of the size of the cost channel. Furthermore, the stabilization trade-off under inflation targeting deteriorates with increasing importance of the cost channel, whereas the stabilization performance under price-level targeting is virtually unaffected. Accordingly, the quantitative difference in the performance of inflation and price-level targeting increases, suggesting that, to the extent that the cost channel is empirically relevant, comparison exercises based on the standard New Keynesian model underestimate the gains from price-level targeting. Both policy regimes differ in their ability to make policy history dependent. This property of price-level targeting turns out to be especially important once we take supply-side effects of monetary policy transmission explicitly into account.

Published by The Berkeley Electronic Press, 2011

11

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

Appendix A Appendix A.1 Proof of Proposition 1 From (8), define γ ≡

λ IT (β +δ κ )−(κ˜ −δ κσ )δ κσ . λ IT +(κ˜ −δ κσ )2

Equilibrium determinacy requires

−1 < γ < 1. If β + δ κ ≤ 1, the condition γ < 1 is equivalent to

λ IT (1 − β − δ κ ) > − (κ˜ − δ κσ ) κ˜ ,

(20)

η IT > 0 and can therefore be omitted. which, given δ ∈ [0, σ + σ ), is satisfied for all λ If β + δ κ > 1, the condition γ < 1 is equivalent to

λ IT <

κ˜ − δ κσ κ˜ , β +δκ −1

(21)

which represents condition (11) in Proposition 1. The condition γ > −1 is equivalent to

λ IT >

κ˜ − δ κσ (2δ κσ − κ˜ ) , 1+β +δκ

(22)

which represents condition (10) in Proposition 1.

Appendix A.2 Price-level targeting under discretion The value function has the form  1 2 pt + λ PT xt2 + β Et V (pt , ut+1 )}. 2 We restrict attention to the minimum-state-variable (MSV) solution V (pt−1 , ut ) = min{

p (pt−1 , ut ) = apt−1 + but x (pt−1 , ut ) = cpt−1 + dut .

(23)

(24) (25)

Combining (1) and (2) rewritten in terms of the price-level with (24) and (25) written one period ahead leads to 1 κ˜ − δ κσ 1 + ((β + δ κ ) b + δ κ d) ρ pt−1 + xt + ut , (26) ω ω ω where ω = 1 + β (1 − a) + δ κ (1 − a − σ c). Solving minimization problem (23) with respect to xt subject to (26) results in the following first-order condition pt =

pt

∂ pt ∂ V (pt , ut+1 ) ∂ pt + λ PT xt + β Et = 0. ∂ xt ∂ pt ∂ xt

http://www.bepress.com/bejm/vol11/iss1/art3

(27) 12

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

It is

∂ V (pt−1 , ut ) ∂ pt ∂ V (pt , ut+1 ) ∂ pt = pt + β Et ∂ pt−1 ∂ pt−1 ∂ pt ∂ pt−1 PT λ = − xt . ˜ κ − δ κσ

(28)

Writing (28) one period ahead and taking expectations, the resulting equation can be substituted into (27) to arrive at

κ˜ − δ κσ β λ PT pt + λ PT xt = Et xt+1 , ω ω

(29)

which is targeting rule (16) representing optimal discretionary price-level targeting. Combining (26) and (29), it follows from (24) that a =

λ PT ω

(30) (κ˜ − δ κσ )2 + λ PT ω 2 + β λ PT (1 − ω a) λ PT ω + λ PT ρ [2β bω − β (1 + β bρ ) + δ κ (b + σ d) (ω − β ρ )] b = . (31) (κ˜ − δ κσ )2 + λ PT ω 2 + β λ PT (1 − ω a) Using (26) and (24), it follows from (25) that (1 − a) (1 − (β + δ κ ) a) κ˜ − (1 − a) δ κσ 1 − b [ω − (β + δ κ ) ρ ] d = − . κ˜ − (1 − ρ ) δ κσ c = −

(32) (33)

Employing the solution functions, the following expressions for the variance of the price level, the inflation rate and the output gap can be derived: (1 + ρ a) b2 Var (εt ) (34) (1 − ρ a) (1 − a2 ) 2 (1 − ρ ) b2 Var (εt ) Var (πt ) = (35) (1 − ρ a) (1 + a)

(1 + ρ a) b2 c2 + (1 − ρ a) 1 − a2 d 2 + 2ρ 1 − a2 bcd Var (εt ) . Var (xt ) = (1 − ρ a) (1 − a2 ) (36)

Var (pt ) =

Published by The Berkeley Electronic Press, 2011

13

The B.E. Journal of Macroeconomics, Vol. 11 [2011], Iss. 1 (Topics), Art. 3

Appendix A.3 Proof of Proposition 2 The characteristic polynomial of Ω is given by P (µ ) = −µ 3 + A2 µ 2 − A1 µ + A0 , where ω (37) A0 = β (β + δ κ ) 1 (κ˜ − δ κσ )2 1+β +δκ A1 = + ω (38) + PT β + δ κ β λ (β + δ κ ) β (β + δ κ ) 1+β +δκ ω δ κσ A2 = + − (39) (κ˜ − δ κσ ) . β +δκ β β λ (β + δ κ ) Furthermore, κ˜ − δ κσ (40) P (1) = −κ˜ β λ PT (β + δ κ )   ω 1+β +δκ κ˜ − δ κσ P (−1) = 2 1 + + (κ˜ − 2δ κσ ) . (41) β β +δκ β λ PT (β + δ κ ) If P (1) < 0 and P (−1) > 0, continuity implies that the characteristic equation has an odd number of roots inside the unit circle. Suppose, all three eigenvalues would lie inside the unit circle, then |det (Ω) | = |A0 | < 1. But det (Ω) = A0 = β (β ω+δ κ ) > 1 which follows from ω ≥ max{1, β + δ κ }. Hence, if P (1) < 0 and P (−1) > 0, Ω has two eigenvalues outside the unit circle and one eigenvalue inside the unit circle. η PT > 0 and that P (−1) > 0 Observing that, given δ ∈ [0, σ + σ ), P (1) < 0 for all λ is equivalent to condition (19) in Proposition 2 completes the proof.

Appendix A.4 Proof of Proposition 3 The solution functions for πt and xt under discretionary inflation targeting have the form
πt = f IT δ , λ IT ut
xt = gIT δ , λ IT ut .

IT κ˜ −δ κσ IT δ , λ IT = − For ρ = 0, we have f IT δ , λ IT = IT λ˜ . 2 and g IT +(κ˜ −δ κσ )2 λ κ − δ κσ ) λ +(  
Define λˆ IT such that gIT 1, λˆ IT = gIT 0, λ IT . This implies the following relationship between λˆ IT and λ IT

η λ IT + κ 2 σ η . σ +η  
IT IT IT IT ˆ
http://www.bepress.com/bejm/vol11/iss1/art3

14

Schmidt: The Cost Channel, Indeterminacy, and Price-Level Stabilization

References Aksoy, Y., H. S. Basso, and J. Coto-Martinez (2009): “Investment cost channel and monetary transmission,” Working Paper. Barth, M. J. and V. A. Ramey (2001): “The cost channel of monetary transmission,” NBER Macroeconomics Annual, 16, 199–240. Brueckner, M. and A. Schabert (2003): “Supply-side effects of monetary policy and equilibrium multiplicity,” Economics Letters, 79, 205–211. Chowdhury, I., M. Hoffmann, and A. Schabert (2006): “Inflation dynamics and the cost channel of monetary transmission,” European Economic Review, 50, 995– 1016. Christiano, L. J., M. Trabandt, and K. Walentin (2010): “DSGE models for monetary policy analysis,” Prepared for the Handbook of Monetary Economics. Clarida, R., J. Gal´ı, and M. Gertler (1999): “The science of monetary policy: A New Keynesian perspective,” Journal of Economic Literature, 37, 1661–1707. Clarida, R., J. Gal´ı, and M. Gertler (2000): “Monetary policy rules and macroeconomic stability: Evidence and some theory,” Quarterly Journal of Economics, 105, 147–180. Gabriel, V., P. Levine, C. Spencer, and B. Yang (2008): “On the (ir)relevance of direct supply-side effects of monetary policy,” Working Paper. Kryvtsov, O., M. Shukayev, and A. Ueberfeldt (2008): “Adopting price-level targeting under imperfect credibility,” Bank of Canada Working Paper 2008-3. Llosa, L.-G. and V. Tuesta (2009): “Learning about monetary policy rules when the cost channel matters,” Journal of Economic Dynamics and Control, 33, 1880– 1896. Ravenna, F. and C. E. Walsh (2006): “Optimal monetary policy with the cost channel,” Journal of Monetary Economics, 53, 199–216. Surico, P. (2008): “The cost channel of monetary policy and indeterminacy,” Macroeconomic Dynamics, 12, 724–735. Svensson, L. E. (2003): “What is wrong with Taylor rules? Using judgement in monetary policy through targeting rules,” Journal of Economic Literature, 41, 426–477. Tillmann, P. (2008): “Do interest rates drive inflation dynamics? An analysis of the cost channel of monetary transmission,” Journal of Economic Dynamics and Control, 32, 2723–2744. Tillmann, P. (2009): “Optimal monetary policy with an uncertain cost channel,” Journal of Money, Credit and Banking, 41, 885–906. Vestin, D. (2006): “Price-level versus inflation targeting,” Journal of Monetary Economics, 53, 1361–1376.

Published by The Berkeley Electronic Press, 2011

15