Inflation-Output Trade-Offs Under Flexible Targeting

Why Flexible Targeting Involves a Trade-Off

A central bank may care both about keeping inflation near its target and about keeping output near potential. When the two objectives point in different directions, meeting the inflation target exactly can require accepting a larger output gap. Flexible inflation targeting makes that trade-off explicit: the central bank assigns a weight to each deviation and chooses a policy that minimizes their combined cost.

The model in this tutorial is deliberately simple. It describes a one-period choice and treats the policy instrument as capable of delivering any desired output gap. The inflation-output relation summarizes the effects of expectations and disturbances; it is not a full model of how policy rates transmit through the economy. In particular, the results are conditional on the stated relation and assumptions.

Definition: Let \(\pi\) be the inflation rate and \(\pi^T\) the inflation target. The inflation deviation is \(\tilde{\pi}=\pi-\pi^T\). Let \(x\) be the output gap, measured as output minus potential output divided by potential output. Thus, \(x=0.02\) is a positive output gap of 2%. In the reduced-form model, \(\tilde{\pi}=\kappa(x-\bar{x})\), where \(\kappa>0\) is the inflation response to the gap and \(\bar{x}\) is the output gap at which inflation is exactly on target.

The parameter \(\bar{x}\) summarizes the position of the inflation-output relation for the period being analyzed. It can differ from zero when a disturbance or the starting conditions shift that relation. If \(\bar{x}>0\), for example, this model says inflation is on target at a positive output gap; choosing \(x=0\) then leaves inflation below target. The model does not explain the source of that shift.

Flexible targeting is represented by a quadratic loss function. Let \(\lambda>0\) measure the relative weight placed on output-gap stabilization. A larger \(\lambda\) means output deviations are more costly relative to inflation deviations.

Definition: The central bank’s one-period loss under flexible targeting is \(L(x)=\tilde{\pi}^{\,2}+\lambda x^2\). Using the model relation, this is \(L(x)=\kappa^2(x-\bar{x})^2+\lambda x^2\). Flexible targeting chooses the value of \(x\) that minimizes this loss. Strict inflation targeting instead imposes \(\tilde{\pi}=0\) exactly.

Deriving the Flexible-Targeting Choice

Substitute the inflation-output relation into the loss function. The first term is smallest when \(x=\bar{x}\), while the second is smallest when \(x=0\). Unless \(\bar{x}=0\), those two goals conflict. Differentiating the loss gives the condition for the minimum:

L(x)=\kappa^2(x-\bar{x})^2+\lambda x^2

\frac{dL}{dx}=2\kappa^2(x-\bar{x})+2\lambda x=0 \quad\Longrightarrow\quad x^*=\frac{\kappa^2}{\kappa^2+\lambda}\bar{x}.

The second derivative is \(2(\kappa^2+\lambda)>0\), so this stationary point is the unique minimum. Substitution into the inflation relation gives the associated inflation deviation. The minimum loss can be expressed directly in terms of the parameters:

\tilde{\pi}^*=-\frac{\kappa\lambda}{\kappa^2+\lambda}\bar{x}, \qquad L^*=\frac{\lambda\kappa^2}{\kappa^2+\lambda}\bar{x}^{\,2}.

These formulas have useful interpretations. The factor \(\kappa^2/(\kappa^2+\lambda)\) lies between zero and one. Thus, flexible targeting moves the output gap partway from zero toward the target-consistent gap \(\bar{x}\). As \(\lambda\) rises, the chosen gap moves closer to zero and the inflation deviation becomes larger in absolute value. As \(\kappa\) rises, a smaller change in the gap can produce a given change in inflation, so the flexible choice places relatively more emphasis on reducing inflation deviation.

Under strict inflation targeting, \(\tilde{\pi}=0\). Since \(\kappa>0\), the relation \(\tilde{\pi}=\kappa(x-\bar{x})\) then requires \(x=\bar{x}\). The strict-target loss is therefore \(L_{\mathrm{strict}}=\lambda\bar{x}^{\,2}\). Flexible targeting has a weakly lower loss in this model because strict targeting is one feasible choice within the flexible minimization problem. If \(\bar{x}\ne0\) and both weights are positive, flexible targeting strictly lowers loss.

Theorem: With \(\kappa>0\), \(\lambda>0\), the stated quadratic loss, and no constraint on the attainable output gap, flexible targeting chooses \(x^*=\kappa^2\bar{x}/(\kappa^2+\lambda)\). Strict targeting chooses \(x=\bar{x}\). If policy choices are constrained, or if the model relation or objective differs, these choices need not apply.

Worked Comparisons

Worked Example: A Positive Target-Consistent Gap

Suppose a hypothetical economy has \(\kappa=0.4\), \(\lambda=0.24\), and \(\bar{x}=0.020\). Compare flexible targeting with strict inflation targeting. Express gaps as decimals in the calculations, then report them as percentages.

Method. Use the reduced-form relation \(\tilde{\pi}=0.4(x-0.020)\) and minimize the stated quadratic loss. The positive \(\bar{x}\) means strict targeting requires a positive output gap.

Flexible choice. First, \(\kappa^2=0.4^2=0.16\), and \(\kappa^2+\lambda=0.16+0.24=0.40\). Therefore,

x^*=\frac{0.16}{0.40}(0.020)=0.008.

The inflation deviation is \(\tilde{\pi}^*=0.4(0.008-0.020)=0.4(-0.012)=-0.0048\), or \(-0.48\) percentage points relative to target. The inflation-loss component is \((-0.0048)^2=0.00002304\). The output-loss component is \(0.24(0.008)^2=0.24(0.000064)=0.00001536\). Thus, \(L^*=0.00002304+0.00001536=0.00003840\).

Strict choice and check. Strict targeting sets \(\tilde{\pi}=0\), so \(0.4(x-0.020)=0\) gives \(x=0.020\), or 2.0%. Its loss is \(0^2+0.24(0.020)^2=0.24(0.0004)=0.000096\). The flexible loss is lower, as expected. The closed-form check gives \(L^*=0.24(0.16)(0.020)^2/0.40=0.00003840\).

Worked Example: A Larger Inflation Weight Relative to Output

Consider a second hypothetical calibration with \(\kappa=0.4\), \(\lambda=0.60\), and \(\bar{x}=0.030\). Calculate both choices and losses. Keep the repeating decimals unrounded until the final reported values.

Method. The flexible choice is the weighted compromise from the derived minimization condition. Strict targeting requires the output gap that sets the inflation deviation to zero.

Flexible choice. Here \(\kappa^2=0.16\) and \(\kappa^2+\lambda=0.76\), so

x^*=\frac{0.16(0.030)}{0.76} =\frac{0.0048}{0.76} =0.006315789\ldots.

The inflation deviation is \(0.4(0.006315789\ldots-0.030)=-0.009473684\ldots\). The inflation-loss component is \(0.000089750693\ldots\). The output-loss component is \(0.60(0.006315789\ldots)^2=0.000023933518\ldots\). Adding these components gives \(L^*=0.000113684211\ldots\), which rounds to \(0.000114\) to six decimal places.

Strict choice and check. Setting \(0.4(x-0.030)=0\) gives \(x=0.030\), or 3.0%. Its inflation deviation is zero and its loss is \(0.60(0.030)^2=0.60(0.0009)=0.00054\). The flexible formula independently gives \(0.60(0.16)(0.030)^2/0.76=0.0000864/0.76=0.000113684211\ldots\), agreeing with the component calculation.

Worked Example: A Negative Target-Consistent Gap

Suppose \(\kappa=0.5\), \(\lambda=0.5\), and \(\bar{x}=-0.020\). Find the flexible and strict choices. This case checks whether the formulas work when the target-consistent gap is negative.

Flexible choice. Since \(\kappa^2=0.25\) and \(\kappa^2+\lambda=0.75\),

x^*=\frac{0.25}{0.75}(-0.020)=-0.006666667\ldots.

The inflation deviation is \(0.5(-0.006666667\ldots+0.020)=0.006666667\ldots\). The loss components are \((0.006666667\ldots)^2=0.000044444444\ldots\) and \(0.5(0.006666667\ldots)^2=0.000022222222\ldots\). Their sum is \(0.000066666667\ldots\), which rounds to \(0.000067\) to six decimal places.

Strict choice and check. Strict targeting requires \(x=\bar{x}=-0.020\), so the output loss is \(0.5(0.020)^2=0.0002\) and inflation loss is zero. The flexible formula gives \(0.5(0.25)(0.020)^2/0.75=0.00005/0.75=0.000066666667\ldots\). The negative sign of \(\bar{x}\) reverses the signs of the chosen gap and inflation deviation, but not the direction of the loss comparison.

Common Traps and Practical Interpretation

The most frequent arithmetic mistake is to report only one component of the flexible loss. In the first worked example, \(0.24(0.008)^2=0.00001536\) is the output component only. It is not the total: the inflation component, \(0.00002304\), must also be added. The correct total is \(0.00001536+0.00002304=0.00003840\). The closed-form minimum-loss formula is a useful independent check, but it does not replace correctly identifying both components.

A second trap is to set \(x=0\) automatically under flexible targeting. The objective penalizes the output gap, but does not require it to be zero. The optimum generally lies between \(0\) and \(\bar{x}\), with the exact location determined by \(\kappa^2/(\kappa^2+\lambda)\). A third trap is to interpret \(x\) as a percentage number in one calculation and a decimal in another. Use \(0.020\) for a 2% gap throughout; switching to 2 without also converting the parameters changes the result by a factor of 100 in the gap and much more in the squared loss.

The framework clarifies what “flexible” means: the target is not ignored, but the central bank accepts some inflation deviation to avoid the full output-gap cost of strict targeting. It is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus. Its numerical results are conditional, not universal policy prescriptions. A different loss function, changing expectations, multiple periods, uncertain data, instrument constraints, or a different inflation-output relation can change the optimal choice.

Key takeaway: Under the stated model, strict targeting sets \(x=\bar{x}\), while flexible targeting chooses \(x=\kappa^2\bar{x}/(\kappa^2+\lambda)\). Calculate inflation deviation from the relation, calculate both squared-loss components, and check the total using \(L^*=\lambda\kappa^2\bar{x}^{\,2}/(\kappa^2+\lambda)\).

Check Your Understanding

For each problem, use the quadratic loss and reduced-form inflation-output relation defined above. Keep decimal gaps as decimals in the calculations.

Let \(\kappa=0.5\), \(\lambda=0.75\), and \(\bar{x}=0.040\). Find the flexible output gap, inflation deviation, flexible loss, and strict-target loss. (Answer: \(x^*=0.010\), \(\tilde{\pi}^*=-0.015\), \(L^*=0.000300\), and \(L_{\mathrm{strict}}=0.001200\).)
Let \(\kappa=0.3\), \(\lambda=0.27\), and \(\bar{x}=-0.050\). Find the flexible output gap, inflation deviation, and both losses. (Answer: \(x^*=-0.0125\), \(\tilde{\pi}^*=0.01125\), \(L^*=0.00016875\), and \(L_{\mathrm{strict}}=0.000675\).)
Let \(\kappa=0.6\), \(\lambda=0.4\), and \(\bar{x}=0.025\). Calculate the flexible choice and compare its loss with strict targeting. (Answer: \(x^*=0.0118421\), \(\tilde{\pi}^*=-0.00789474\), \(L^*=0.000118421\), and \(L_{\mathrm{strict}}=0.000250\).)
Let \(\kappa=0.5\) and \(\bar{x}=0.024\). Compare flexible targeting for \(\lambda=0.25\) and \(\lambda=1\), calculating each output gap, inflation deviation, and loss. (Answer: for \(\lambda=0.25\), \(x^*=0.012\), \(\tilde{\pi}^*=-0.006\), and \(L^*=0.000072\); for \(\lambda=1\), \(x^*=0.0048\), \(\tilde{\pi}^*=-0.0096\), and \(L^*=0.0001152\).)
Suppose \(\kappa=0.4\), \(\bar{x}=0.030\), and the flexible choice is observed to be \(x^*=0.006\). Find the implied \(\lambda\), then calculate the inflation deviation and minimum loss. (Answer: \(\lambda=0.64\), \(\tilde{\pi}^*=-0.0096\), and \(L^*=0.0001152\).)

What You'll Learn