Costa Rica: Selected Issues and Analytical Notes

Abstract

Costa Rica: Selected Issues and Analytical Notes

Monetary Policy Stance1

This note presents an empirical assessment of the current monetary policy stance in Costa Rica. We find that monetary policy is appropriately expansionary after eight consecutive policy rate decreases since early 2015. All the employed estimates of the neutral monetary policy rate support this conclusion. The decrease in inflation below the lower bound of the target range in 2015 can largely be explained by the fall in international commodity prices—which in turn resulted in lower domestic prices of regulated goods and services, and agricultural products—and a widening output gap. While the current easing cycle is projected to support the return of inflation within the new target range and of economic activity to its potential level over the short and medium term respectively, the BCCR should stand ready to start reversing it should signs emerge that price pressures are materializing, also in light of the prospective normalization of global interest rates.

1. The central bank seized the fall of inflation below the lower limit of the target range to reduce the target inflation band. Headline inflation was negative and core inflation stood below the central bank’s target range of 3-5 percent during most of 2015, while expectations fell within the target range for the first time. To stimulate the economy, the central bank has cut the monetary policy rate by 350 cumulative basis points since the beginning of 2015, to 1.75 percent in January 4, 2016, undoing the three interest rate increases implemented in 2014. The central bank also took advantage of the negative inflation shock to decrease the inflation target range to 2-4 percent, in line with trading partners.

2. To assess the adequacy of the current monetary policy stance this note estimates the neutral monetary policy interest rate. Looking at the difference between the actual policy rate and the estimated short-run operational neutral interest rate (NIR)—which would yield a stable inflation under closed output gap conditions—we can assess the monetary stance taking into account the economy’s current position in the cycle.

3. We use five different empirical approaches to estimating the neutral monetary policy interest rate. Following Magud and Tsounta (2012), we employ: (1) an Uncovered Interest Parity (UIP) condition; (2) a specification of the Taylor rule augmented for inflation expectations; and (3) a general equilibrium model that focuses on aggregate demand and supply equations. In addition, we estimate: (4) a forward looking monetary model à la Clarida, Galí, and Gertler, 1998, and Galí and Monacelli, 2005; and (5) a linear semi-structural new-Keynesian Quarterly Projection Model (QPM) with model-consistent expectations, of the kind used in several Central Banks and other institutions, including the IMF (Berg, Karam and Laxton, 2006), to help set an appropriate level of the policy interest rate given an inflation target and the macroeconomic conditions. The first three models use monthly data from March 2006 to November 2015; the fourth model from January 1994 to August 2015; and the fifth model uses quarterly data from 1996Q1 to 2016Q1.

4. The UIP condition points to a neutral nominal interest rate of 4.6 percent. This value assumes an implicit annual nominal depreciation in line with the inflation differential with the U.S. to maintain the real exchange rate constant, and a country risk premium. The “model” comprises the following equations:

it=it*+E^+ρE^=RER+(ππ*)

where it is the neutral policy rate in Costa Rica, it* is the current policy rate in the U.S., Ê is the expected nominal depreciation of the colon vis-à-vis the dollar, ρ is the risk premium as captured in the country’s external bond spreads, rer is the real exchange rate, π and π* are current end 2016 inflation projections in Costa Rica and the U.S respectively.

5. The expected-inflation augmented Taylor rule model estimates a neutral nominal interest rate of 5.4 percent. The model incorporates information from the yield curve and inflation expectations, in addition to the standard output gap and deviations from the inflation target in standard Taylor rule models. The real neutral level for the monetary policy rate estimated under this model is 2.4, which corresponds to a neutral nominal interest rate of 5.4 percent with the staff’s projected inflation of 3 percent at the end 2016.2 These results should be interpreted with caution, however, given that the model implicitly assumes a certain degree of sophistication of a country’s financial markets. The model comprises the following system of equations:

rt=rt*+πt+6e+β(πt+5eπt*)+θy˜t+εt1Rt=rt*+α+πt+6e+εt2rt*=rt1*+gt1gt=gt1+ϑt1

where rt is the short-term interest rate (rate on the central bank’s open-market operations), rt* is the neutral real policy rate, πet is the end-of-year inflation expectation at time t, πet+6π*t is the deviation of the end-of-year inflation expectation at time t+6 from the target, y˜t is the output gap, Rt is the long-term rate (approximated by a long-term time deposit rate), and α is the term premium. All disturbance terms (ɛt1,ɛt2andϑt1) are assumed to have zero mean and constant variance. The transition process for the NIR is defined as a random walk process with drift, gt being the growth rate of the unobserved state variable rt*. The model parameters, unobserved variables, and residual terms are estimated using a Kalman filter based on a log likelihood function as in Durbin and Koopman (2001).

6. The general equilibrium model puts the nominal neutral rate at 5.6 percent. This model includes an Investment-Savings (IS) equation—that relates the output gap to its own lags and lags of deviations of the monetary policy rate from neutral levels—and a Phillips curve that relates inflation to the output gap. It depends less than the previous one on the structure of financial markets; however, it still assumes that the monetary transmission channel works efficiently. The model consists of the following system of equations:

(ytyt*)=s=1Sαsy(ytsyts*)+v=1Vαvr(rtvrtv*)+x1,tα+εtyπ^t=p=1Pβpππ^tp+q=1Qβqy(ytqytq*)+x2,tβ+εtπyt=yt*+εtcyt*=yt1*+gt1gt=gt1+εtgrt*=rt1*+εtr

where ytyt* is the output gap, rtrt* is the deviation of the nominal policy rate from the neutral policy rate, π^t is the deviation in core inflation from the inflation target, x1 is the cyclical deviations of the oil prices and x2 is a vector of two variables, the cyclical deviations of the food price index and the cyclical deviations of the real effective exchange rate. All disturbance terms (ɛty,ɛtπ,ɛtc,ɛtg and ɛtr) are assumed to have zero mean and constant variance. The (unobserved) NIR and potential GDP are estimated, along with the model parameters, with a Kalman filter assuming that the NIR follows a random walk, the potential GDP grows at a rate gt, and real GDP is given by stochastic deviations from its potential level, using a log likelihood function à la Durbin and Koopman (2001). 3

7. The forward-looking monetary model yields a neutral interest rate of 4.8 percent. The model includes three structural equations derived for a small open economy. Specifically: a New Keynesian Philips curve with international oil prices; a forward-looking IS curve with real exchange rate and foreign demand; and a standard Taylor rule with a smoothing parameter. Parameters are estimated by the generalized method of moments (GMM). The system was then solved and found to be saddle-path stable using Blanchard and Kahn’s method.4 The model consists of the following equations:

xt=xt+6+α(rtr¯)+δΔst+ψΔyt*+εtxπt=(1ϕ1)πt1+ϕ1πt+6+ϕ2xt+6+ϕ3πtoil+εtπrt=ρrt1+(1ρ)[r¯+β(πt+6π¯t)+γxt+6]+εtr

where xt is the output gap at time t, Δst is the annual rate of change of the real exchange rate index between t and t–1, Δyt* is the annual growth rate of foreign demand (approximated with the US GDP growth rate), πt is the inflation rate, calculated as the annual rate of change of the CPI, and πtoil the annual rate of change of the international oil price index; rt is the monetary policy rate and r¯t is the neutral (nominal) interest rate. Finally, the error term in each equation, εt, is a linear combination of forecast errors and an exogenous disturbance (by assumption, this error term is orthogonal to the set of instruments).

8. The QPM model gives a neutral interest rate of 4.4 percent. The model consist of four basic behavioral equations—for aggregate demand (IS curve), (short term) aggregate supply (Phillips curve), the UIP condition, and a Monetary Policy rule—and several identities. The aggregate demand equation includes a monetary condition index which combines deviations of both the interest and exchange rates from their equilibrium (neutral) levels. The aggregate supply equation accounts for shocks to energy and food prices, as well as to core inflation. The UIP condition and monetary policy rule are modeled to accommodate nominal exchange rate persistency and imperfect control of the money market due to FX Central Bank interventions, and imperfect capital mobility. The model’s main equations are:

Aggregate demand:

y^t=b1y^t1b2mcit+b3y^t*+ɛtymcit=b4(r^t+cr_premt)+(1b4)(z^t)

Aggregate supply:

πtcore=a1πt1core+(1a1)Etπt+1+a2rmct+ɛtπrmct=a3y^t+(1a3)z^tπt=woilπtoil+wfoodπtfood+(1woilwfood)πtcore

UIP:

st+1st=e2(π¯tπ¯t*+Δz¯t)+(1e2)(it*it+premt)+ɛts

Monetary Policy Rule:

it=h1(Δst+1+it*+premt)+(1h1)(g1it1n+(1g1)(itn+g2(Etπt+4πT)+g3y^t))+ɛti

Where y^t is the output gap at time t, mcit is the real monetary conditions index, y^t* is the foreign output gap, r^t and z^t are the real interest rate and RER gap, rmct are real marginal costs, cr_premt and premt are credit and sovereign premium terms respectively. πt,πtcore,πtoil,and πtfood are headline, core, oil and food inflation respectively, and w the respective weight in the basket. π¯t and π¯t* are the domestic and foreign (target) inflation objective, and Δz¯t the ‘desired’ change in the real exchange rate path along which the CB smoothes the actual nominal exchange rate. ɛty,ɛtπ,ɛts and ɛti are demand, cost-push, exchange rate, and monetary policy shocks.5 The model is solved using a variant of the Blanchard and Khan (1982) algorithm. The model’s structural shocks and unobservable variables—i.e. trends and gaps, whose dynamics are jointly described by a VAR (1) representation—are estimated using a Kalman smoother as described in Hamilton (1994, chapter 13).

9. To conclude, the monetary stance seems broadly adequate, however, vigilance is required going forward. Averaging the results from the five models above, the nominal neutral interest rate for Costa Rica is estimated at 4.6 percent. This value is lower than what was estimated in 2014 (5.1 percent), consistent with comparatively lower estimates of potential output and inflation. The current nominal monetary policy interest rate of 1.75 percent is below the estimated neutral monetary policy rate. With a positive output gap and inflation projected to remain below target in the near term, the monetary policy is appropriately expansionary. However, the BCCR should remain vigilant and stand ready to raise interest rates should inflation increase faster than anticipated due to: (i) faster U.S. growth; (ii) a rebound in international commodity prices; (iii) upward food price pressures stemming from regional droughts; and (iv) possible second-round effects from currency depreciation following the normalization of global interest rates.

Table 1.

Costa Rica: Neutral Interest Rate, Latest Estimates

article image
Sources: National authorities and Fund staff estimates.Notes:

All units expressed as percent points unless otherwise stated.

(bps): Basis points

References

  • Berg, A., P. Karam, and D. Laxton, 2006, “A Practical Model-Based Approach to Monetary Policy Analysis—Overview,Working Paper 06/80 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Blanchard, O. J., and C. M. Kahn, 1980, “The Solution of Linear Difference Models under Rational Expectations,Econometrica, 48, pp. 13051311.

    • Search Google Scholar
    • Export Citation
  • Clarida, R., J. Galí and M. Gertler, 1998, “Monetary Policy Rules in Practice Some International Evidence,European Economic Review, 42, pp. 10331067.

    • Search Google Scholar
    • Export Citation
  • Durbin, J., and S. J. Koopman (2001). Time Series Analysis by State Space Methods. Oxford University Press, Oxford.

  • Galí, J and T. Monacelli, 2005, “Monetary Policy and Exchange Rate Volatility in a Small Open Economy,Review of Economic Studies, 72, pp. 707734.

    • Search Google Scholar
    • Export Citation
  • Hamilton, J., (1995). Time Series Analysis. Princeton, NJ: Princeton University Press.

  • Magud, N., and E. Tsounta, 2012, “To Cut or Not to Cut? That is the (Central Bank’s) Question: In Search of the Neutral Interest Rate in Latin America,Working Paper 12/243 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
1

Prepared by Valentina Flamini and Rodrigo Mariscal Paredes

2

The estimated timeframe for the transmission from monetary policy rates to inflation is 5 to 6 months in Costa Rica.

3

See Magud and Tsounta (2012) for additional methodological details.

4

This condition guarantees that the system will converge to the steady state for any given initial value in the state variables and any given change in the value of the control variables that satisfy the feasibility constrains.

5

See Berg and others (2006) for additional model equations and properties.