Monetary Policy in Disaster-Prone Developing Countries
Author:
Mr. Alessandro Cantelmo
Search for other papers by Mr. Alessandro Cantelmo in
Current site
Google Scholar
PubMed
Close
,
Nikos Fatouros
Search for other papers by Nikos Fatouros in
Current site
Google Scholar
PubMed
Close
,
Mr. Giovanni Melina
Search for other papers by Mr. Giovanni Melina in
Current site
Google Scholar
PubMed
Close
, and
Mr. Chris Papageorgiou
Search for other papers by Mr. Chris Papageorgiou in
Current site
Google Scholar
PubMed
Close

This paper analyzes monetary policy regimes in emerging and developing economies where climate-related natural disasters are major macroeconomic shocks. A narrative analysis of IMF reports published around the occurrence of natural disasters documents their impact on important macroeconomic variables and monetary policy responses. While countries with at least some degree of monetary policy independence typically react by tightening the monetary policy stance, in a sizable number of cases monetary policy was accommodated. Given the lack of consensus on best practices in these circumstances, a small open-economy New-Keynesian model with disaster shocks is leveraged to evaluate welfare under alternative monetary policy rules. Results suggest that responding to inflation to an extent sufficient to keep inflation expectations anchored, while allowing temporary deviations from its target is the welfare maximizing policy. Alternative regimes such as strict inflation targeting, exchange rate pegs, or Taylor rules explicitly responding to economic activity or the exchange rate would be welfare-detrimental.

Abstract

This paper analyzes monetary policy regimes in emerging and developing economies where climate-related natural disasters are major macroeconomic shocks. A narrative analysis of IMF reports published around the occurrence of natural disasters documents their impact on important macroeconomic variables and monetary policy responses. While countries with at least some degree of monetary policy independence typically react by tightening the monetary policy stance, in a sizable number of cases monetary policy was accommodated. Given the lack of consensus on best practices in these circumstances, a small open-economy New-Keynesian model with disaster shocks is leveraged to evaluate welfare under alternative monetary policy rules. Results suggest that responding to inflation to an extent sufficient to keep inflation expectations anchored, while allowing temporary deviations from its target is the welfare maximizing policy. Alternative regimes such as strict inflation targeting, exchange rate pegs, or Taylor rules explicitly responding to economic activity or the exchange rate would be welfare-detrimental.

1 Introduction

Climate-related natural disaster shocks are major determinants of macroeconomic outcomes in disaster-prone Emerging and Developing Economies (EMDEs). The countries at the top of the list by incidence of natural disasters per square kilometer (see Table A.1 in Appendix A and Cantelmo et al., 2019) are mainly small islands located in the Pacific or the Caribbean, or low-income countries. In other words, countries that are either geographically small or have a small economy. Their size and lack of economic diversification makes them more vulnerable to natural disasters, because when disruptive events occur, they typically affect the whole economy, with their damages representing a large fraction of their GDP (approximately 7% of GDP on average versus 0.5% in their peers less exposed to natural disaster, with extreme events causing damages beyond 200% of GDP).

For instance, Belize was hit by hurricane Keith in October 2000 and by hurricane Iris in October 2001. Both hurricanes caused damages of the tune of 30 percent of GDP each, and GDP growth in 2001 and 2002 was about 8 percentage points lower than in the pre-shock year (Figure 1-a). To put things in perspective, at the time of the oil crisis of the early 1970s—often regarded as a typical large exogenous shock in macroeconomics—U.S. GDP growth in 1974 and 1975 was about 6 percent lower than in 1973 (Figure 1-b). This is to say that, in disaster-prone EMDEs, natural disaster shocks are of the same, if not greater, importance as those that are typically regarded as major macroeconomic shocks in larger or richer countries.

Figure 1:
Figure 1:

Change in Annual GDP Growth Rate in the Aftermath of a Large Macroeconomic Shock

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Source: IMF International Financial Statistics.

It is then no surprise that many central banks respond to these shocks and, in this paper, we show that the monetary policy regime in place makes a sizable difference in terms of welfare. While monetary policy is not a substitute for structural and financial climate adaptation policies, welfare losses from ill-devised monetary policy rules may compound with those deriving from the devastating impacts of disasters. Establishing the adequate monetary policy regime is not a trivial task because, in the aftermath of these events, at least two policy challenges typically arise. The first is that many disaster-prone EMDEs adopt pegs or exchange rate anchors and thus lack full monetary policy independence. The second is that the occurrence of a natural disaster often behaves like a supply shock, generating an increase in inflation and a decrease in GDP (Figure 2). Hence, a trade-off arises between stabilizing inflation and sustaining output. Consequently, the monetary policy response to these events has been rather heterogeneous and there is no consensus on what best practices should be.

Figure 2:
Figure 2:

Distribution of Changes in Key Macroeconomic Variables in the Aftermath of Natural Disasters in Disaster-Prone Countries

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Sources: IMF World Economic Outlook Database and authors’ calculations.The horizontal line inside each box represents the median; the upper and lower edges of each box show the top and bottom quartiles, respectively; and the top and bottom markers denote the maximum and the minimum, respectively. The sample is restricted to cases that suffered a cumulative loss of at least 5 percent of GDP in a given year.

This paper focuses on two research questions. The first is: how is monetary policy set in disaster-prone countries? To answer this question, we build stylized facts using a narrative analysis of IMF staff reports on disaster-prone EMDEs over the past 20 years, published around the occurrence of natural disasters. Issues covered in the analysis include features that might change over time, such as the exchange rate regime and monetary policy independence; changes on key macroeconomic variables, such as GDP and inflation, in the aftermath of disasters; monetary policy responses; and the IMF evaluation and advice on these policy actions.

The second question is: what should be the optimal policy rule? To provide an answer, we use a well-established New-Keynesian model augmented with features capturing important aspects of disaster-prone EMDEs. In particular, we take the rather standard model in Fernández-Villaverde and Levintal (2018), in which disaster shocks affect the capital stock and productivity and we use the same solution method—Taylor projection—which proves to be accurate and tractable in a stochastic environment with large shocks. We extend this framework along three dimensions: (1) we allow the effect of disasters in productivity to have both a permanent and a temporary component as in Gourio (2012), in line with empirical findings, and to affect export demand, so as to capture the experience of many disaster-prone EMDEs; (2) we introduce a small-economy set-up along the lines of Gali and Monacelli (2005) as, again, this is an important aspect of disaster-prone EMDEs; and (3) we consider an array of alternative Taylor-type interest rate rules capturing several possible monetary policy regimes and evaluate the associated welfare outcomes.

The main results can be summarized as follows. The narrative analysis suggests that natural disasters are typically followed by a decline in output and often by an increase in inflation. If there is at least some degree of monetary policy independence, central banks generally change their monetary policy stance in the aftermath of disasters. While monetary policy is commonly tightened, there is a sizable minority of cases in which it is accommodated. Policy appraisals and advice by IMF staff have also been mixed, possibly underscoring that while tightening is a direct consequence of concerns toward inflation and currency depreciations, stimulating economic activity might have been prioritized in certain cases. The model analysis demonstrates that, from a welfare standpoint, a flexible inflation targeting regime—whereby inflation can depart temporarily from target—is superior both to extreme regimes, such as strict inflation targeting or hard pegs, and to hybrid regimes in which monetary policy reacts also to output and the exchange rate, besides inflation.Under flexible inflation targeting, welfare is maximized by keeping the interest rate response at the lowest level needed to keep inflation expectations anchored while minimizing the impact on real activity.

This paper is related to four main strands of the literature. The first strand is the growing body of empirical studies on the economic impacts of climate change and natural disasters (e.g., IPCC, 2018; Hsiang and Jina, 2014; Burke et al., 2015; IMF, 2017; Nordhaus, 2019; Kamber et al., 2013; Cashin et al., 2017; De Winne and Peersman, 2021; Kabundi et al., 2022, among many others). Our paper contributes to this research area with our novel narrative analysis on the monetary policy responses to disaster shocks. The second strand includes macroeconomic models with disaster shocks (Barro, 2006; Gabaix, 2012; Gourio, 2012; Isoré and Szczerbowicz, 2017; Fernández-Villaverde and Levintal, 2018). As discussed, our model builds on this literature by extending the model of Fernández-Villaverde and Levintal (2018) with features of disaster-prone EMDEs that are key for the analysis of monetary policy. The third strand comprises macroeconomic models for disaster-prone developing economies (Adam and Bevan, 2020; Isoré, 2018; Marto et al., 2018; Cantelmo et al., 2019). Ours, however, is the first study that analyzes monetary policy regimes in these countries. The fourth strand comprises both empirical and theoretical contributions on monetary policy in the presence of natural disaster shocks (Keen and Pakko, 2011; Fratzscher et al., 2020; Klomp, 2020; Jorda et al., 2020; McKibbin et al., 2021; Cantelmo, 2022). Our novel angle is the welfare-based ranking of alternative monetary policy rules capturing the experience of disaster-prone EMDEs.

Investigating monetary policy in the presence of weather shocks could not be more central in a global context where climate change is projected to make natural disasters even more frequent and severe (IPCC, 2018). For example, the frequency of hurricanes of category 4 or greater, are expected to increase by 39–87% over the 21st century (Knutson et al. 2013). In addition, with a few notable exceptions discussed above, the macroeconomic literature on climate-related disaster shocks has investigated important policy aspects (including investment in resilient infrastructure, pre-disaster and post-disaster donor support and insurance) from a fiscal viewpoint, almost neglecting the monetary policy angle. With the COVID-19 pandemic that left many disaster-prone countries, especially those dependent on tourism, with alarmingly high amounts of public debt and very limited fiscal space, it is even more important that monetary policy is conducted in a welfare-maximizing manner.

The remainder of the paper is structured as follows. Section 2 outlines the design of the narrative analysis and summarizes its main findings. Sections 3 and 4 present the model and its calibration, respectively. Section 5 discusses the model results, including welfare outcomes associated with alternative monetary policy rules. Section 6 provides a sensitivity analysis of the findings. Finally, Section 7 concludes. A thorough documentation of the narrative analysis is appended to the paper.

2 Narrative Analysis

In this section, we describe the methodology and report the findings of a narrative analysis on the response of monetary authorities following the occurrence of a climate-related natural disaster. We obtain the relevant information from IMF staff reports prepared after the so-called “Article IV” consultations in the year of, and one year following, the occurrence of a disaster, covering the macroeconomic and inflation performance in the aftermath of climate-related disasters, and IMF’s evaluations and advice on the monetary policy stance.1 We focus on disaster-years where annual damages were at least 1 percent of GDP, subject to staff report availability. For countries in currency unions (such as the Eastern Caribbean Currency Union), we cross-reference Article IV staff reports of the IMF mission to the union’s central bank. Our final sample consists of 34 disaster-years, that occurred in 16 disaster-prone countries from 1999 to 2017. Table 1 shows the complete list of countries and disasters used in our dataset, as well as the annual damages (as a percentage of GDP).

Table 1:

List of Disasters Used in the Narrative Analysis and Corresponding Year of IMF Article IV Staff Report

article image
Source: EM-DAT (EM-DAT: The Emergency Events Database – Universite Catholique de Louvain (UCL) – CRED, D. Guha-Sapir – www.emdat.be, Brussels, Belgium.) and Cantelmo et al. (2019). Notes:* Authors combined Article IV staff reports for the country in question, as well as the ECCU (Eastern Caribbean Currency Union). Both Article IVs are dated at the same year.

Authors combined Article IV staff reports for the country in question, as well as the ECCU (Eastern Caribbean Currency Union). The ECCU Article IV is dated a year before the country one.

Authors combined Article IV staff reports for the country in question, as well as the ECCU (Eastern Caribbean Currency Union). The ECCU Article IV is dated a year after the country one.

Dominica received IMF support (Catastrophe Containment and Relief Trust) under the financial instruments designed for these circumstances, in 2015. The Catastrophe Containment and Relief Trust (CCRT) allows the IMF to provide grants for debt relief for the poorest and most vulnerable countries hit by catastrophic natural disasters or public health disasters. The relief on debt service payments frees up additional resources to meet exceptional balance of payments needs created by the disaster and for containment and recovery. Established in February 2015 during the Ebola outbreak and modified in March 2020 in response to the COVID-19 pandemic.

This disaster led to damages <1% of GDP, while the other disaster let to damages >1% of GDP. Cumulative damages encompass both disasters.

The magnitude of the damages for this particular disaster is unreported, therefore they are excluded from the cumulative damages.

2.1 Methodology

The narrative analysis covers the macroeconomic and monetary policy performance of countries after the disaster, as well as the monetary policy tools that might have been mobilized to mitigate the negative impact that disasters had on the economy. The assessment is conducted by recording the nature of the mobilized monetary policy tools, whether policy was accommodative or tight, the appraisal of the monetary policy stance by IMF staff and/or Board of Directors, and the IMF’s advice on the monetary policy stance for the near future. Table 2 shows the complete set of questions answered to construct our dataset. Some questions relate to structural features that might change over time, such as the exchange rate regime and monetary policy independence. For example, El Salvador had its own legal tender when Hurricane Mitch struck in October 1998, but did not possess this feature when Hurricane Adrian struck in May 2005, because effective January 1, 2001, the U.S. dollar became its legal tender. Other questions are on the changes in key macroeconomic variables such as the GDP growth rate and the inflation rate, in the aftermath of the disaster. Others relate to the monetary policy response in countries where the monetary policy regime allows to mobilize it. In this respect, we classify as “independent” a monetary policy regime in which a country has full control on their monetary policy; “not independent” a regime of an economy that does not have its own legal tender or it has a hard peg; and “mixed” a regime where, although there is peg or exchange rate anchor, limited capital mobility still allows room for monetary policy. The final set of questions is on the IMF evaluation of these policy actions, and on its advice on future adjustments. The answers to these questions are especially important, because both in the literature and in policy circles, there is no consensus about how monetary policy should be conducted in the aftermath of a disaster.

Table 2:

Questions Posed to Conduct the Narrative Analysis

article image

Table 3 illustrates how the questions are answered using the example of Hurricane Iris that hit Belize on October 4, 2001. In Appendix B we document the whole process by reporting quotes extracted from the relevant IMF Article IV staff reports, for all disaster-country observations. This procedure enables us to construct a complete dataset of qualitative data.

Table 3:

Example of Narrative Analysis Documentation: Belize, 2001

article image
Sources: Authors and 2002 Article IV IMF Staff Report for Belize.

2.2 Results

The main findings of the narrative analysis are summarized in Figures 3 to 5. Figure 3 illustrates the results on the economic performance shortly after a natural disaster occurrence, as well as some features of the affected countries. Figure 4 summarizes the monetary policy stance adopted and Figure 5 presents the IMF staff appraisal and advise on monetary policy. In most cases, GDP growth declined and, often, inflation increased. Most disasters occurred in countries with at least partial control of their monetary policy, as only 12 percent of cases refer to disasters occurred in countries with dollarized economies. In addition, in non-peg countries the impact on the exchange rate was mixed, while in the other countries, the sustainability of the peg was often discussed.

Figure 3:
Figure 3:

Narrative Analysis: Impact of Natural Disasters and Features of Affected Countries

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Sources: IMF staff reports and authors’ calculations.Notes: Estimates are based on a narrative analysis of IMF staff reports on disaster-prone developing countries over the period 1999 to 2017. The analysis is restricted to weather-related natural disasters with associated damages of at least 1% of GDP (according to the EMDAT database), subject to IMF staff report availability. These criteria lead to a sample of 34 incidents that occurred in 16 countries. Please note that if we were to consider also non pegged countries, the percentage of countries that experienced an impact on their reserves would go down to 35 percent. Answers referring to Panel B come from crossing the answers to questions 1, 2 and 3. The characterization of monetary policy as being independent does not take possible fiscal dominance into account.El Salvador switched regimes in 2001 as U.S. Dollar replaced the local Colón as the legal tender.
Figure 4:
Figure 4:

Narrative Analysis: Monetary Policy Stance

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Sources: IMF staff reports and authors’ calculations.Notes: Estimates are based on a narrative analysis of IMF staff reports on disaster-prone developing countries over the period 1999 to 2017. The analysis is restricted to weather-related natural disasters with associated damages of at least 1% of GDP (according to the EMDAT database), subject to IMF staff report availability. These criteria lead to a sample of 34 incidents that occurred in 16 countries. The time horizon considered in IMF staff’s assessment of the monetary policy stance is within one year after the occurrence of each disaster. Constraints to changes in the monetary policy stance are typically hard pegs or dollarized economies. The aftermath of a disaster is defined as the period, generally shorter than one year, between the occurrence of the disaster and the IMF mission to the country. IMF Staff provide an appraisal of the MP stance adopted, and advice on the stance to adopt in the near future, with a time horizon usually not longer than one year after the completion of the IMF mission.
Figure 5:
Figure 5:

Narrative Analysis: IMF Staff Appraisal and Advice

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Sources: IMF staff reports and authors’ calculations.Notes: Estimates are based on a narrative analysis of IMF staff reports on disaster-prone developing countries over the period 1999 to 2017. The analysis is restricted to weather-related natural disasters with associated damages of at least 1% of GDP (according to the EMDAT database), subject to IMF staff report availability. These criteria lead to a sample of 34 incidents that occurred in 16 countries. The time horizon considered in IMF staff’s assessment of the monetary policy stance is within one year after the occurrence of each disaster. Constraints to changes in the monetary policy stance are typically hard pegs or dollarized economies. The aftermath of a disaster is defined as the period, generally shorter than one year, between the occurrence of the disaster and the IMF mission to the country. IMF Staff provide an appraisal of the MP stance adopted, and advice on the stance to adopt in the near future, with a time horizon usually not longer than one year after the completion of the IMF mission.

Figure 4 summarizes the monetary policy stance adopted in the aftermath of disasters, in countries where monetary policy could be mobilized. The monetary policy stance was changed in virtually all cases where there was room for maneuver. This finding highlights the perceived importance of monetary policy as a tool for mitigating the adverse effects of natural disasters. When changed, the monetary policy stance was tightened in slightly more than half of the cases (almost 56 percent of disasters), and accommodated in the remaining cases, signaling heterogeneous importance attributed to inflation and exchange rate considerations on one hand, and to output losses on the other. The main monetary policy tool utilized in the aftermath of disasters was the interest rate, but there were several cases where other policy tools, such as the money supply, where mobilized.

Figure 5 presents the IMF appraisal and advice on monetary policy.2 IMF staff and/or directors always agreed with authorities when they adopted a tight monetary policy stance, but also with loosening in a number of cases (about half of instances in which authorities adopted a loose monetary policy stance). A tight monetary policy stance is a natural consequence of concerns toward spikes in inflation and sharp exchange rate movements. However, the fact that IMF staff also appraised positively cases of accommodative monetary policy stance shows that, in certain cases, expert judgment might have underscored the importance of stimulating the economy in the aftermath of disasters, in order to avoid recession traps. Even though IMF staff did not oppose to accommodative monetary responses in their appraisal of policies adopted in the aftermath of certain disasters, there are no cases where the advice was to switch from a tight to a loose monetary policy stance in the near future, while the reverse is true. This outcome is likely due to concerns about inflation derailment and anchoring of expectations, central bank’s credibility, availability of reserves and exchange rate stabilization.

The heterogeneity in the monetary policy conduct and advice, besides the challenges triggered by natural disasters in terms of inflation, business cycle and exchange rate stabilization, raise questions on what policymakers’ priorities should be. We investigate these issues using the model outlined in the following section.

3 The model

The framework is a SOE New-Keynesian model with stochastic trend growth and disaster shocks. Households supply labor and decide on the optimal level of consumption and investment. The economy’s consumption and investment basket include domestic and imported goods, with a set up along the lines of Gali and Monacelli (2005). Firms combine capital and labor to produce a domestic good. Differently from a standard NK model, households feature Epstein-Zin preferences (Epstein and Zin, 1989), which help capture appropriately the effects of disaster risk, and disaster shocks hit the capital stock and total factor productivity as in Gourio (2012) and Fernández-Villaverde and Levintal (2018), besides impacting the demand for exports. Finally, an array of alternative Taylor-type interest rate rules captures a number of possible monetary policy regimes.

3.1 Disasters

The modeling of disasters closely follows Fernández-Villaverde and Levintal (2018). Investment, xt, is subject to quadratic adjustment costs S[xtxt1]=κ2(xtxt1z^tz^)2 as in Christiano et al. (2005), where z^t=(AtAt1)11α is the technological stochastic trend growth and At is the permanent component of productivity. It follows that the law of motion of capital is:

kt*=(1δ)kt+(1S[xtxt1])xt,(1)

with:

kt=kt1*edtθt,(2)

where kt is the actual capital stock in period t, equal to the capital stock kt1* chosen by households in period t — 1 net of a possible disaster shock, as governed by the term kt1*. In particular, the dummy variable dt takes value 1 with probability pd, in case of a disaster realization, and 0 with probability (1 — Pd) otherwise. When a disaster occurs, the capital stock falls by a quantity θt, which follows an autoregressive process:

1ogθt=(1ρθ)1ogθ¯+ρθ1ogθt1+σθεθ,t,(3)

where the random variable θt takes a log-normal distribution with average disaster size θ¯, persistence parameter ρθ, and stochastic volatility σθεθ,t.3

It is important to note that a disaster realization is a one-off event, i.e. it occurs only in one quarter (when dt = 1). Conversely, disaster risk shocks are persistent. Equation (3) implies that agents may temporarily expect the average disaster size θ¯ to be higher or lower, with ρθ governing the persistence of the risk shock.

In addition to destroying part of the capital stock, disaster shocks affect also total factor productivity (TFP), Atagg. Along similar lines as Gourio (2012) and Cantelmo (2022), TFP has both a permanent, At, and a temporary component, AtT, meaning that disasters might be followed by a partial recovery.4 The permanent component is specified as a random walk with a drift while the temporary component follows a AR(1) process:

1ogAtagg=1ogAt+1ogAtT,(4)
1ogAt=1ogAt1+ΛA+σAεA,tω(1α)dtθt,(5)
logAtT=ρAlogAt1T(1ω)(1α)dtθt,(6)

where ΛA is the steady-state TFP growth, σAεA,t, is the Gaussian component of permanent TFP and ρA is the persistence of temporary TFP. Parameter ω ∈ [0,1] governs the relative impact of disasters on the two components of TFP. Moreover, disaster variables in the two processes of TFP are rescaled by the labor share of income, (1 — α), to ensure that capital and output fall by the same proportion.

3.2 Households

The representative household’s utility reads as:

Vt1ψ=Ut1ψ+βEt(Vt+11γ)1ψ1γ,(7)

where the period-c utility Ut is defined over consumption ct and labor lt, Ut = eξtct (1 — lt) , while Vt+1 is its continuation value. Parameter γ governs risk aversion while 1/ψ^ is the inter-temporal elasticity of substitution, where ψ^=1(1+ν)(1ψ) is its inverse. As noted by Caldara et al. (2012), the importance of recursive preferences is twofold. First, they allow for a distinction between γ and ψ^.5 Second, they imply a trade-off between current and a certainty equivalent of future utility. Households therefore have preference for early (γ>ψ^) or later (γ<ψ^) resolution of uncertainty. These features are particularly appealing in our context where agents face the risk of natural disasters, which induces precautionary savings captured by the recursive structure of preferences.

Households consume a constant-elasticity-of-substitution (CES) basket (ct) of home (ctH) and foreign goods (ctF). Thus,

ct=[φ1χc(ctH)χc1χc+(1φ)1χc(ctF)χc1χc]χcχc1,(8)

where ϕ indicates the home good bias and χc > 0 is the intratemporal elasticity of substitution.

The consumption basket is the numeraire of the economy, with the unit price of this basket corresponding to:

1=[φ(ptHpt)1χc+(1φ)(ptFpt)1χc]11χC,(9)

where ptH represents the price of home goods, ptF represents the price of foreign goods, and pt is the price of the composite consumption good. The relative price of home goods will then be p˜tHptHpt. The relative price of foreign goods is stptFpt=etpt*pt, where et is the nominal exchange rate and pt* is the price level of foreign goods expressed in foreign currency. Assuming that the law of one price holds, st corresponds also to the real exchange rate, defined as the price of one unit of foreign consumption basket in units of the domestic basket.

The definition of the real exchange rate pins down the following purchasing power parity relationship linking domestic to foreign inflation:

stst1=etet1Πt*Πt,(10)

where Πtptpt1 is the gross domestic inflation rate and Πt*pt*pt1* is the gross foreign inflation rate, which is exogenous and follows an autoregressive process,

1og(Πt*Π*)=ρΠ*1og(Πt1*Π*)+εtΠ*,(11)

where ρΠ∗ is the autoregressive parameters, and εtΠ* is a mean zero, normally distributed random shock with standard deviation σty*.

Minimizing total consumption expenditures subject to the consumption basket (8) yields the following demand functions for each good:

ctH=φ(p˜tH)χcctandctF=(1φ)(st)χcct.(12)

Each period, the household’s budget constraint (in real terms) reads as:

ct+xt+bt+1pt+etbt+1*pt=wtlt+rtkt+Rt1btpt+etRt1*Ψt1bt*pt+Ft+Tt,(13)

where xt denotes investment in capital, wt is the real wage, rt is the rental rate on capital kt, Ft are profits earned from firms, Tt is a lump-sum transfer from the government, bt represents private domestic bonds which pay a gross return, Rt, and bt* are net foreign assets denominated in foreign currency paying a gross return Rt*, which is exogenous and follows an autoregressive process:

1og(Rt*R*)=ρR*1og(Rt1*R*)+εtR*,(14)

where ρR* is the autoregressive parameters, and ϵtR* is a mean zero, normally distributed random shock with standard deviation σtR*. To prevent bt* from being a unit-root process, there exists a premium for holding net foreign assets (as in Schmitt-Grohe and Uribe, 2003), Ψtψ0exp{ψ1(bt*b*)}, inversely related to the deviations of national foreign asset holdings to GDP, yt, from their steady state. While ψ0 captures the average wedge between Rt and Rt*,ψ1>0 makes the interest rate paid on foreign debt instruments elastic to net foreign asset holdings.

The household determines the optimal capital stock, kt*, which depreciates at a rate δ, and the investment xt needed to achieve it.

Optimal choices of consumption, domestic and net foreign assets, labor supply, capital stock, and investment are taken to maximize utility (7), subject to (13), and (1), thus leading to the following first-order conditions:

1=Et[Mt+1RtΠt+1],(15)
1=Et[Mt+1et+1etΨtRt*Πt+1],(16)
wt=νct1lt,(17)
qt=Et(Mt+1edt+1θt+1[rt+1+qt+1(1δ)]),(18)
1=qt[1S[xtxt1]S[xtxt1]xtxt1]++EtMt+1qt+1S[xt+1xt](xt+1xt)2.(19)

Equations (15) and (16) are the Euler equations, where Mt+1βλt+1λtVt+1ψγEt(Vt+11γ)ψγ1γ is the stochastic discount factor with Epstein-Zin preferences and λt is the Lagrange multiplier on the budget constraint (13). Equation (17) represents the marginal rate of substitution between consumption and leisure, while equations (18) and (19) define the asset price and investment decisions, respectively.

Combining equations (15) and (16) yields the uncovered interest rate parity condition, whereby the domestic and foreign nominal interest rates are equal up to the nominal exchange rate depreciation and the risk premium:

RtRt*=ΨtEt[et+1et]=ΨtEt[st+1stΠt+1Πt+1*].(20)

Similarly to private consumption, investment xt is also a CES basket of home, xtH, and foreign goods, xtF. For simplicity, the elasticity of substitution and the distributional parameter between the home and foreign components of investment are the same as in the consumption aggregator:

xt=[φ1χc(xtH)χc1χc+(1φ)1χc(xtF)χc1χc]χcχc1.(21)

Minimizing total investment expenditures subject to the consumption basket (21) yields the following demand functions for each type of investment goods:

xtH=φ(p˜tH)χcxtandxtF=(1φ)(st)χcxt.(22)

3.3 Firms

The firms’ side of the model is completely standard and borrowed from Fernández-Villaverde and Levintal (2018), except for the fact that the small-open-economy aspect needs to be taken into consideration (along the lines of Gali and Monacelli, 2005). Perfectly competitive final good producers combine i domestic intermediate goods according to

yt=(01yi,tε1)εε1,(23)

where ε is the elasticity of substitution.6 Intermediate goods producers combine labor and capital according to a Cobb-Douglas production function:

yi,t=Ataggki,tαli,t1α,(24)

where α ∈ [0,1] is the capital share of income. Intermediate firms choose inputs and prices to maximize profits Fi,t=pi,tHptyi,twi,tli,tri,tki,t, subject to the production function (24) and a Dixit-Stiglitz demand function yi,t=(pi,tHptH)εyt, and are subject to Calvo price stickiness. At the symmetric equilibrium all i firms are equal, hence the first-order conditions of the profit-maximization problem imply the following relationships:

ktlt=α1αwtrt,(25)
gt1=mctyt+θpEtMt+1[(ΠtH)χΠt+1H]εgt+11,(26)
gt2=(ΠtH)oyt+θpEtMt+1[(ΠtH)χΠt+1H]1ε[(ΠtH)O(Πt+1H)O]gt+12,(27)
εgt1=(ε1)gt2,(28)
1=θp[(Πt1H)χΠtH]1ε+(1θp)[(ΠtH)O]1ε,(29)
utp=θp[(Πt1H)χΠtH]1εut1p+(1θp)[(ΠtH)O]1ε,(30)
p˜tHmct=(11α)1α(1α)αwt1αrtαAtagg,(31)

where θp [0,1] denotes the per-period probability of not resetting the price; χ ∈ [0> 1] governs the degree of indexation to past inflation of home good prices, ΠtH=ptHpt1H;(ΠtH)O=(ptH)o is the ratio between the optimal reset price and the price of the final domestic good; mct is the marginal cost expressed in units of domestic goods; g1 and g2 are auxiliary variables; and finally υtp denotes price dispersion.

3.4 Monetary Policy

The central bank sets the interest rate according to a feedback rule, generalized as follows:

RtR=(ΠtΠ¯)γΠ(ytyt1exp(Λy))γy(etet1)γe.(32)

We explore a number of alternative monetary policy regimes in line with the experience of disaster-prone countries, analyzed in Section 2, and the literature. Each case, obtained by means of appropriate parametrization, is labeled and discussed below.7

  • 1. Flexible Inflation targeting (FIT). In this case the central bank is concerned exclusively with inflation stabilization, although temporary deviations from the inflation objective are allowed, hence inflation is stabilized at a longer horizon (see, e.g., Svensson, 2000). The larger the responsiveness (γΠ) of the nominal interest rate to inflation deviations from target ((Π¯)), the sooner inflation is brought back to target in the aftermath of shocks. Conversely, in the limiting case where γΠ is very close but above 1, the Taylor principle is satisfied, hence inflation expectations are anchored, while keeping the monetary policy stance as mild as possible:
    RtR=(ΠtΠ¯)γΠ.(33)
  • 2. Strict inflation targeting (SIT). We label strict inflation targeting the limiting case in which the responsiveness of inflation is very large (γΠ = ∞) and the central bank keeps the inflation rate constant, i.e. inflation is stabilized in the very short run (Svensson, 2000):
    RtR=(ΠtΠ¯)γΠ,γΠ=.(34)
  • 3. Hard Peg (HP). In this regime, the central bank’s objective is to keep the nominal exchange rate constant (i.e. a fixed exchange rate regime as in Benigno, 2004). In practice, this outcome can be achieved by setting a very large responsiveness of the nominal interest rate to changes in the nominal exchange rate (γe → ∞):
    RtR=(etet1)γe.(35)
  • 4. Taylor rule (TR). This rule follows the standard practice of many central banks that respond to both inflation developments and economic activity. The specific formulation is borrowed from Fernández-Villaverde and Levintal (2018) who, relative to equation (33), include also a responsiveness (γy) of the nominal interest rate to output growth:
    RtR=(ΠtΠ¯)γΠ(ytyt1exp(Λy))γy.(36)
  • 5. Exchange-rate-augmented Taylor rule (ERTR). Relative to the previous regime, this rule allows the central bank to respond also to changes in the nominal exchange rate (γe > 0), (see McCallum and Nelson, 1999, Batini et al., 2003 and Justiniano and Preston, 2010, among many others). This case captures concerns regarding the fact that depreciations may harm welfare via increases in the prices of imports:
    RtR=(ΠtΠ¯)γΠ(ytyt1exp(Λy))γy(etet1)γe.(37)

Section 6 provides robustness checks to alternative specifications of the rules listed above by allowing also for interest rate inertia, by replacing the interest rate responsiveness to CPI inflation (Πt) with a responsiveness to inflation of domestic consumption goods prices ((ΠtH)), and by targeting nominal GDP.

3.5 Equilibrium

Imports consist of the sum of purchases of foreign goods for consumption and investment,

impt=ctF+xtF=(1φ)(st)χc(ct+xt).(38)

Exports consist of the foreign demand for home goods, assumed to have an analogous algebraic expression as domestic demand, and to be subject to downward shifts when the economy is hit by natural disasters, ψddtθt, where parameter ψd governs the impact of disasters on external demand:8

expt=φ*(ptHetpT**)χc*yt*ψddtθt,(39)

where ϕ* and χc* are the foreign distributional parameter and elasticity of substitution, respectively. Aggregate foreign demand, yt*, follows an autoregressive process:

1og(yt*y*)=ρy*1og(yt1*y*)+εty*,(40)

where ρy* is the autoregressive parameter, and ϵty* is a mean zero, normally distributed random shock with standard deviation σty*.

Therefore, the resource constraint reads as follows:

p˜Hyt=ct+xt+p˜Hexptstimpt.(41)

The balance of payments equilibrium requires the current account balance to be equal to the change in net foreign assets:

ptHexptptFimpt+(Rt1*Ψt11)etbt1*=et(bt*bt1*).(42)

By using the definitions of relative prices, p˜tHptHptandstptFpt=etpt*pt, foreign inflation, Πt*pt*pt1*, and the purchasing power parity condition (10), equation (42) can be rewritten in real terms as follows:

p˜tHexptstimpt+st(Rt1*Ψt11)b˜t1*Πt*=st(b˜t*b˜t1*Πt*),(43)

where b˜t*bt*pt* denotes the real net foreign assets.

4 Calibration and Solution Method

We calibrate the model to an average disaster-prone EMDEs at a quarterly frequency. Table 4 reports the choice of all parameter values for the baseline calibration.

Table 4:

Baseline Calibration

article image

Households. The discount factor (β) is set at 0.9838, such that it yields a steady-state annual interest rate of 8.52%, as reported by Garcia-Cicco et al. (2010) for a set of emerging market economies. Moreover, this value falls also in the range considered by Shen et al. (2018) for low-income countries. As conventional in the business cycle literature, the inverse of the intertemporal elasticity of substitution, Ψ^, is calibrated to the value of 0.5, and the leisure preference parameter, v, is set at 1.1, such that agents work 1/3 of their time. Given the scant evidence on risk aversion within Epstein-Zin preferences for developing economies, we set γ = 3.8, as Gourio (2012) and Fernández-Villaverde and Levintal (2018) do for the U.S. economy.9 Some experimental evidence in countries hit by natural disasters (Cassar et al., 2017 and Cameron and Shah, 2015) suggests that their economic agents tend to exhibit a more risk averse behavior, although these findings are difficult to translate into a value of γ.10 We therefore see the calibration of risk aversion based on the U.S. economy as a lower bound for disaster-prone countries. Following Justiniano and Preston (2010), the intratemporal elasticity of substitution between the home and foreign good, χc, is set to 0.67, while the home good bias, ϕ, is set to 0.5502, in order to match the imports-to-GDP ratio of 55 percent in disaster-prone countries over the 1997–2017 sample. The average wedge between Rt and Rt* ψo, is calibrated at 1.0084 in line with a spread between the average deposit rate for disaster-prone countries and the average effective Federal Funds rate of 336 annual basis points over the same period. The interest rate elasticity to net foreign assets, ψ1, is set to 0.001, given that its presence is only necessary to eliminate the unit root that there would otherwise be in net foreign assets (see, e.g., Schmitt-Grohe and Uribe, 2003).

Foreign demand. The scaling parameter in foreign demand, ϕ*, is normalized to one, the steady-state export-to-GDP ratio, expy, is set to 0.3231, in order to match the data for disaster-prone countries over the 1997–2017 sample. The elasticity of demand, χc*, is set to 0.58, following Justiniano and Preston (2010), and the parameter governing the impact of disaster shocks on export demand, ϕd, is set equal to 0.25, to deliver an one-percent increase in the annualized CPI inflation rate in response to an average disaster shock, in line with the experience of disaster-prone countries reported in Section 2.

Firms. We follow Garcia-Cicco et al. (2010) also in setting the total capital share of income, a, to 0.32, while we set trend TFP growth, ΛA, to 0.0035, as suggested by Araujo et al. (2016). For the baseline calibration, we assume that the shock is distributed equally between the permanent and stationary components of TFP (ω = 0.5), given the uncertainty surrounding this parameter. However, we check the extent to which the results are robust to alternative choices.11 The parameter governing investment adjustment costs, κ, is set to 12, in line with the calibration of Schubert and Turnovsky (2011) for a set of developing economies. The private capital depreciation rate, δ, is borrowed from Shen et al. (2018) who set it equal to a value of 0.025. Following the calibration of Justiniano and Preston (2010) for small-open economies, the automatic price adjustment, χ, is set to 0.11, and the Calvo price stickiness parameter is set to 0.68. Lastly, the elasticity of substitution of demand faced by final good producers, ε, is set to the conventional value of 6, adopted also by Isoré and Szczerbowicz (2017) in the context of a DSGE model with natural disasters.

Monetary Policy. The inflation target parameter, Π¯, is calibrated to 1.0122 to match the average annual inflation rate for disaster-prone countries of 4.87 percent, while the steady state of foreign inflation, Π¯*, is set at 1.0053 to match the average annual U.S. inflation rate of 2.12 percent. For baseline illustrative results, the parameter governing the responsiveness of the interest rate to inflation in the Taylor rule, γΠ, is set to 1.5, a conventional value that satisfies the Taylor principal (Taylor, 1993), whereas the remaining parameters in the Taylor Rule (γy, γR, γe) are set equal zero, essentially shutting down any additional monetary policy objective besides inflation targeting. However, we activate these objectives in various policy experiments and discuss the calibration of the relevant parameters in the appropriate sections.

Disaster Shocks. Absent evidence specific for EMDEs, we calibrate the persistence of the disaster risk shock, ρθ, to 0.90, following Gourio (2012), Isoré and Szczerbowicz (2017) and Fernández-Villaverde and Levintal (2018). The standard deviation, σθ = 0.1270, matches the quarterly dispersion of damages to GDP in disaster-prone countries of 28 percent. In accordance with the evidence found for disaster-prone countries (Cantelmo et al., 2019), we set the annual disaster probability, pd, to 16.2 percent and the average loss, θ¯=0.0344, so that the average disaster destroys about 7 percent of GDP, when the the disaster affects also export demand.12

Other Shocks. We set the persistence of the temporary component of TFP affected by disaster shocks, ρA, equal to 0.71 as in Gourio (2012), while the standard deviation of the shock hitting the permanent component of TFP, σA, equal to 0.0280 to match the average for disaster-prone countries of the standard deviation of the cyclical component of the logarithm of real GDP, which amounts to 2.87 percent at an annual frequency. In order to calibrate the persistence and standard deviations of shocks to the foreign interest rate, inflation and demand, we estimate AR(1) processes for the U.S. quarterly CPI inflation rate, Federal Funds rate and cyclical components of GDP (computed with a standard HP filter). This leads to the following persistence parameters for the foreign inflation rate, ρΠ∗, the foreign interest rate, ρR*, and foreign demand, ρy*, of 0.2144, 0.8085 and 0.8751, respectively; and the following standard deviations of shocks to the the same variables, and of 0.0052, 0.0095 and 0.0023, respectively.

Solution Method. To simulate our model, we resort to Taylor projection, a solution method proposed by Levintal (2018) and Fernández-Villaverde and Levintal (2018) to solve DSGE models with rare disasters. Fernández-Villaverde and Levintal (2018) demonstrate that a Taylor projection up to third order is more accurate and generally faster to compute than perturbation methods up to a fifth order of approximation and projection methods (Smolyak collocation) up to a third order to solve a wide range of DSGE models with rare disasters. Taylor projection essentially combines the setup of standard projection methods (e.g. Judd, 1992) with approximation methods via Taylor expansions. The method yields a solution that, although not global, is possible to approximate at many points of the state-space, and this makes it accurate in dealing with large nonlinearities. These features of Taylor projection are particularly appealing for studying natural disasters within a DSGE model and motivate our choice of using a third-order Taylor projection over alternative methods.

5 Results

5.1 Effects of a Natural Disaster Shock in a Small Open Economy

We start from analyzing the effects that the realization of an average natural disaster shock has on disaster-prone small-open economies. In this subsection we present results assuming a flexible CPI inflation targeting, that is, the central bank targets CPI inflation allowing for temporary deviations from the target (alternative monetary policy regimes are presented in Subsection 5.2). As explained in Section 3, the disaster affects the stock of capital and productivity as in other contributions with closed-economy models (Gourio, 2012; Fernández-Villaverde and Levintal, 2018; Cantelmo et al., 2019), with the addition of the export demand channel (illustrated in Subsection 3.5 and analyzed in Subsection 6.1).

Figure 6:
Figure 6:

Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Notes: X-axes are in quarters. Y-axes are in percent deviations from the stochastic steady state. The stochastic steady state is obtained by simulating the model in the absence of shocks for 100 quarters.

As common in the related literature, we obtain the stochastic steady state by simulating the model in the absence of shocks for 100 quarters. Soon afterwards, the model is perturbed by a one-off disaster shock of average size and impulse response functions (IRFs) are traced for 20 quarters.

Output, consumption, investment, exports, imports, net exports and net foreign assets are non-stationary and are plotted in percent deviations from the stochastic steady state. These variables grow each period at the same growth rate as TFP. Given that disasters hit both components of TFP, the growth rate of TFP initially falls and then experiences an overshooting before gradually reverting to its pre-disaster level (Subsection 3.1). Therefore, real output and consumption fall in the aftermath of the shock and then grow, initially at a higher rate, before converging to their balanced growth path. Investment increases to rebuild the destroyed stock of capital, exports and imports fall, but the fall in exports is more pronounced than that of imports, leading to a contraction in net exports.

Given that the disaster shock affects domestic output and export demand, it acts both as a demand and as a supply shock. The disaster impacts domestic production and incomes while the export channel reduces import capacity via the balance of payment condition (equation 43). Given that the elasticity of substitution between domestic and imported goods c) is less than unity, the contraction in import demand is less than proportional than the fall in exports, which requires the real exchange rate to depreciate (shown as an increase in the figure) in order to further curb import demand, stimulate exports and induce a net inflow of capital (i.e. a fall in net foreign assets).

This real exchange rate depreciation is facilitated by an initial sharp depreciation (increase) in the nominal exchange rate, which makes CPI inflation increase. The fall in the demand for home goods causes a contraction in (sticky) home good prices. Since domestic goods inflation remains below its steady state level for a prolonged period, CPI inflation experiences an undershooting following the initial increase. Given that the central bank targets CPI inflation, the response of the monetary policy rate tracks its dynamics. However, the strength of this response clearly depends on the calibration of parameter γΠ. For example, setting γΠ = 1.1, which is the welfare-maximizing response, subject to satisfying the Taylor principle (see Figure 8 in Section 5.3), implies that the central bank accommodates the disaster shock to a large extent. In that case, the responses of CPI inflation and the monetary policy rate are close to those implied by a Taylor rule that responds also to output, while the fall of GDP is mitigated.13

Figure 7:
Figure 7:

Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Alternative Specification of the Monetary Policy Regime

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Notes: X-axes are in quarters. Y-axes are in percent deviations from the stochastic steady state. The stochastic steady state is obtained by simulating the model in the absence of shocks for 100 quarters. Bold blue lines represents the effect of an average natural disaster shock in a disaster-prone country, under the baseline assumption of flexible inflation targeting. Dashed red lines and dotted black lines represent alternative monetary policy regimes.
Figure 8:
Figure 8:

Welfare Level as a Function of Responsiveness Parameters to Inflation, Output and the Exchange Rate in the Interest-Rate Rule, under Alternative Assumptions on the Frequency and Severity of Natural Disasters Shocks

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Notes: Bold black lines represent the baseline calibration. Dashed red lines represent the case of no natural disasters, while dotted blue and green lines represent the case of more severe and more frequent natural disasters, respectively.

5.2 Differences Associated with Alternative Monetary Policy Regimes

In this subsection we analyze the impact of alternative monetary policy regimes, mimicked by the alternative interest rate rules outlined in Subsection 3.4, in the context of a natural disaster realization. Figure 7 depicts the impulse responses of key macroeconomic variables to the same average natural disaster shock analyzed in the previous subsection, under alternative assumptions on the monetary policy regime.

In Subfigure 7-(a) we compare the baseline flexible inflation targeting regime (γΠ = 1.5, γy = 0, γe = 0), in which the central bank targets only CPI inflation, but allows for temporary departures of inflation from target, with strict inflation targeting, and a hard peg. Strict inflation targeting is achieved by setting a very large interest rate responsiveness to inflation to keep it virtually constant (γΠ → ∞, γy = 0, γe = 0). A hard peg is a fixed exchange regime achieved by setting a very large interest rate responsiveness to the exchange rate (γΠ = 0, γy = 0, γe → ∞). The differential impact of these three regimes on output is limited because the natural disaster shock dominates its dynamics, but both a hard peg and strict inflation targeting magnify the GDP loss to an extent. Expectedly, the specific monetary policy regime has significant implications for nominal variables. The peg, by definition, eliminates the shock-absorbing effect of the exchange rate, thus exacerbating the recession and causing a fall in aggregate demand and inflation. The central bank accommodates the shock by lowering the policy rate but still the initial output loss is larger than under flexible inflation targeting. Strict inflation targeting requires a more prolonged increase in the interest rate to keep inflation constant. The exchange rate still depreciates, but to a smaller extent than in the flexible inflation targeting.

In Subfigure 7-(b) we compare the baseline flexible inflation targeting regime (γΠ = 1.5, γy = 0, γe = 0) with a conventional Taylor rule, whereby the central bank reacts to inflation and output (γΠ = 1.5, γy = 0.5, γe = 0), and an exchange-rate-augmented Taylor rule whereby the central bank also reacts to the exchange rate (γΠ = 1.5, γy = 0.5, γe = 0.5). Again, the differential impact of these three regimes on output is limited but the effects on nominal variables are very different. While with flexible inflation targeting monetary policy is tightened following the disaster shock, the responsiveness to output in the conventional Taylor rule leads to a monetary policy accommodation, which mildly mitigates the output contraction and leads to a stronger exchange rate depreciation and higher inflation. If the central bank is also concerned with the stability of the exchange rate, this leads to intermediate responses, between those delivered by flexible inflation targeting and a conventional Taylor rule.

5.3 Welfare Outcomes

In the previous subsections, results are based on the analysis of impulse responses to a disaster shock, conditional on monetary policy regimes. This is especially useful to highlight tradeoffs among alternative monetary policy reactions to disasters. The model, and the economies under investigation, however, are subjected by several other shocks, in addition to natural disasters. Therefore, it is informative to simulate the model with all shocks activated and to evaluate welfare outcomes.14

Table 5 reports output and inflation volatilities, welfare levels and welfare gain/losses associated with the various monetary policy regimes vis-à-vis the flexible inflation targeting regime. Output and inflation volatilities are captured by the standard deviations of the percent fluctuations of output around its trend and the CPI inflation rate, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ utility (equation 7). Finally, the consumption equivalent (C.E.) welfare gain represents the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

Table 5:

Output and Inflation Volatilities, and Welfare Levels Associated with Alternative Monetary Policy Regimes

article image
Notes: Parameters γΠ, γy and γe represent the responsiveness to inflation, output and the exchange rate, respectively, in the interest-rate rule. Output and inflation volatilities are the standard deviations of the percent fluctuations around their respective trends, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ welfare. The consumption-equivalent (C.E.) welfare gain represent the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

Under the baseline calibration, flexible inflation targeting dominates all other regimes. Relative to this regime, strict inflation targeting leads to a lower inflation volatility, a slightly higher output volatility and a welfare loss equivalent to a permanent loss in consumption of about 0.3 percent. A hard peg, by removing the shock-absorbing properties of a flexible exchange rate, is associated with higher output and inflation volatilities, and to a C.E. welfare loss of about 0.7 percent. In this sense, we extend the results of the small-open economy of Kollmann (2002) to one subject to natural disasters and our findings agree with those of Elekdag and Tuuli (2022) who find that exchange-rate flexibility mitigates the negative impact of weather shocks relative to a fixed-exchange rate regime. Both the conventional and the exchange-rate-augmented Taylor rule deliver an increase in the output and inflation volatilities and a C.E. welfare loss of about one percent relative to flexible inflation targeting.

Given that the various monetary policy regimes are based on illustrative, and possibly suboptimal, parameterizations, in Figure 8, we report the welfare level as a function of the responsiveness parameters to inflation, output and the exchange rate in the interest-rate rule. In the simulations, these parameters are changed one at a time, leaving the other two set at their baseline values (i.e. γπ = 1.5, γy = γe = 0). The same exercise is replicated also under alternative assumptions on the frequency and severity of natural disasters shocks: (i) no disaster shocks; (ii) larger damages (1.5 larger than baseline); and (iii) higher disaster frequency (1.5 higher than baseline). As expected, the no-disaster scenario delivers a higher welfare level, while higher disaster frequency or severity lead to lower welfare levels than the baseline scenario. However, regardless of the assumptions on the disaster-shock calibration, a flexible inflation targeting regime remains the welfare-optimal regime, with a small interest-rate responsiveness to inflation (γΠ ≈ 1.1) being the welfare-maximizing value. This means that the central bank can set the monetary stance at the minimum to keep inflation expectations anchored, ultimately accommodating a disaster shock to a large extent (see also discussion in Section 5.1). Positive values for the monetary policy responsiveness parameters to output or the exchange rate deliver a decrease in the level of welfare. In other words, it is optimal for the central bank to focus only on inflation stabilization, although departures of the inflation rate from target are allowed for in the aftermath of shocks. This way the central bank is able to effectively absorb both demand and supply shocks by stimulating aggregate demand and firms production, respectively, while keeping inflation under control. These results are consistent with the empirical findings of Fratzscher et al. (2020) who show that countries adopting an inflation targeting regime suffer lower output losses and milder surges in inflation than in countries adopting alternative regimes.

6 Sensitivity Analysis

6.1 Excluding the Export Demand Channel

The baseline results include the effects of the export demand channel illustrated in Subsection 3.5, capturing the typical case of tourism-dependent small islands hit by cyclones, which experience an avalanche of cancellations when these episodes ensue. Given that this model feature departs from the closest contributions to this paper in the literature, it seems appropriate to disentangle its role and assess the sensitivity of the results to its removal.

Besides the baseline case with the export channel activated (ϕd = 0.25), in Figure 9 we also present a counterfactual with no direct impact of the disaster shock on export demand (ϕd = 0). As far as real variables are concerned, the export channel of natural disasters has an amplification role but does not affect the sign of the responses. However, this feature is especially important for the effects that natural disasters have on nominal variables, particularly CPI inflation.

Figure 9:
Figure 9:

Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Alternative Assumptions on the Effect of a Natural Disaster on Export Demand

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Notes: X-axes are in quarters. Y-axes are in percent deviations from the stochastic steady state. The stochastic steady state is obtained by simulating the model in the absence of shocks for 100 quarters. Bold blue lines represents an average natural disaster shock in a disaster-prone country, assuming that natural disasters affect the demand for exports (ϕd = 0.25). Dashed red lines represents a natural disaster shock of the same intensity, assuming that the disaster has no effect on export demand (ϕd = 0).

When the export demand channel is deactivated, the disaster shock behaves as a pure supply-side shock with the decline in home good production leading to an increase in domestic inflation. With export demand effectively insulated from the disaster shock, domestic import capacity is also partially insulated. The supply side shock has an income effect and, given the relatively low elasticity of substitution between the home and imported goods, the adjustment requires an appreciation (decrease) of the real exchange rate to shift the fall in aggregate demand on the domestic good. The real appreciation is achieved by an impact appreciation (decrease) in the nominal exchange rate, which leads to a decline in CPI inflation. Since, in this case, domestic goods inflation remains above its steady state level for a prolonged period, CPI inflation experiences an overshooting following the initial decrease. The response of the monetary policy rate closely tracks that of CPI inflation.

Since, following disasters, we observe an increase in CPI inflation on average (Figure 2), it seems appropriate to leave this channel activated for the baseline calibration. However, given the empirical heterogeneity in the responses of CPI inflation, monetary policy and the real exchange rate in the aftermath of disasters (documented in Section 2), the intensity of the export demand channel of disaster shocks (captured by parameter ϕd) represents an effective lever to align responses of these key variables to the experience of specific countries and/or disasters.

Table 6 reports output and inflation volatilities, welfare levels and welfare gain/losses associated with the various monetary policy regimes vis-à-vis the flexible inflation targeting regime, when the export demand channel is deactivated (ϕd = 0). Relative to the baseline case, reported in Table 5, the welfare-based ranking of the various regimes remains unaltered, with flexible inflation targeting dominating all other cases.

Table 6:

Output and Inflation Volatilities, and Welfare Levels Associated with Alternative Monetary Policy Regimes–No Export Demand Channel (ϕd = 0)

article image
Notes: Parameters γΠ, γy and γe represent the responsiveness to inflation, output and the exchange rate, respectively, in the interest-rate rule. Output and inflation volatilities are the standard deviations of the percent fluctuations around their respective trends, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ welfare. The consumption-equivalent (C.E.) welfare gain represent the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

6.2 Excluding one Shock at a Time

Another sensitivity experiment worth conducting is switching off one shock at a time, while keeping all other shocks activated (including natural disaster shocks) and computing welfare outcomes across alternative monetary policy regimes. This exercise is meant to rule out that the results presented earlier in the paper, hinge on the presence of one specific shock. As shown in Table 7, irrespective of the shock being deactivated, the flexible inflation targeting regime continues to dominate all other regimes. The welfare ranking among the other regimes changes to an extent when the foreign interest rate shock or the TFP shock are excluded, leaving the bottom line of the analysis unaltered, i.e. that flexible inflation targeting is the welfare maximizing regime.

Table 7:

Output and Inflation Volatilities, and Welfare Levels Associated with Alternative Monetary Policy Regimes–Excluding One Shock at a Time

article image
Notes: Parameters γΠ, γy and γe represent the responsiveness to inflation, output and the exchange rate, respectively, in the interest-rate rule. Output and inflation volatilities are the standard deviations of the percent fluctuations around their respective trends, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ welfare. The consumption-equivalent (C.E.) welfare gain represent the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

6.3 CPI Inflation Targeting versus Domestic Inflation Targeting

We now analyze how sensitive our results are to the measure of inflation targeted by the central bank. Specifically, we replace CPI inflation (Πt) with domestic inflation (ΠtH) in each monetary policy rule. We start by assessing the impulse responses to an average natural disaster shock in Figure 10. Relative to the baseline, where CPI inflation is targeted (blue-solid lines), targeting domestic inflation (red-dashed lines) has the obvious effect that the latter is stabilized while the former is allowed to increase. This is reflected in the opposite response of the central bank rate, which is lowered to avoid the fall in domestic inflation. The nominal exchange rate increases more than under CPI inflation targeting. However, in real terms the exchange rate appreciates only slightly more than in the baseline. As a result, net exports only marginally deteriorate but, given the monetary policy accommodation, the initial fall in output is reduced.

Figure 10:
Figure 10:

Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Alternative Measures of Inflation in the Monetary Policy Rule

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Notes: X-axes are in quarters. Y-axes are in percent deviations from the stochastic steady state. The stochastic steady state is obtained by simulating the model in the absence of shocks for 100 quarters. Bold blue lines represents an average natural disaster shock in a disaster-prone country, assuming that the central bank targets CPI inflation. Dashed red lines represents a natural disaster shock of the same intensity, assuming that the central bank targets domestic inflation.

Next, we analyze the welfare properties of the monetary policy regimes when the central bank targets domestic inflation. Results are reported in Table 8. In general, both output and inflation volatilities are reduced while welfare level is higher, relative to targeting CPI inflation.15 Welfare losses relative to FIT are likewise smaller, except for the case of a hard peg. Therefore, targeting domestic inflation improves welfare relative to targeting CPI inflation, which is a result consistent with Gali and Monacelli (2005). Crucially, the welfare ranking is preserved under the different measures of inflation to target, implying that FIT is still superior to the alternative monetary policy regimes.

Table 8:

Output and Inflation Volatilities, and Welfare Levels Associated with Alternative Monetary Policy Regimes–Domestic Inflation Targeting

Excluding the foreign inflation shock

article image
Notes: Parameters γΠ, γy and γe represent the responsiveness to inflation, output and the exchange rate, respectively, in the interest-rate rule. Output and inflation volatilities are the standard deviations of the percent fluctuations around their respective trends, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ welfare. The consumption-equivalent (C.E.) welfare gain represent the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

6.4 Nominal GDP Targeting

In this subsection, we assess the properties of nominal GDP targeting (NGT). This regime has received attention in the literature on optimal monetary policy, although no central banks has yet attempted to follow such a strategy. Some studies (McCallum and Nelson, 1999, Garin et al., 2016, Bullard and Singh, 2020 and McKibbin et al., 2021, among others) argue that NGT offers several advantages relative to inflation targeting. First, by targeting the growth rate of nominal GDP, it requires knowledge of easily observable variables, instead of, e.g. the output gap. Second, it does not suffer from indeterminacy issues because, in the long run, NGT is equivalent to price level targeting, which supports a determinate equilibrium for any level of trend inflation. Third, McKibbin et al. (2021) argue that, since climate change will increase the variability of inflation and output because more supply shocks will occur due to disaster strikes, NGT can be more effective than other alternatives at stabilizing the economy. However, these contributions generally neglect the effects of NGT on exchange rate dynamics hence their results do not necessarily extend to a small-open-economy setting. Moreover, Jensen (2002) and Billi (2017) show that the desirability of NGT arises only in the presence of supply shocks, i.e. when the central bank faces a trade-off between stabilizing inflation and output. Since in our setting, there are both demand and supply shocks, it is worth exploring whether NGT is welfare improving relative to other regimes or not.

We follow Garin et al. (2016) in choosing an appropriate parametrization of the Taylor rule to obtain NGT:

RtR=(ΠtΠ¯)γΠ(ytyt1exp(Λy))γy,γΠ=,,γy=.(44)

In Figure 11, we compare the baseline flexible inflation targeting regime (γΠ = 1.5, γy = 0, γe = 0) with nominal GDP targeting (γΠ → ∞, γy → ∞, γe = 0). By targeting the growth of nominal GDP, this regime is very effective at mitigating the output collapse in the aftermath of the disaster realization. This outcome is achieved through an accommodating monetary policy, a large exchange rate depreciation and a spike in inflation, which then returns to its steady state, essentially implying a shift in the price level.

Figure 11:
Figure 11:

Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Inflation vs Nominal GDP Targeting

Citation: IMF Working Papers 2022, 067; 10.5089/9798400204371.001.A001

Notes: X-axes are in quarters. Y-axes are in percent deviations from the stochastic steady state. The stochastic steady state is obtained by simulating the model in the absence of shocks for 100 quarters. Bold blue lines represents the effect of an average natural disaster shock in a disaster-prone country under the baseline assumption of flexible inflation targeting. Dashed red lines the effect of an average natural disaster shock in a disaster-prone country under nominal GDP targeting.

Table 9 compares welfare under the two regimes. We find that, by substantially increasing both output and inflation volatilities, NGT is suboptimal relative to FIT. One reason behind this results is that, as shown by Figure 11, NGTentails too large shifts in the exchange rate and hence of inflation.

Table 9:

Output and Inflation Volatilities, and Welfare Levels Associated with Alternative Monetary Policy Regimes–Nomical GDP Targeting

article image
Notes: Parameters γΠ, γy and γe represent the responsiveness to inflation, output and the exchange rate, respectively, in the interest-rate rule. Output and inflation volatilities are the standard deviations of the percent fluctuations around their respective trends, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ welfare. The consumption-equivalent (C.E.) welfare gain represent the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

6.5 Alternative Modeling Assumptions

Our final sensitivity checks concern specific modeling assumptions. In particular, we assess welfare under the alternative monetary policy regimes and: (i) CRRA utility function, whereby risk aversion (γ) equals the inverse of the elasticity of intertemporal substituion (Ψ^) and the role of risk is dampened; (ii) more permanent or transitory effects of disasters on TFP by setting ω to 0.75 and 0.25, respectively (relative to the baseline calibration, ω = 0.50); (iii) inertial interest rate rule, with a smoothing parameter γR = 0.80.

Table 10 reports the results under each alternative modeling assumption. The welfare ranking of monetary policy strategies carries through the various modifications hence FIT remains superior to the alternatives. However, a few remarks are in order. First, employing a CRRA utility function does not dramatically alter volatilities but increases welfare under all rules and reduces the welfare losses relative to FIT. Underestimating welfare costs of natural disasters with CRRA utility is also highlighted by Douenne (2020).16 Consistently, since our baseline calibration of risk aversion (i.e. γ = 3.8) already likely entails underestimating the welfare effects of natural disasters on disaster-prone countries, further reducing it would probably miss much of these effects. Second, even when assuming more permanent or transitory effects of disaster shocks on TFP, the flexible inflation targeting regime is the welfare maximizing policy. Finally, adding the interest rate inertia in the monetary policy rule mostly reduces the volatility of inflation and slightly increases welfare relative to the baseline case of no-interest rate smoothing, a result in line with the literature (see, e.g., Schmitt-Grohé and Uribe, 2007). However, the welfare ranking of the various regimes remains unaltered.

Table 10:

Output and Inflation Volatilities, and Welfare Levels Associated with Alternative Monetary Policy Regimes–Alternative Modeling Assumptions

article image
Notes: Parameters γΠ, γy and γe represent the responsiveness to inflation, output and the exchange rate, respectively, in the interest-rate rule. Output and inflation volatilities are the standard deviations of the percent fluctuations around their respective trends, simulated for 900 quarters, after running the model in the absence of shocks for 100 quarters. The welfare level is the average of the simulated recursive definition of households’ welfare. The consumption-equivalent (C.E.) welfare gain represent the permanent increase in consumption (in percent) necessary to make agents as well off as in the flexible inflation targeting regime (with a minus sign representing a welfare loss).

7 Conclusions

In this paper we assessed the role of monetary policy in emerging and developing economies where climate-related natural disasters are a major source of macroeconomic fluctuations.

First, we conducted a narrative analysis documenting the macroeconomic effects of natural disasters, the monetary policy regimes in place, central banks’ responses and IMF policy advices. This analysis shows that natural disasters are typically followed by a decline in output and often by an increase in inflation. If there is at least some degree of monetary policy independence, central banks generally change their monetary policy stance in the aftermath of disasters. While monetary policy is commonly tightened, there is a sizable number of cases in which it is accommodated. Policy appraisals and advice by IMF staff have also been mixed, possibly underscoring that while tightening is a direct consequence of concerns toward inflation and sharp currency movements, stimulating economic activity has been prioritized in certain cases.

We then compared these empirical facts with simulations obtained with a small-open-economy New Keynesian model augmented with disaster shocks, in which we laid out alternative monetary policy regimes and evaluated their welfare outcomes. The model analysis demonstrates that, from a welfare standpoint, a flexible inflation targeting regime—whereby inflation can depart temporarily from target—is superior both to extreme regimes, such as strict inflation targeting or hard pegs, and to hybrid regimes in which monetary policy reacts also to output and the exchange rate, besides inflation. In other words, even in the presence of natural disaster shocks, it remains suboptimal for central banks to directly target the exchange rate or economic activity. Under flexible inflation targeting, welfare is maximized by keeping the interest rate response at the lowest level needed to keep inflation expectations anchored while minimizing the impact on real activity. These findings underscore the importance of reforms geared at strengthening the monetary policy transmission mechanism and the credibility of central banks.

While monetary policy is not a substitute for structural and financial climate adaptation policies, welfare losses from ill-devised monetary policy rules may be sizable and may compound with those deriving from the devastating impacts of disasters. While focusing on monetary policy, this paper abstracts from fiscal responses or donor support, which we investigated in previous research (Cantelmo et al., 2019), and does not consider monetary-fiscal policy interactions that may affect welfare, which we leave for future research.

References

  • Adam, C. and Bevan, D. (2020). Tropical cyclones and post-disaster reconstruction of public infrastructure in developing countries. Economic Modelling, 93:8299.

    • Search Google Scholar
    • Export Citation
  • Araujo, J. D., Li, B. G., Poplawski-Ribeiro, M., and Zanna, L.-F. (2016). Current account norms in natural resource rich and capital scarce economies. Journal of Development Economics, 120:144156.

    • Search Google Scholar
    • Export Citation
  • Barro, R. J. (2006). Rare Disasters and Asset Markets in the Twentieth Century. The Quarterly Journal of Ergonomics, 121(3):823866.

  • Barro, R. J. (2009). Rare disasters, asset prices, and welfare costs. American Economic Review, 99(1):24364.

  • Barro, R. J. (2015). Environmental protection, rare disasters and discount rates. Economica, 82(325):123.

  • Batini, N., Harrison, R., and Millard, S. P. (2003). Monetary policy rules for an open economy. Journal of Economic Dynamics and Control, 27(11):20592094.

    • Search Google Scholar
    • Export Citation
  • Benigno, G. (2004). Real exchange rate persistence and monetary policy rules. Journal of Monetary Economics, 51(3):473502.

  • Billi, R. M. (2017). A note on nominal GDP targeting and the zero lower bound. Macroeconomic Dynamics, 21(8):2138a2157.

  • Brown, P., Daigneault, A. J., Tjernstrom, E., and Zou, W. (2018). Natural disasters, social protection, and risk perceptions. World Development, 104:310325.

    • Search Google Scholar
    • Export Citation
  • Bullard, J. and Singh, A. (2020). Nominal GDP targeting with heterogeneous labor supply. Journal of Money, Credit and Banking, 52(1):3777.

    • Search Google Scholar
    • Export Citation
  • Burke, M., Solomon M. H., and Miguel, E. (2015). Global non-linear effect of temperature on economic production. Nature, 527(15725).

  • Caldara, D., Fernandez, J., Rubio-Ramirez, J., and Yao, W. (2012). Computing DSGE models with recursive preferences and stochastic volatility. Review of Economic Dynamics, 15(2):188206.

    • Search Google Scholar
    • Export Citation
  • Cameron, L. and Shah, M. (2015). Risk-taking behavior in the wake of natural disasters. Journal of Human Resources, 50(2):484515.

  • Cantelmo, A. (2022). Rare disasters, the natural interest rate and monetary policy. Oxford Bulletin of Economics and Statistics.

  • Cantelmo, A., Melina, G., and Papageorgiou, C. (2019). Macroeconomic Outcomes in Disaster-Prone Countries. IMF Working Papers 19/217, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Cashin, P., Mohaddes, K., and Raissi, M. (2017). Fair weather or foul? The macroeconomic effects of el nino. Journal of International Economics, 106:3754.

    • Search Google Scholar
    • Export Citation
  • Cassar, A., Healy, A., and von Kessler, C. (2017). Trust, risk, and time preferences after a natural disaster: experimental evidence from Thailand. World Development, 94(C):90105.

    • Search Google Scholar
    • Export Citation
  • Christiano, L. J., Eichenbaum, M., and Evans, C. L. (2005). Nominal rigidities and the dynamic effects of a shock to monetary policy. Journal of Political Economy, 113(1):145.

    • Search Google Scholar
    • Export Citation
  • Dang, D. A. (2012). On the sources of risk preferences in rural Vietnam. MPRA Paper 38738, University Library of Munich, Germany.

  • De Winne, J. and Peersman, G. (2021). The adverse consequences of global harvest and weather disruptions on economic activity. Nature Climate Change, 11(8):665672.

    • Search Google Scholar
    • Export Citation
  • Douenne, T. (2020). Disaster risks, disaster strikes, and economic growth: the role of preferences. Review of Economic Dynamics, 38:251272.

    • Search Google Scholar
    • Export Citation
  • Elekdag, S. and Tuuli, M. (2022). Weather shocks and exchange rate flexibility. IMF Working Papers 22/XX, International Monetary Fund.

  • Epstein, L. G. and Zin, S. E. (1989). Substitution, risk aversion, and the temporal behavior of consumption and asset returns: A theoretical framework. Econometrica, 57(4):937969.

    • Search Google Scholar
    • Export Citation
  • Fernandez-Villaverde, J. and Levintal, O. (2018). Solution methods for models with rare disasters. Quantitative Economics, 9(2):903944.

    • Search Google Scholar
    • Export Citation
  • Fiala, O. (2017). Experiencing Natural Disasters: How This Influences Risk Aversion and Trust, pages 4383. Springer International Publishing, Cham.

    • Search Google Scholar
    • Export Citation
  • Fratzscher, M., Grosse-Steffen, C., and Rieth, M. (2020). Inflation targeting as a shock absorber. Journal of International Economics, page 103308.

    • Search Google Scholar
    • Export Citation
  • Gabaix, X. (2012). Variable rare disasters: An exactly solved framework for ten puzzles in macro-finance. The Quarterly Journal of Economics, 127(2):645700.

    • Search Google Scholar
    • Export Citation
  • Gali, J. A. and Monacelli, T. (2005). Monetary policy and exchange rate volatility in a small open economy. Review of Economic Studies, 72(3):707734.

    • Search Google Scholar
    • Export Citation
  • Garcia-Cicco, J., Pancrazi, R., and Uribe, M. (2010). Real business cycles in emerging countries? American Economic Review, 100(5):251031.

    • Search Google Scholar
    • Export Citation
  • Garin, J., Lester, R., and Sims, E. (2016). On the desirability of nominal GDP targeting. Journal of Economic Dynamics and Control, 69(C):2144.

    • Search Google Scholar
    • Export Citation
  • Gourio, F. (2012). Disaster risk and business cycles. American Economic Review, 102(6):273466.

  • Hsiang, S. M. and Jina, A. S. (2014). The causal effect of environmental catastrophe on long-run economic growth: Evidence from 6,700 cyclones. NBER Working Papers 20352, National Bureau of Economic Research, Inc.

    • Search Google Scholar
    • Export Citation
  • Intergovernmental Panel on Climate Change (2018). Special Report: Global Warming of 1.5 C. Technical report.

  • International Monetary Fund (2017). The effects of weather shocks on economic activity: How can low income countries cope? World Economic Outlook, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Isore, M. (2018). Changes in Natural Disaster Risk: Macroeconomic Responses in Selected Latin American Countries. Economies, 6(1):112.

    • Search Google Scholar
    • Export Citation
  • Isore, M. and Szczerbowicz, U. (2017). Disaster risk and preference shifts in a New Keynesian model. Journal of Economic Dynamics and Control, 79(C):97125.

    • Search Google Scholar
    • Export Citation
  • Jensen, H. (2002). Targeting nominal income growth or inflation? American Economic Review, 92(4):928956.

  • Jorda, O., Sing, S. R., and Taylor, A. M. (2020). Longer-run economic consequences of pandemics. NBER Working Papers 26934, National Bureau of Economic Research, Inc.

    • Search Google Scholar
    • Export Citation
  • Judd, K. L. (1992). Projection methods for solving aggregate growth models. Journal of Economic Theory, 58(2):410452.

  • Justiniano, A. and Preston, B. (2010). Can structural small open-economy models account for the influence of foreign disturbances? Journal of International Economics, 81(1):6174.

    • Search Google Scholar
    • Export Citation
  • Kabundi, A., Mlachila, M., and Yao, J. (2022). How persistent are climate-related price shocks? IMF Working Papers 22/XX, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Kamber, G., McDonald, C., Price, G., et al. (2013). Drying out: Investigating the economic effects of drought in New Zealand. Technical report, Reserve Bank of New Zealand Wellington.

    • Search Google Scholar
    • Export Citation
  • Keen, B. D. and Pakko, M. R. (2011). Monetary policy and natural disasters in a DSGE model. Southern Economic Journal, 77(4):973990.

  • Klomp, J. (2020). Do natural disasters affect monetary policy? A quasi-experiment of earthquakes. Journal of Macroeconomics, 64:103164.

  • Knutson, T. R., Sirutis, J. J., Vecchi, G. A., Garner, S., Zhao, M., Kim, H.-S., Bender, M., Tuleya, R. E., Held, I. M., and Villarini, G. (2013). Dynamical downscaling projections of twenty-first-century atlantic hurricane activity: CMIP3 and CMIP5 model-based scenarios. Journal of Climate, 26(17):65916617.

    • Search Google Scholar
    • Export Citation
  • Kollmann, R. (2002). Monetary policy rules in the open economy: Effects on welfare and business cycles. Journal of Monetary Economics, 49(5):9891015.

    • Search Google Scholar
    • Export Citation
  • Levintal, O. (2018). Taylor projection: A new solution method for dynamic general equilibrium models. International Economic Review, 59(3):13451373.

    • Search Google Scholar
    • Export Citation
  • Marto, R., Papageorgiou, C., and Klyuev, V. (2018). Building resilience to natural disasters: An application to small developing states. Journal of Development Economics, 135(C):574586.

    • Search Google Scholar
    • Export Citation
  • McCallum, B. T. and Nelson, E. (1999). Nominal income targeting in an open-economy optimizing model. Journal of Monetary Economics, 43(3):553578.

    • Search Google Scholar
    • Export Citation
  • McKibbin, W. J., Morris, A. C., Wilcoxen, P. J., and Panton, A. J. (2021). Climate change and monetary policy: Issues for policy design and modelling. Oxford Review of Economic Policy, 36(3):579603.

    • Search Google Scholar
    • Export Citation
  • Nakamura, E., Steinsson, J., Barro, R., and Ursua, J. (2013). Crises and recoveries in an empirical model of consumption disasters. American Economic Journal: Macroeconomics, 5(3):3574.

    • Search Google Scholar
    • Export Citation
  • Nordhaus, W. (2019). Climate change: the ultimate challenge for economics. American Economic Review, 109(6):19912014.

  • Okafor, L. E., Adeola, O., and Folarin, O. (2021). Natural disasters, trade openness and international tourism: the role of income levels across countries. Tourism Recreation Research, 0(0):119.

    • Search Google Scholar
    • Export Citation
  • Rossello, J., Becken, S., and Santana-Gallego, M. (2020). The effects of natural disasters on international tourism: A global analysis. Tourism Management, 79:104080.

    • Search Google Scholar
    • Export Citation
  • Schmitt-Grohe, S. and Uribe, M. (2003). Closing small open economy models. Journal of International Economics, 61(1):163185.

  • Schmitt-Grohe, S. and Uribe, M. (2007). Optimal simple and implementable monetary and fiscal rules. Journal of Monetary Economics, 54(6):17021725.

    • Search Google Scholar
    • Export Citation
  • Schubert, S. F. and Turnovsky, S. J. (2011). The impact of oil prices on an oil-importing developing economy. Journal of Development Economics, 94(1):1829.

    • Search Google Scholar
    • Export Citation
  • Shen, W., Yang, S.-C. S., and Zanna, L.-F. (2018). Government spending effects in low- income countries. Journal of Development Economics, 133:201219.

    • Search Google Scholar
    • Export Citation
  • Svensson, L. E. O. (2000). Open-economy inflation targeting. Journal of International Economics, 50(1):155183.

  • Taylor, J. B. (1993). Discretion versus policy rules in practice. In Carnegie-Rochester conference series on public policy, volume 39, pages 195214. Elsevier.

    • Search Google Scholar
    • Export Citation
  • van den Berg, M., Fort, R., and Burger, K. (2009). Natural hazards and risk aversion: experimental evidence from Latin America. Technical report.

    • Search Google Scholar
    • Export Citation

Appendix

A List of Disaster-Prone Countries

Table A.1:

Disaster-Prone Countries: Fourth Quartile (75%-100%) of the Annual Probability Distribution of Natural Disasters.

article image
Source: Cantelmo et al. (2019). Notes: Countries are ordered by the annual probability of a natural disaster per 1000 squared kilometers over the sample 1998–2017. EM-DAT provides damages in US dollars. Damages in percent of GDP are obtained dividing damages by GDP of the year of the event. Damages (% of GDP) are computed for each country by using data for each single event over the sample 1998–2017. Small economies comprise small states and low-income countries.Denotes Small states which are countries with a population below 1.5 million that are not advanced economies or high-income oil exporting countries (IMF). ∗∗ Denotes Low-income-countries which are countries with a GNI per capita below $995 in 2017 (World Bank).

B Narrative Analysis Documentation

Table B.1:

Narrative Analysis Documentation

article image
article image
article image
article image
article image
article image
article image
article image
article image
article image
article image
article image
article image
Note: Authors’ comments are provided in square brackets.
1

After downloading all the relevant archived IMF staff reports (pairs of disaster occurrences-countries), we read the documents to answer the survey questions covered in Subsection 2.1. Article IV consultations are part of the IMF’s country surveillance, an ongoing process that culminates in regular (usually annual) comprehensive consultations with individual member countries. These consultations are known as Article IV consultations because they are required by Article IV of the IMF’s Articles of Agreement. During an Article IV consultation, an IMF team of economists visits a country to assess economic and financial developments and discusses the country’s economic and financial policies with government and central bank officials. Due to staff report availability, in a few cases we base our answers on consultations occurred two (El Salvador, 2011; Micronesia, 2015 and Solomon Islands, 2014) and three years (Samoa, 2012) after the disaster.

2

The results of this narrative analysis were discussed with IMF mission chiefs and desk economists covering about a third of the countries in our sample. These teams found that the findings are generally in line with their experience in the field, and stressed the monetary policy dilemmas often posed by natural disasters. Some teams noted the omission of the fiscal response or donor support in our study, as well as the exclusion of lower impact disasters (with damages smaller than 1 percent of GDP) in the sample, which might compound and affect macroeconomic outcomes due to their high frequency. On the former, Cantelmo et al. (2019) conducted earlier research, hence it was a deliberate choice of focusing this paper on the understudied issue of monetary policy responses in disaster-prone countries. On the latter, the inclusion of higher frequency/lower impact disasters in the narrative analysis would prove unfeasible, as these largely go unreported in IMF staff reports, and a change of the monetary policy stance typically requires events of greater magnitude.

3

Epidemics and pandemics are expected to work differently because they are not associated with a destruction of capital.

4

See discussion in Section 4. This specification nests that of Isoré and Szczerbowicz (2017) and Fernández-Villaverde and Levintal (2018). The latter assumes that only the permanent component of TFP is subject to disasters hence, by construction, disasters have permanent effects.

5

The more standard case of expected utility can be achieved by setting γ=ψ^.

6

For simplicity the model abstracts from imported intermediate goods, although the capital stock, owned by households, is built with investment goods that are partly imported. For a setting featuring imported intermediate goods explicitly, see Justiniano and Preston (2010), among others. Moreover, the setting is standard in that monopolistic competition is at the level of intermediate firms, which are distinct from final goods producers to allow for Calvo price stickiness.

7

These monetary policy rules imply that the central bank has acquired sufficient credibility and a functioning transmission mechanism between the monetary policy rate to interest rates that affect borrowing and lending, which may be weak, especially in low-income countries.

8

This channel captures, e.g., the fall in external demand for exports in the tourism sector when small island countries are impacted by hurricanes or similar natural disasters and the rise in trade barriers as crucial mobility infrastructure (such as harbors and airports) is disrupted. Empirical evidence (e.g., Rossello et al., 2020, among others) finds that events such as tsunamis, floods and volcanic eruptions generally reduce tourist arrivals and may divert tourist flows from one destination to another. This effect may be persistent, especially in low-income countries (Okafor et al., 2021).

9

Values of risk aversion between 3 and 4 are needed to replicate the average equity premium, see Barro (2009; 2015) and Gourio (2012).

10

See also van den Berg et al. (2009), Dang (2012) and Brown et al. (2018). Fiala (2017) reviews this evidence in more detail and reports also some contrasting results. Cantelmo (2022) shows that sufficiently temporary higher risk aversion in the aftermath of disasters might generate large demand-side in addition to supply-side effects.

11

The extreme cases of ω = 0 and ω = 1 imply that disasters only have a temporary or a permanent effect, respectively. Hsiang and Jina (2014) estimate that tropical cyclones have a highly persistent effect on the growth rate and reject hypothesis of “creative destruction” or “build-back better.” Moreover, a peculiarity of disaster-prone countries is that they are subject to recurrent natural disasters, hence even if a single disaster alone would not be very persistent, when more events compound the effects might become virtually permanent. With a focus on other types of disasters, Nakamura et al. (2013) show that disasters are followed by partial recoveries, hence with a temporary higher growth rate of output after the disaster relative to the pre-disaster growth rate. By appealing to their evidence, our baseline calibration assumes that natural disasters have both a short-run and a long-run impact on productivity, hence the aftermath of disasters is characterized by faster growth and a partial recovery.

12

Note that θ¯=log(1Δ), where ∆ is the loss in terms of GDP.

13

Impulse responses are available upon request.

14

In Section 6.2 we show that the results are not driven by a specific shock.

15

Obviously, volatilities and welfare level are unaffected in case of hard peg. However, the consumption equivalent gain changes because welfare changes in the FIT case.

16

In particular, Douenne (2020) shows that lowering risk aversion to equal the inverse of the elasticity of intertemporal substituion leads to underestimate the welfare costs of natural disasters. Conversely, increasing the inverse of the elasticity of intertemporal substituion to equal risk aversion leads to conclude that natural disasters foster growth. All in all, these two parameters have empirically very different values hence Epstein-Zin preferences are more appropriate for the quantitative assessment of disasters.

  • Collapse
  • Expand
Monetary Policy in Disaster-Prone Developing Countries
Author:
Mr. Alessandro Cantelmo
,
Nikos Fatouros
,
Mr. Giovanni Melina
, and
Mr. Chris Papageorgiou
  • View in gallery
    Figure 1:

    Change in Annual GDP Growth Rate in the Aftermath of a Large Macroeconomic Shock

  • View in gallery
    Figure 2:

    Distribution of Changes in Key Macroeconomic Variables in the Aftermath of Natural Disasters in Disaster-Prone Countries

  • View in gallery
    Figure 3:

    Narrative Analysis: Impact of Natural Disasters and Features of Affected Countries

  • View in gallery
    Figure 4:

    Narrative Analysis: Monetary Policy Stance

  • View in gallery
    Figure 5:

    Narrative Analysis: IMF Staff Appraisal and Advice

  • View in gallery
    Figure 6:

    Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country

  • View in gallery
    Figure 7:

    Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Alternative Specification of the Monetary Policy Regime

  • View in gallery
    Figure 8:

    Welfare Level as a Function of Responsiveness Parameters to Inflation, Output and the Exchange Rate in the Interest-Rate Rule, under Alternative Assumptions on the Frequency and Severity of Natural Disasters Shocks

  • View in gallery
    Figure 9:

    Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Alternative Assumptions on the Effect of a Natural Disaster on Export Demand

  • View in gallery
    Figure 10:

    Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Alternative Measures of Inflation in the Monetary Policy Rule

  • View in gallery
    Figure 11:

    Impulse Responses of Selected Macroeconomic Variables to an Average Natural Disaster Shock in a Disaster-Prone Country, under Inflation vs Nominal GDP Targeting