A. Identifying staples in SSA

The staple foods are identified based on their average contribution to individual diets. Our analysis focuses on staple foods across a diverse group of 15 Sub-Saharan African countries (Table 1). This set of countries—home to approximately 70 percent of the population in the region—includes low-income countries, as well as frontier and emerging economies.⁵ Due to data limitations, a larger set of Sub-Saharan countries was not included in the analysis.

Using consumption data from FAO food balance sheets, the most consumed staple foods are selected by averaging the daily per capita caloric intake of each staple across countries. A typical consumption basket in the region is broadly made up of two types of staples: cereals and starchy roots. On average, cereals—such as maize, rice, and wheat—and starchy roots—such as cassava, sweet potatoes, and potatoes—account for almost two thirds of daily caloric intake. The other third of daily food consumption is mainly comprised of vegetable oils, sugar, fruits, meat, and pulses. The five most consumed staples in SSA are maize, rice, cassava, wheat, and palm oil, which contribute to about 54 percent of the average daily per capita caloric intake (Figure 1).⁶ Within the region, maize—19.7 percent of the daily per capita caloric intake—is the top staple in Eastern and Southern Africa, while rice—11.4 percent of the daily per capita caloric intake—is widely consumed in Western Africa. The consumption of cassava, wheat, and palm oil—which account for around 10.5, 8.3, and 3.5 percent of the daily per capita caloric intake, respectively—seems more homogeneous across the region, albeit with some differences between certain countries. For instance, the top 5 staples make up more than 65 percent of total daily caloric contribution in Angola and Mozambique, whereas they account for less than 40 percent in Ethiopia, Namibia, or Rwanda, reflecting differences in diet composition and missing data in some countries. In Ethiopia, other cereals (15.3 percent) such as teff, sorghum (9.6 percent), and roots (6.0 percent) are largely consumed staples alongside maize (19.3 percent) and wheat (13.3 percent). In Namibia, wheat (17.3 percent) and maize (16.9 percent) are the top 2 staples, followed by roots (11.5 percent), sugar (9.1 percent), and milk (7.8 percent). In Rwanda, the top staples are beans (14.6 percent), cassava (11.8 percent), sweet potatoes (10.1 percent), maize (8.8 percent) and potatoes (8.6 percent). The percent caloric contribution of each staple to the daily diet has remained virtually unchanged over the last 5 years of the sample period, suggesting a limited temporal substitution between these top five staples in the food consumption basket.⁷

View Full Size

Daily caloric contribution and net import dependence of top 5 staples

Citation: IMF Working Papers 2022, 135; 10.5089/9798400216190.001.A001

Source: FAOSTAT Food Balances and authors’ calculations.

• Download Figure
• Download figure as PowerPoint slide

There is substantial cross-country heterogeneity in the domestic production and net import dependence of staple foods.⁸ For example, Côte d’Ivoire is a net exporter of palm oil—with a net import dependence of -73.5 percent, whereas Ghana is a net importer—with a net import dependence of 36.4 percent (Table 2). About three-quarters of Ghana’s annual imports of palm oil come from Malaysia and Indonesia—two leading global producers—and only 13 percent from Côte d’Ivoire.⁹ Despite this heterogeneity, reliance on domestic production in the region is broadly low for wheat, palm oil, and rice, reflected in relatively high net import dependence of 81.9, 80, and 52.9 percent, respectively. As regards import trading partners, Russia is the dominant source for wheat (34.7 percent), while palm oil and rice mainly come from Malaysia (47.8 percent) and India (36.3 percent), respectively (Table 3). By contrast, maize and cassava, which are mainly domestically sourced subsistence crops, have low net import dependence ratios at 8.5 and 8 percent, respectively.¹⁰ The main regional suppliers of maize and cassava are Uganda and Tanzania, accounting for 33.8 and 41.2 percent of imported maize and cassava, respectively.¹¹ The net import dependence of a staple food, which reflects its supply composition in terms of net imports versus domestic production, is expected to play a crucial role in shaping local staple food prices and food security in the region.

B. Overview of staple food price developments during COVID-19

Global food prices have increased substantially during the COVID-19 crisis, as the pandemic dealt a massive humanitarian and economic blow to countries worldwide. The upward trend in global food prices, which rose 25 percent over 2020–21, coincided with higher relative staple food prices—nominal food prices deflated by general CPI—in local markets in SSA (Figure 2.A-B). In SSA, where food makes up nearly 40 percent of the consumption basket, the local relative prices of staples surged by an average 19.3 percent in 2020–21 (Figure 2.C).¹² This was the sharpest increase in the staple food prices since the global financial crisis.

View Full Size

View Full Size

Evolution of Staple Food Prices

Citation: IMF Working Papers 2022, 135; 10.5089/9798400216190.001.A001

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

The recent step change in the local costs of staple foods was broad-based, albeit of varying magnitudes. The largest real price increases were seen for cassava (83.2 percent), wheat (18.6 percent) and maize (11.3 percent). The average price changes for rice and palm oil have been more muted. When consumption shares are factored in, there is a 8.5 percent real increase in the cost of a typical food consumption basket in the region (Figure 2.D).

While mainly imported staple food prices have risen more than the costs of domestically produced staples over the last decade (2012–21), the prices of locally sourced staples have spiked during the pandemic (2020–21) in some countries. For example, the prices of both cassava and maize (mainly produced locally) in Nigeria have more than doubled. In Ghana, the relative prices of cassava and maize have surged by 78 and 66 percent, respectively.

Stricter mobility restrictions in groceries and pharmacies, retail facilities, and workplaces during the pandemic have disrupted local food markets and accentuated the dual shock of rising food prices and falling incomes (Figure 2.E). Countries in the region have gradually eased these mobility restrictions towards the end of the period. Within countries, market-level data reveal different staple food price dynamics between urban and rural areas (Figure 2.F). The relative prices of maize and cassava—predominantly produced domestically in rural areas—have become more expensive in urban than rural areas. In contrast, the cost of wheat and rice—mostly imported staples—have been relatively cheaper in urban than rural local markets. This partly reflects better infrastructure and lower transport costs of imported goods in urban areas. On balance, the costs of the five most consumed staples are on average cheaper in urban area than rural area—see subsequent discussions of estimation results by location.

A. Supply

Local production. The domestic food production is a key component of staple food supply in SSA, where the agricultural sector makes up about one-fifth of the value added (Fuglie et al. 2020). Food production in the region is predominated by small-scale farming—it absorbs 80 percent of all farmers and accounts for 90 percent of food production (Unsal et al. 2020). Smallholder farmers typically operate small-scale farms and resort to unmechanized labor-intensive production (Lowder et al. 2021). Despite recent headways, agricultural (land and labor) productivity in SSA has remained low and lagged other regions (Gollin 2010). Over the last half century, the increase in crop yields in Sub-Saharan Africa has been muted, on average at 1.5 tons per hectare (t/ha), whereas yields have tripled to 3 t/ha in South Asia and jumped six-fold to 6 t/ha in East Asia (Fuglie et al. 2020). Moreover, there is limited storage capacity in SSA despite recent progress, which prevents many countries from maintaining sufficient food inventories. In some areas, this leads to sizeable food losses and limits the ability of countries to cushion supply disruptions and food price volatility that are exacerbated by frequent natural disasters and conflicts (ECOWAS 2020, Tadesse et al. 2014). The post-harvest handling and storage food losses in the region have been estimated at about 15 percent of total production (Affognon et al. 2015; Hodges et al. 2010).

The knock-on effects of the rising costs of food production inputs—fertilizers used to improve crop yields and fuel, combined with localized seed shortages have contributed to bouts of food price volatility in some countries (Schmidhuber et al. 2020). Input and production subsidies also shape the supply of staple foods. These production subsidies aim at raising agricultural production and supporting the price competitiveness of domestic food products by dampening production costs, and therefore, reducing import demand. From a food security perspective, governments have used subsidies to boost food availability, provide a form of social assistance to farmers, facilitate the access of vulnerable people to basic staple foods, and limit food price volatility (Morris et al. 2007). Despite their good intentions, ill-designed, broadly targeted, and prolonged subsidies can lead to bad outcomes. Poorly targeted food subsidies often distort productive investments toward low-productivity agriculture sector, imply high fiscal costs, and crowd out pressing public spending in priority sectors such as education, health, and infrastructure. Many subsidy programs are also socially regressive, that is, they actually benefit the rich much more than the poor (Coady et al. 2004). Recognizing these potentially negative outcomes, some countries have taken steps to reduce or remove subsidies, while improving their targeting and efficiency.

Net imports. In SSA, net imports complement the domestic production to ensure a sufficient supply of staple foods in local markets. Some countries rely more heavily on imports to bridge the food supply and demand gap of certain staples such as wheat and rice. The net import dependence—net food imports as a share of total food supply—synthesizes the information about net imports and local production. A high net import dependence for some staples potentially entails two opposite effects. On one hand, international trade disruptions can accentuate the risk to food security in high food import dependent countries. This was particularly critical in the context of COVID-19, as the pandemic-related restrictions have disrupted midstream and downstream components of the food supply chains and compressed the imports of some foodstuffs (Mogues 2020). For instance, export restrictions by the region’s main food trading partners (e.g., the European Union, Pakistan, Thailand) may have contributed up to 25 percent decline in food imports into the region (Zeufack et al. 2020). On the other hand, countries with access to global food markets can hedge against domestic food supply-side shocks, such as extreme weather conditions and insect infestations that damage harvests and dent food production (Nchanji et al. 2021). The net effect may depend on the degree of net import dependence of a country, diversification of supply sources, and substitutability of imported staples.

Transport and geographic challenges. Transport costs affect the quantity and quality of domestically available food items. Logistical constraints including weak transport infrastructures and geographical challenges—e.g., mountainous or rough terrain—weigh on the ability of market intermediaries to ship food at reasonable costs (Burke and Myers 2014, Nunn and Puga 2012). There are various transport costs, including international shipping costs and domestic distribution costs. These frictions add to staple food prices in local markets and can contribute to inflationary pressures (Gollin and Rogerson 2016). Typically, the closer the consumption centers to the main supply points—farms for locally sourced foodstuffs or borders for imported staples—the lower the logistical costs. This could imply cheaper transport costs for highly imported staples in large SSA cities where transport infrastructures are often better. By contrast, the transport costs for locally produced staples would be cheaper in rural areas where most farms are situated.

B. Demand

Consumption. The demanded quantities of staples are influenced by local dietary habits. These are reflected in the caloric contribution of staple food items to individuals’ daily diets—measured by their consumption shares— and shape prices in local markets (Timmer 2000). Data from FAO food balance sheets show that a typical consumption basket in SSA countries is predominantly comprised of cereals—maize, rice, and wheat—and starchy roots—cassava, sweet potatoes, and potatoes. Cereals and starchy roots make up nearly two-thirds of the daily per capita caloric intake, while other staples such as vegetable oils, sugar, fruits, meat, and pulses account for the rest. Moreover, substitutability among demanded staples allows consumers to switch from expensive products to cheaper ones, which in turn can help ease price pressures (Haggblade et al. 2017). Over the last 5 years, the share of each of the top five staples in the daily caloric intake has remained almost unchanged, pointing to limited substitution effects for the five most consumed staples.

Income. The purchasing power of consumers also affects food demand and prices. Households at the lower end of the income distribution spend a larger share of their incomes on food (Arezki et al. 2014, Ehui 2020). The very poor typically live from hand to mouth and devote a substantial part of their incomes to buy subsistence foods (Timmer 2010). An erosion of income, as seen during the pandemic, imposes a heavy cost on poor households and further hinders their access to sufficient and nutritious staple foods in SSA (World Bank 2021). By contrast, better-off households can afford more diverse diets and tend to spend more on non-food items. Thus, food consumption shares subsume the information on income differences and dynamics.

C. Global food prices and exchange rates

Global food prices. Net imports shape the pass-through of global food prices to domestic food prices. There are both direct and indirect effects. The direct effect occurs through wholesale purchases of foodstuffs on international markets. Indirectly, global food prices affect prices in local food markets through purchases of agricultural inputs—fertilizers, fuels, seeds, etc.—on global markets. In the absence of subsidies and tariffs, the higher the net import dependence, the stronger the link between global and domestic food prices. In practice however, policy interventions affect the pass-through from international market prices to domestic consumers prices. For instance, import tariffs aimed at insulating domestic producers from foreign competition can weaken the price transmission from international to local markets. By contrast, export tariffs are often used to ensure sufficient supply of locally sourced staples and alleviate price pressures on domestic markets. There have been significant price pressures during the pandemic, which contributed to the sharpest increase in staple food prices since the global financial crisis. In 2020–21, global food prices have risen sharply by 25 percent, alongside a nearly 20 percent jump in local staple food prices. Despite some heterogeneity across countries, staple food markets in SSA experienced dramatic price increases for major staples, in particular cassava (83.2 percent), wheat (18.6 percent) and maize (11.3 percent).

Exchange rates. The change in local prices may also reflect the relative changes in exchange rates. The relative strength of a country’s currency with respect to its trading partners affects the costs of imported food items in domestic markets. This effect depends on the exchange rate pass-through to domestic prices for tradable staple foods (Alper et al. 2016). After the onset of the pandemic, there were initially sharp broad-based currency depreciations in 2020 in our sample, followed by different speed recoveries (Unsal et al. 2020).

D. Adverse events

Adverse events such as natural disasters and wars exacerbate food price volatility. These events often cause severe human (deaths, injuries, displacements), physical (destruction of infrastructures, assets, etc.), and economic damages, and inflict deep scars on the affected countries (Dieppe et al. 2020). Absent public or international assistance, these shocks typically compress food production (disruption of planting and harvesting cycles, destruction of crops, etc.), dislocate supply chains, and degrade the living conditions in affected areas. Recurring bouts of violence in some areas (e.g., insurgencies in the Sahel) and armed conflicts push food prices higher by putting stress on food security from both supply and demand sides (McGuirk and Burke 2020, van Weezel 2016). Of particular concern is the greater frequency of natural disasters, especially climate disasters that are increasingly fueled by climate change.

A. Baseline specification

This section describes the empirical specification and estimation approach used to analyze the factors driving the observed change in the relative price of food staples. The analysis starts with the following baseline linear specification:

where y_cmi,t is the annualized (monthly year-on-year) change in the relative price (nominal prices deflated by general CPI) of staple food i, in market m, country c, month t, NID is the net import dependence defined as (imports – exports)/(production + imports – exports) in percent, CS is the consumption share reflecting the percent contribution of a staple food in the daily diet, ΔGFPI is the annualized (monthly year-on-year) change in the global food price index in real term, y^high is a dummy that takes 1 if the change in staple food price exceeds 50 percent and 0 otherwise, α_cmi stands for the country-market-item fixed effects, and ε is the error term.¹³ Note that for the dependent, we use the retail prices of staples in local markets.¹⁴ The model includes lagged net import dependence, consumption share, and change in the global real price to mitigate the bias from potential endogeneity and reverse causality. Following Calderón and Schmidt-Hebbel 2010, we control for episodes of high food price increases (y^high). Linear models tend to overestimate the effects when based on samples that include high food price episodes compared to estimations based on samples without these extreme episodes (Fischer et al. 2002). Thus, controlling for episodes of food price surges helps circumvent the exclusion bias of coefficients estimates for other determinants of food price inflation and improves the overall fit.¹⁵ However, it is important to stress that while lagged regressors are commonly used instruments in panel regressions, they may not fully guard from endogeneity concerns in reduced form estimations. For instance in cases where the lagged explanatory variables and unobserved characteristics have the same temporal dynamics, changes in local relative food prices could affect consumption and imports, and using lagged regressors would not ensure strict exogeneity (Bellemare et al. 2017; Leszczensky and Wolbring 2019). Therefore, our estimation results should be interpreted with care, as the estimated coefficients are not pure causal effects from strictly exogenous factors (Arellano and Bond 1991; Imbens 2014).

Equation (1), which assumes slope homogeneity across food staples, is estimated using fixed effects panel data regressions, which account for time-invariant unobserved heterogeneity across food items.¹⁶

B. Extensions

The baseline model in Equation (1) is augmented to encompass the changes in real effective exchange rate for highly imported staples, natural disaster and war shocks, location, along with interaction terms. The extended specification writes:

where ${\overline{M}}_{c,t}$ is a vector of additional variables including the percent change in the real effective exchange rate AREER, dummies for COVID-19, natural disaster, war, and location (largest city), and interaction terms. The binary COVID-19 variable takes 1 during the pandemic from January 2020 through the end of our sample period in September 2021. Natural disasters dummies are defined based on adverse events comprising climate, biological, and geophysical disasters recorded in the Emergency Disasters Database. Using the Correlates of War (COW) and Peace Research Institute Oslo (PRIO) databases, wars are identified as civil and inter-state armed conflicts with at least 1,000 battle-related deaths over the entire episode. We refer the reader to Dieppe et al. (2020) for detailed definitions of COVID-19, natural disaster, and war dummies.¹⁷ An increase in the real effective exchange rate level (ΔREER>0) indicates a real appreciation, reflecting a loss in price competitiveness in the home country vis-à-vis its trade partners. Simply put, an increase in effective exchange rate implies that the home country’s exports become more expensive while imports become cheaper. An uptick in real effective exchange rate is expected to lower the real price of highly imported staples whereas a real effective exchange rate depreciation is expected to raise the real cost of highly imported staples. Therefore, we expect the coefficient of the interaction between ∆REER and NIDhigh—a dummy for highly imported staples that takes 1 if the net import dependence of a staple food exceeds 75 percent and 0 otherwise—to be negative. Moreover, the coefficients associated with COVID-19, natural disaster, and war dummies are expected to be positive because these shocks typically entail large-scale disruptions in the supply and consumption of staple foods. For location effects, we expect relatively lower average staple food prices in largest cities, as better infrastructure in largest cities facilitates food transport, in particular the shipping of imported staples.

A. Baseline estimations

The empirical estimations use an unbalanced panel of 15 countries covering the period January 2012 to September 2021. The sample selection was guided by data availability. The primary sources of data used are the FAO FPMA and food balance sheets, IMF AREAER, Emergency Events Database, Correlates of War and Peace Research Institute Oslo armed conflicts database, IMF World Economic Outlook, and World Bank WDI (Table A1).¹⁸ The regressions are conditioned on lagged regressors to mitigate the bias from potential endogeneity and reverse causality. Nonetheless, the estimated coefficients should not be interpreted as pure causal effects from strictly exogenous regressors.

Empirical assessment of food price determinants in SSA

What key factors drive staple food prices in local markets in SSA? Taking Equation (1) to data, we run our baseline specification using fixed effects estimations. The two columns in Table 4 show fixed effects estimations of the baseline model excluding and including a dummy for high food price increases.¹⁹ As discussed after Equation (1), we control for episodes of high food price increases to limit the exclusion bias of coefficient estimates (Calderón and Schmidt-Hebbel 2010). This is confirmed by the large and significant coefficient values and the substantial improvement in the model fit. Conditional on other regressors, the real cost of staples rises by an average 89.2 percent in periods of food price surges (Table 4, column 2). The net import dependence has a positive and significant coefficient as expected. A one standard deviation increase in net import dependence from the panel sample mean of 24 percent to about 59 percent, would raise the relative prices of local staple foods by 3.9 percent (Figure 3.A).²⁰ As expected, the consumption share is also associated significantly, both in statistical and economic terms, with higher staple food prices. Following a one percent increase in the consumption share of a staple food, the corresponding relative prices in local markets are expected to rise, on average, by 0.7 percent. Turning to global food prices, the results show a positive and significant effect in line with our prior. A one standard deviation uptick of about 8.6 percent in the global food price index would increase the average real cost of staples by 1.3 percent. In the baseline specification, domestic factors (net import dependence and consumption share) contributed more (65 percent) to model fit than the external factor (global food prices).²¹

View Full Size

View Full Size

Estimated effects

Citation: IMF Working Papers 2022, 135; 10.5089/9798400216190.001.A001

Source: COW, EMDAT, FAOSTAT food balance sheets, FPMA GIEWS Tool, IMF WEO and AREAER, Nunn and Puga (2012), PRIO, Unsal et al. (2022), World Bank WDI, and authors’ calculations.Notes: The round markers show coefficient estimates and horizontal bars represent 95 percent confidence intervals. The vertical red line draws a reference line at zero.

• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Next, the baseline model is extended to include other variables such as the change in the real effective exchange rate AREER, dummies for COVID, natural disaster, war, and location (largest city), and interaction terms.

B. Exchange rates effects for imported staples

We now turn to the price effects for imported food staples (Alper et al. 2016). The extended specification includes the change in REER interacted with a dummy for highly imported staples, that takes 1 if the net import dependence of a staple exceeds 75 percent and 0 otherwise. As expected, the estimated coefficient of ΔREER * NIDhigh is significantly negative, as shown in Table 5. The estimate in column 1 in Table 5 suggests that a one percent appreciation in the REER—making imports cheaper and exports more expensive—is associated with about 0.3 percent decline in the relative price of imported staple foods in local markets.

The other columns (2–4) in Table 5 show interactions for the baseline drivers.²² The positive and significant interaction coefficients suggest that the average price effects are broadly higher for highly imported staples than locally produced food items. Simply put, an increase in consumption shares and global food prices would raise the relative price of highly imported staples more than those produced locally. Rescaling the estimated interaction coefficients by the standard deviations of the regressors reveals that the change in global food prices has the largest effect on the price of highly imported staples (Figure 3.B). Pass-through estimates in column 4 of Table 5, show that increases in global staple food prices could bring almost one-to-one (0.966=0.189+0.777) increases in the local relative prices of highly imported staples.²³ Thus, there are potential gains in promoting a competitive domestic production of highly imported staple foods in SSA countries.

C. Largest cities

The price of a commodity should be the same across frictionless markets. However, frictions such as transport costs typically lead to price differentials across food markets and location.²⁴ In SSA, the infrastructure gap is particularly acute in rural areas. Underdeveloped road networks complicate the access to many rural areas and very few rural residents have access to electricity or piped water. This pushes transport and market access costs higher in rural areas where about 60 percent of the population live and depend on subsistence agriculture (Gollin and Rogerson 2016). The first column of Table 6 shows that the average real cost of staple foods is 2.4 percent lower in largest cities. This points to better infrastructures in largest cities that facilitate food transport. Interestingly, the reduction in staple food prices in largest cities is smaller (1.9 percent in column 3 vs. 2.4 percent in column 1) in more geographically challenged countries—identified by an index which measures how rough in elevation is the terrain in a country. Nunn and Puga (2012) use this index of terrain ruggedness as an exogenous proxy for geographic isolation and transport challenges in Africa.²⁵ In more geographically challenged areas, food prices are often higher because transportation over irregular terrain is slower and more expensive and crop production is more difficult (Allen, Bourke, and Gibson 2005). Indeed, steep slopes hinder the transportation of crops to markets and make earthwork, seeding, watering, and harvesting more costly (FAO 1993). For highly imported staples in the second column of Table 6, the average relative price differential is larger: relative prices are 3 percent lower in largest cities. The standardized price effects are shown in Figure 3.C. These findings are in line with the implications of the spatial macro-model of Baptista et al. (2022). Although our analysis does not explicitly explore this channel, real income gaps between (urban and rural) areas could also entail differences in consumption shares and explain spatial differences in the relative prices of staple foods. Because wealthier households can afford more diversified diets, a larger concentration of affluent households in large cities may entail smaller consumption shares, and thus, lower relative prices of the top staple foods in urban areas than in rural areas.

D. Natural disasters and wars

We analyze what happens to staple food prices when countries are hit by natural disasters and wars. These adverse events often leave short-term socio-economic impacts and long-term scarring effects on the affected countries (Marto et al. 2018; Dieppe et al. 2020). The estimation results are reported in Table 7 and shown in Figure 3.D. Following natural disasters (climate, biological, and geophysical) and wars (intra, extra, and inter), the relative prices of staples in local markets surge by 1.8 and 4 percent, respectively, conditional on other variables. McGuirk and Burke (2020) and van Weezel (2016) document similar food price spikes in the aftermath of natural disasters and conflicts. Wars also exhibit more persistent effects on local staple food prices than natural disasters. Columns 1–3 in Table 8 show that the effects of natural disasters on the relative prices of staples taper-off to about 0.4 percent and become statistically insignificant after 3 months. By contrast, wars (columns 4–7) have more protracted effects on relative staple food prices, as reflected by the statistically significant coefficient of 1.5 percent associated with the 2-year lagged war dummy.²⁶ It is worth stressing that our estimations bundle various types of natural disasters together. While we leave finer analyses of the impact of different types of natural disasters for future work, let’s stress as in Dieppe et al. (2020) that the definition and type of a natural disaster matters because different categories of natural disasters (type, magnitude, frequency) can entail various price effects. For instance, Kabundi et al. (2022) argue that droughts tend to push staples inflation higher whereas floods tend to have deflationary impacts.²⁷ Moreover, attrition around natural disasters and wars entails that data might not be collected in affected food markets at times where staple food prices could rise the most. As a result, our estimated effects of natural disasters and wars on the local relative prices of staples could be interpreted as lower bounds.

E. Effects of the COVID pandemic

Focusing on COVID-19—a large-scale biological disaster—we assess to what extent the relative price effects of the drivers may have changed during the pandemic (Bogmans et al. 2021; Unsal et al. 2020). To this end, we interact our explanatory variables and a dummy for the COVID period. In Table 9 and Figure 3.E, the interaction coefficients for NID, CS, and ΔGFPI are all positive and significant. The negative coefficient associated with ΔREER * NIDhigh * COVID suggests that a one percent depreciation in REER would have raised the real cost for highly imported staples by 0.7 percent more during the pandemic, conditional on other variables. Thus, the price effects of these factors have been amplified during the pandemic, which led to massive disruptions of global and domestic food supply chains, unprecedented policy support, and currency depreciations in some countries.

F. Drivers of cross-country food price differences

To explore the country characteristics that may explain the differences in staple food prices across SSA countries, we fit the following model:

where ${\tilde{y}}_{cmi,t}=y_{cmi,t}-\left({\mathrm{\Sigma }}_{cm}{y}_{cmi,t}\right)/N_{cm}$ is the deviation of the relative food price change of staple i, in market m, and country c from the regional average, observed for each month t. N_cm stands for the number of country-market pairs. For the monetary variable, we use a dummy reflecting the quality of the monetary policy frameworks MPF. In each time period, we create a dummy that takes 1 if the Independence, Accountability, Policy and Operational Strategy, and Communications (IAPOC) index of monetary policy frameworks of a country is above the median and 0 otherwise (Unsal et al. 2022).²⁸ Our fiscal variable Debt is the debt-to-GDP ratio. ΔGDPPC is the change in GDP per capita (constant 2010 USD). GEO is an index measuring the roughness of the landscape in a country (Nunn and Puga 2012), and we expect food transport to be more difficult and expensive in geographically challenged countries. In Equation (3), GEO is not lagged because the overall roughness of a country’s terrain is expected to be nearly time-invariant and would likely not change in the short run. y^high is a dummy (1 if y > 50% and 0 otherwise), τ and α are time and item fixed effects, u is the error term.

Table 10 reports the estimates coefficients and Figure 3.F shows the standardized price effects. On average, the real cost of staple food is about 2.5 percentage points lower for countries above the median of monetary policy framework index.²⁹ This suggests that central banks can effectively curb food price inflationary pressures, and in turn control general inflation, by making strides to establish credibility, fostering independence, enhancing operational strategy, and improving communication (Unsal et al. 2022). Moreover, monetary authorities, should monitor and credibly calibrate their actions to account for both direct and second-round inflationary effects— through non-food item prices—of food price increases. This is of prime importance for low-income SSA countries which have faced higher inflation relative to richer countries, owing to less developed monetary policy frameworks, larger share of food in the consumption basket, and higher exposure to volatile food prices (World Bank 2021).³⁰ On the fiscal side, a one percent increase in debt to GDP ratio, would raise relative staple food prices by about 0.1 percent. A higher debt in percent of GDP may reflect weaker fiscal management and/or heavy debt burden, which could weaken the domestic currency in SSA countries. In this case, the relative prices of staple foods would rise in local markets, as nominal staple food prices—boosted by more expensive imported staples—increase faster than non-food components of the CPI.

Turning to income per capita—a proxy for overall economic development, a one percent increase is associated with 1.2 percent lower relative food prices, consistent with the findings in Calderón and Schmidt-Hebbel (2010). This negative and significant effect of income growth on real food prices is consistent with Engle’s law, which hinges on the view that wealthier households spend a smaller proportion of their income on food than their poorer counterparts. Thus, in growing developing countries, the relative decline in the demand for food would lead to lower relative food prices (Baffes and Etienne 2016).

Finally, a country with one-point higher geographic challenge index—or 100 meters higher terrain elevation, e.g., from Zambia-0.5 to Ethiopia-1.6—has, on average, 2.5 percent higher relative food prices. Typically, food transport is more challenging in countries where the terrain is very rugged (e.g., Rwanda, Ethiopia).

A. High food price increase

We start by assessing the sensitivity of our estimations to episodes of high food price increases. To do this, we define the episodes of staple food price surges using a higher threshold. Tables 11 and 12 show the estimated models in Equations (2) and (3), where y^high is a dummy that takes 1 if the change in staple food price exceeds 75 percent and 0 otherwise.³¹ As expected, the coefficient estimates of y^high is higher in Table 11 than Table 9. In Table 11, the sign and magnitude of the coefficients (e.g., in column 1) associated to NID (0.348 from 0.143), CS (0.846 from 0.558), ΔGFPI (0.334 from 0.393), and ΔREER * NIDhigh (-0.248 from -0.222) are similar to those from Table 9—where y^high is defined using a lower threshold of 50 percent. Similarly in column 2 of Table 12, the coefficients of MPF (-1.807 from -2.504), Debt (0.122 from 0.113), ΔGDPPC (-1.990 from -1.200), and GEO (2.523 from 2.532) are in line with their estimated values in Table 10. This demonstrates the robustness of the estimated effects to the choice of the threshold of price surges.

B. Net import dependence

To further establish the robustness of our analysis, we take an extra step to estimate the sensitivity of the price effects of the drivers to different levels of net import dependence. Table 13 reports the estimation of Equations (2) using a lower threshold of net import dependence to define the dummy for highly imported staples—here, NIDhigh takes the value of 1 if the net import dependence of a staple exceeds 50 percent and 0 otherwise. These robustness results are reported in Table 13 and are similar to the estimates in Table 9. In particular, the estimated coefficients associated with NID ranges from 0.099 in column 1 of Table 13 to 0.143 in column 1 of Table 9. Our robustness exercises lend additional confidence in the estimated effects of these factors on local staple prices.