This Selected Issues paper quantifies the importance of international food and oil price inflation for Uruguayan inflation. The paper employs the vector autoregressive (VAR) model and Phillips-curve estimations techniques to address these questions. Similar estimations are carried out for a broader set of emerging markets to put the Uruguayan results into an international context. The paper examines spillovers from commodity price increases to measures of core inflation, and reveals that world food prices have had a strong but stable impact on Uruguay’s inflation.

Abstract

This Selected Issues paper quantifies the importance of international food and oil price inflation for Uruguayan inflation. The paper employs the vector autoregressive (VAR) model and Phillips-curve estimations techniques to address these questions. Similar estimations are carried out for a broader set of emerging markets to put the Uruguayan results into an international context. The paper examines spillovers from commodity price increases to measures of core inflation, and reveals that world food prices have had a strong but stable impact on Uruguay’s inflation.

I. Commodity Prices: Their Impact on Inflation in Uruguay1

1. As in many other countries around the world, the recent surge in commodity prices has contributed to pushing up prices in Uruguay. This has coincided with domestic demand pressures associated with strong consumption and investment growth. For policy purposes it is, however, important to be able to disentangle the relative magnitude of these two forces: while domestic overheating clearly requires a contractive policy response, in the case of external price shocks the challenge for monetary policy is to limit second-round effects. In this context, it is necessary to assess the magnitude of such second-round effects.

2. This paper seeks to quantify the importance of international food and oil price inflation for Uruguayan inflation and to draw some international comparisons. We employ both VAR and Phillips-curve estimations techniques to address these questions. Moreover, we carry out similar estimations for a broader set of emerging markets to put the Uruguayan results into an international context. We also examine spillovers from commodity price increases to measures of core inflation.

uA01fig01

Uruguay CPI, world oil and food developments

(y/y change, percentage)

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

3. We find that world food prices have had a strong but stable impact on Uruguay’s inflation. A 10 percentage point increase in world food prices raises domestic CPI inflation by about 1.2 percentage points after a full year. This implies that about 2-2½ percentage points of current inflation can be attributed to world food price increases. This pass-through is in the upper range of those found for other emerging markets. Second-round effects are also considerable, as there has been a sizeable spillover from world food price increases to different measures of core inflation. An appreciating currency has, however, helped dampen the effects on domestic prices. The impact of world fuel price increases on headline inflation could, however, not be statistically confirmed.

4. Over a longer time horizon, however, commodity prices explain only a moderate fraction of the total variation in headline inflation in Uruguay. The variance decomposition suggests that in the last 10 years, variations in world food and oil prices have contributed only about 15-20 percent of the variation in total inflation. Aside from inertia and other factors not captured by the model, exchange rate changes explain nearly a third of the variation in headline inflation.

A. Evidence From Phillips-Curve Estimations

5. As a first way of assessing the impact of commodity prices on domestic inflation dynamics, we estimate an augmented Phillips-curve. Estimating a Phillips-curve allows for assessing the impact of commodity shocks while controlling for other factors such as demand pressures and exchange rate pressures. The approach is similar in spirit as Hooker (2002) and De Gregorio, Landerretche, and Neilson (2008). The estimation uses quarterly data on CPI inflation, with lags of inflation, the unemployment rate gap (as a commonly used proxy of the output gap) and the nominal effective exchange rate as well as changes in the peso price of world fuel and food prices as explanatory variables:

(1)πt=α+δΣi=12πti+φΣi=02unempl+γNEER+θWorld¯oil+θWorld¯food+εt

6. Results show a stable impact of food prices on CPI inflation. The fit of the estimation is good; all coefficients have their expected signs and are significant, with the exception of world fuel prices. The coefficient onworld food prices of around 0.05 suggests that a 10 percentage point increase in world food prices is associated with an immediate 0.5 percentage point increase in domestic CPI inflation. The overall importance of world food prices in explaining the variation of total inflation within the sample is, however, small. The pass-through of oil prices has been reported to be low for other developing countries (see, for example, Deutsche Bank, 2008, and Duma, 2008). For a larger set of countries, De Gregorio, Landerretche, and Neilson (2008) find a declining pass-through, which stabilizes at around 0.015.

uA01fig02

Recursive Estimates of World Food Price Coefficient

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

Table 1.

Phillips Curve Estimation

article image
Note: Dependent variable: headline inflation. Sample covers 1998 Q1-2008 Q1, with 41 observations. Crisis dummy is a dummy which is equal to one during the 2002-03 crisis. Includes quarterly dummies. Reported standard errors are adjusted to correct for heteroskedasticity and serial correlation (Newey-West).

7. We also estimate hybrid Phillips-curves with forward- and backward looking expectations, obtaining less stable estimates. To proxy for inflation expectations, we use annual inflation expectations from Consensus Forecasts.2 These expectations are available monthly, but only for end-of-year inflation, and we proxy 12-month ahead expectations by appropriately weighting expected current and next-year inflation for each month. To address the potential problem of endogeneity we estimate the equation with the generalized method of moments (GMM). The coefficient on food prices turns out to be rather unstable, while the one on fuel prices is around 0.01 (a 10 percentage point increase in world fuel prices would be associated with a 0.1 percent rise in headline inflation).

B. Evidence From Vector Autoregressions

8. To assess the impact of external and domestic shocks over time, we estimate a vector autoregressive model (VAR). This method allows to control for factors such as economic activity or exchange rate movements in a dynamic setting when examining the impact of commodity prices on inflation. The approach is similar to several recent studies (see, for example, Blanchard and Gali, 2008, De Gregorio, Landerretcghem and Neilson, 2008, and Deutsche Bank 2007, 2008). We estimate a six variable recursive VAR with quarterly data, covering the period 1996Q1-2008Q2 for nominal prices of fuel and food commodities (in US$ dollars), economic activity, the nominal effective exchange rate, monetary aggregates, and headline, food, fuel and underlying (non food and fuel) consumer prices. Aggregate demand shocks are proxied by the output gap (measured again as deviations of unemployment from its long-term trend). (See appendix for more details).

9. Impulse responses derived from the VAR estimations indicate how inflation has reacted to different types of shocks. These responses take into consideration not only the direct impact of the shocks, but also their indirect impact through their feedback on other endogenous variables. We impose exogeneity conditions on both oil and food prices—these variables are assumed not to be affected by domestic conditions in Uruguay. For the remaining variables, a Cholesky factorization is used, with the following ordering: domestic economic growth; money growth; exchange rate growth; and domestic inflation. To explore spillovers to other prices, we estimate the response of different inflation measures to a shock of 1 percent to three variables: commodity fuel inflation; commodity food inflation; and the output gap.

uA01fig03

Response of Headline Inflation to World Food Shock

order 1, w_food, headline

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

Graphs by irfname, impulse variable, and response variable

10. Food commodity price shocks have a positive and significant impact on headline inflation. The impact of these shocks on headline inflation is fairly protracted, with the response dying out within the course of a year. A 10 percentage point shock in world food commodity prices initially raises domestic headline inflation by 0.5 percentage points. After one year, headline inflation is higher by 1.2 percentage points as a result of the same shock.3 These results are in line with the Phillip’s curve estimates (¶6). The estimated impact is higher when using indices in local currency, namely 1.8 percentage points after one year.

uA01fig04

Pass through of World Food Price Shocks to Headline inflation 1/

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

1/ Authors' calculations (maximum impact within 12 quarters) in response to a 10 percentage point shock

11. The pass-through of food price inflation is relatively high compared to other emerging markets. We estimated similar VARs for nine other countries; the pass-through for Uruguay is higher than in all but two countries in the sample. Comparing the estimated pass through for Uruguay with those found in other studies (see, for example, Deutsche Bank 2008) also confirms the relatively strong impact on CPI in the Uruguayan case. Possibly, this reflects the relatively rare use of price controls or voluntary price restrictions in Uruguay, as well as the relatively high share of tradable food items in CPI (14.3 percent).

12. Furthermore, there is evidence of spillover to most measures of core inflation. The impact of food price shocks on the different measures of core inflation varies, but is generally positive, large, and even more persistent than the impact on headline inflation. The duration of the response of core inflation to the initial shock typically lasts between 1½ and 2 years. On average, a 10 percentage point shock raises core inflation by 0.8 percentage points immediately, and close to 2 percentage points by the end of the end of one year. In particular, measures of core inflation that exclude regulated prices show a larger acceleration in inflation in response to the original shock.

Table 2.

Pass through of Food Price Shocks to Inflation

article image
Source: authors’ estimates1/ Estimates are the percentage point change in inflation in response to a 10 percentage point increase in food commodity prices2/ Official measures of core inflation

13. The impact of food commodity price shocks on the food sub-component of the CPI has also been sizeable. As expected, domestic food inflation responds significantly to the shocks, reflecting the large component of tradable food items in the total food basket. A 10 percentage point increase in world food prices raises domestic food prices immediately by 1.6 percentage points, and by 4 percentage points after a full year.

14. We do not find strong evidence of pass-through from world fuel price shocks to headline inflation. Fuel commodity price shocks do not have a statistically significant impact on headline inflation—the impact is positive and significant in the first two months following the shock, but becomes negative and insignificant thereafter. World fuel prices do have a positive and significant impact on domestic fuel inflation. However, there appears to be no statistically measurable spillover from fuel inflation to non-fuel inflation (i.e., headline inflation excluding fuel) when using the US dollar-based index in the estimations.4 In addition to the low share of fuel in the CPI,5 this also likely reflects delays in adjustment of domestic fuel prices as well as regulated price—changes in domestic energy prices have not kept pace with increases in world fuel prices.

uA01fig05

World Commodity Price Inflation and Domestic Inflation

(in local currency: percent)

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

uA01fig06

Response of Headline Inflation to World Fuel Shock

order1, w_fuel, headline

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

Graphs by irfname, impulse variable, and response variable
uA01fig07

Response of Fuel Inflation to World Fuel Shock

order1, w_fuel, fuel

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

Graphs by irfname, impulse variable, and response variable

15. The model finds evidence of demand side pressures on inflation. As expected, headline inflation is affected by our measure of the output gap. In particular, a widening (fall) of the output gap has a negative (positive) and significant impact on headline inflation. A negative (positive) shock of 10 percentage points6 to the output gap measure is expected to lower (raise) headline inflation by 0.4 percentage points after 1 year.

uA01fig08

Response of Headline Inflation to Output gap shock

order1, ugap, headline

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

Graphs by irfname, impulse variable, and response variable
uA01fig09

Response of Headline Inflation to Exchange Rate shock

order1, neer, headline

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

Graphs by irfname, impulse variable, and response variable

16. Exchange rate shocks also contribute to explaining CPI movements. The initial impact of a currency appreciation on consumer prices is negative, as expected, and remains so for at least a year. By the end of two years, the response is imprecisely estimated. The pass through is quite large in the first year, with a 10 percentage point appreciation lowering headline inflation by about 3.8 percentage points. Hence, the strong rise of Uruguay’s nominal effective exchange rate (17 percent since mid-2007) has helped to buffer the impact of external commodity shocks on domestic inflation.

17. Over a longer time horizon, commodity prices explain only a moderate fraction of the total variation in headline inflation in Uruguay. The variance decomposition allows us to assess how much of the forecast variance in domestic inflation over the estimated period can be attributed to external vs. domestic factors. In the last 10 years, variations in world food and oil prices have contributed only about 15-20 percent of the variation in total inflation. Aside from inertia and other factors not captured by the model, exchange rate changes explain around 28 percent of the variation in headline inflation.

uA01fig10

Contribution to Variation in Headline Inflation

(in percent)

Citation: IMF Staff Country Reports 2009, 103; 10.5089/9781451839432.002.A001

C. Conclusions

18. The rise of world food prices is clearly an important factor behind the rise in inflation in Uruguay. The link between world food prices and domestic inflation is significant and robust, and stronger than in many other emerging markets. Roughly one third of current headline inflation levels can be traced to world food price inflation.

19. However, domestic demand pressures have also played an important role, and spillovers to core inflation support the need to contain second-round effects. Tighter fiscal and monetary policy would help dampen demand pressures. More exchange rate flexibility would also help, as evidenced in the sizeable contribution of exchange-rate movements in explaining variations in inflation. Lastly, avoiding backward-looking indexation mechanisms is critical to prevent exogenous shocks from becoming entrenched in inflation and inflation expectations.

Appendix: VAR Methodology

We examine the sources of inflationary pressures in Uruguay using a six variable recursive vector autoregressive (VAR) model to isolate the impact of domestic and external shocks on domestic consumer prices. This approach captures the dynamic structure underlying the interaction between strictly exogenous world fuel and food commodity price shocks, as well as (endogenous) economic activity, the exchange rate, monetary policy and domestic consumer prices. The model simultaneously regresses each endogenous variable on lags of itself and all other variables in the model. The underlying recursive structure of the variance-covariance matrix allows for identification of shocks stemming from external and domestic developments, and estimation of their effects on consumer prices.

VAR models have been widely used in the literature analyzing the response of an economy to commodity price shocks. Recent studies include De Gregorio, Landerretche and Neilson (2007) and Pincheira and Garcia (2007) who analyze the effects of oil shocks on inflation in Chile; Blanchard and Gali (2007) who explain changes over time in the macroeconomic effects of oil shocks in industrialized countries; and Mishkin and Schmidt-Hebbel who analyze the macroeconomic effects of external shocks under inflation targeting.

A VAR approach is relevant for analyzing a phenomenon, such as rising inflation, without imposing a prior theoretical approach. Hence the approach suits the purposes of the current study given that it does not require necessarily having strong priors about competing explanations for rising inflation, instead allowing the regularities found in the data to tell a story. World fuel and food prices are treated as exogenous variables, and hence assumed to be determined independently of the rest of the system. Ordering of the endogenous variables in the VAR to achieve identification is informed by economic theory and supported by Granger causality tests. Shocks in the system are identified according to the following recursive specification7:

πtfuel=Et1[πtfuel]+εtfuel(1)πtfood=Et1[πtfood]+εtfood(2)Δyt=Et1[Δyt]+β1εtfuel+β2εtfood+εtΔy(3)Δet=Et1[Δet]+γ1εtfuel+γ2εtfood+γ3εtΔy+εtΔe(4)Δmt=Et1[Δmt]+θ1εtfuel+θ2εtfood+θ3εtΔy+θ4εtΔe+εtΔm(5)πtCPI=Et1[πtCPI]+δ1εtfuel+δ2εtfood+δ3εtΔy+4εtΔe+εtΔm+εtCPI(6)

The system incorporates the dynamic effect of a commodity price shock on domestic prices, with shocks affecting prices both directly as well as indirectly via previous stages. The domestic price variable contains the expected inflation at month t, the supply and demand shocks, the exchange rate and monetary policy shocks, as well as shocks due to CPI inflation.

References

  • Blanchard, Olivier and Jordí Galí, 2008, “The Macroeconomic Effects of Oil Price Shocks: Why are the 2000s so Different from the 1970s?NBER Working Paper.

    • Search Google Scholar
    • Export Citation
  • De Gregorio, José, Oscar Landerretche, and Christopher Neilson (2007), “Another Pass-Through Bites the Dust? Oil Prices and Inflation?Economia.

    • Search Google Scholar
    • Export Citation
  • Deutsche Bank, 2007, “Inflation Pass-Through Composition in EM,Global Markets Research Economics Special Report, May.

  • Deutsche Bank, 2008, “Commodity Pass-Through and 2nd Round Effects in EM InflationEmerging Markets Monthly, May.

  • Duma, Nombuelo, 2008Pass-Through of External Shocks to Inflation in Sri Lanka,IMF Working Paper 08/78 (Washington, D.C.: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Hooker, Mark, 2002, “Are Oil Shocks Inflationary? Asymmetric and Nonlinear Specifications versus Changes in RegimeJournal of Money, Credit, and Banking, 34 (2), 540-61.

    • Search Google Scholar
    • Export Citation
  • Mihaljek, Dubravko, and Marc Klau, 2000, “A Note on the Pass-Through from Exchange rte and Foreign Price Changes to Inflation in Selected Emerging Market EconomiesBIS Papers 8, 6981.

    • Search Google Scholar
    • Export Citation
  • Mishkin, Frederic, Exchange Rate Pass-Through and Monetary PolicyNBER Working Paper.

1

This paper was prepared by Rita Babihuga and Gaston Gelos.

2

See Gelos and Rossi (2007) and Celasun, Gelos, and Prati (2004).

3

We also examined the impact on the producer price index (PPI), finding that world food price shocks have a large positive and significant impact on producer prices, with the impulse response of the PPI to an increase in world food prices exceeding, in magnitude and duration, the response of consumer prices to the same shock. As in the case of CPI (¶14), the impact of world fuel prices on the PPI is minimal.

4

When using a peso-based price index, the initial impact is positive and significant, but after the first quarter, the impulse response looses significance.

5

Gasoline products account for just under 3 percent of the total CPI basket.

6

On average, this would be equivalent to a 0.5 percentage point increase (decrease) in the unemployment rate.

7

πtfuel and πtfood are world fuel and food price inflation respectively, Δyt is the first log difference of the activity index, Δet is the first log difference of the nominal effective exchange rate, Δmt is the first log difference of the monetary aggregates and πtCPI is the rate of CPI inflation

Uruguay: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Uruguay CPI, world oil and food developments

    (y/y change, percentage)

  • View in gallery

    Recursive Estimates of World Food Price Coefficient

  • View in gallery

    Response of Headline Inflation to World Food Shock

    order 1, w_food, headline

  • View in gallery

    Pass through of World Food Price Shocks to Headline inflation 1/

  • View in gallery

    World Commodity Price Inflation and Domestic Inflation

    (in local currency: percent)

  • View in gallery

    Response of Headline Inflation to World Fuel Shock

    order1, w_fuel, headline

  • View in gallery

    Response of Fuel Inflation to World Fuel Shock

    order1, w_fuel, fuel

  • View in gallery

    Response of Headline Inflation to Output gap shock

    order1, ugap, headline

  • View in gallery

    Response of Headline Inflation to Exchange Rate shock

    order1, neer, headline

  • View in gallery

    Contribution to Variation in Headline Inflation

    (in percent)