Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Mauritius: Selected Issues

Author(s):
International Monetary Fund
Published Date:
July 2008
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

II. Inflation and the Role of Administered Prices in Mauritius15

A. Introduction

37. After fifteen years of single-digit inflation rates, price increases in Mauritius began to accelerate in 2006 and remained high in 2007 (Figure 1). In November 2006, end-period consumer price inflation reached 12.3 percent, up from 6.1 percent just six months earlier. Half a year later, end-period inflation was still hovering high at 11.1 percent. This development coincided with the implementation of measures associated with the 2006/07 budget, including the relaxation of certain price controls in mid-2006.

Figure 1.CPI inflation, 2004–07

Source: Mauritius Central Statistics Office.

38. This paper analyzes the relationship between measured inflation and relative price variability, particularly affected through changes in administered prices, and investigates the determinants of inflation over the period 2002–07. The analysis suggests that the mid- 2006 sharp price liberalization of several administered prices contributed around one half of overall inflation three months later. In general, administered price adjustments account for a large share of the monthly cross-sectional variability of prices, especially during episodes of marked adjustment as in fiscal year 2006/07. The administered price regime in Mauritius and the discrete price changes associated with it leads to increased inflation through higher relative price variability (i.e., higher variance and skewness of the inflation distribution), although money growth and the nominal effective exchange rate (NEER) also play an important role.

39. The study combines a micro-level analysis of price variations with a macro-level analysis of the determinants of inflation. We use highly disaggregated consumer price index (CPI) data to investigate the relationship between inflation, and the variance and skewness of price changes. The analysis is undertaken for 163 goods and services in the Mauritian representative consumer basket, an approach that relies on individual product-level inflation. 16 The weights used to compute the CPI were derived by the Central Statistics Office of Mauritius (CSO) from the 2001/02 Household Budget Survey and remained unchanged during the sample period.17 The determinants of inflation—including monetary policy and the nominal exchange rate—are analyzed through Ordinary Least Squares (OLS) in a simple cointegrating framework.

40. In recent months, demand pressures and supply conditions have led to a worldwide surge in food prices while many commodity prices are at an all-time high. In response to rising imported food prices and to contain cost of living increases, Mauritius more than doubled its consumption subsidies on rice, wheat flour, and cooking gas in April 2008. Other countries took measures such as reducing or eliminating customs fees on some food imports, introducing VAT and import duty exemptions, introducing price controls (e.g., ceilings on retail prices), banning food exports, increasing state involvement in procurement, or increasing direct transfers to vulnerable households.18 These developments have put inflation back on the agenda of policymakers, rendering it important to better understand the dynamics of inflation and the monetary transmission mechanism to identify appropriate policy responses.

41. In what follows, Section B describes administered prices and post-liberalization inflation developments in Mauritius, and Section C assesses the relationship between relative price variability and inflation. Section D presents a simple empirical model of the determinants of inflation, focusing on the interplay between features of the price change distribution (inflation, variance, and skewness), while controlling for monetary policy variables and the nominal exchange rate. Concluding remarks are offered in Section E.

B. Drivers of Inflation in Mauritius

The Scope of Administered Prices

42. The prices of several imported basic foodstuffs, including rice, wheat flour, and vegetable oils, have traditionally been administered in Mauritius. Administered prices have long represented a form of social protection in Mauritius due to the country’s dependence on imported staples such as rice and wheat flour. Price controls were introduced in the late 1960s to benefit consumers by providing essential commodities at prices lower than market levels (Ministry of Industry, 2006).

43. Historically, administered prices have played an important role in the inflation-exchange rate interplay in Mauritius. Price controls for goods other than the main imported staples were introduced in November 1967 following a devaluation of the Mauritian rupee, and several waves of price adjustments have subsequently ensued. Prior to the November 1967 devaluation (by 14.3 percent) against the pound sterling, average inflation had been very low at 1.3 percent per annum (1962–67). Two more devaluations (in 1979 and 1981) coupled with gradual rises in prices of imported staples and high wage settlements, led to a rise in inflation to 26.5 percent in 1980–81. Following the pegging of the rupee to a basket of currencies reflecting the country's trade patterns, inflation remained relatively low over the following two decades, at an average annual rate of 5.1 percent (1985–05).

44. The administration of prices in Mauritius is implemented either through a ceiling on the retail price level or the mark-up. Fixed prices are communicated to the public in the massmedia and to the importers through written correspondence. Retail prices set up by maximum mark-up are determined on the basis of shipment arrivals, depending on the CIF value and the exchange rate. The Price Control Unit of the Ministry of Industry, Small & Medium Enterprises, Commerce and Cooperatives oversees price controls. Some goods, such as petroleum products, ration rice, and wheat flour, are entirely imported through the parastatal State Trading Corporation (STC), which handles government interventions in the market. The STC and the privately-owned Mauritius Portland Cement Company Ltd. share the importation of cement. The Meat Authority is in charge of meat imports while the Agricultural Marketing Board oversees imports of food products that compete with domestically produces ones, for which there are administered prices (primarily onions, garlic, and potatoes). The enforcement agency is the Consumer Protection Unit which checks prices in the market.

45. As of end-2006, around 21 percent of the CPI basket (by value) was subject to price controls (Table 1). All fuel products and around one third of tradable goods have administered prices. Under the maximum price regime fall the following ten product categories: bread, cement19, sugar, fertilizer, rice (excluding basmati), flour (including wheat flour), onions, cooking gas, iron/steel bars, and petroleum products (including kerosene). The mark-up regime is applied to imported fresh fruit, milk powder, pharmaceutical products, some drugs, timber, tires and tubes, corned beef, corned mutton, and pilchards (sardines).

Table 1.Classification of 163 CPI items
overallfuelnon-fuelfoodnon-foodtradablenon-tradable
Administered295241217263
% weight in CPI21516912202
Free1340134409410034
% weight in CPI7907926535524
Source: Mauritius Central Statistics Office, Price Control Unit, and authors’ estimates.
Source: Mauritius Central Statistics Office, Price Control Unit, and authors’ estimates.

46. In Mauritius, the price control mechanism—notably for staple foods—is characterized by infrequent, large price adjustments. This is a consequence of maximum retail prices being determined on a yearly basis for rice, wheat flour, bread, cooking gas, and cement. Figure 2 compares international and domestic prices for imported staples, namely rice and wheat (using the bread and flour CPI for the latter). It shows that the two series generally comove, but often domestic prices lag behind world prices (left panels). Furthermore, in contrast to the international price series which exhibit steady volatility, domestic prices are characterized by large and infrequent adjustments (right panels).

Figure 2.Domestic and International Commodity Prices, 2003–07

Source: Mauritius Central Statistics Office, IMFWorld Economic Outlook, DataStream, and authors’ estimates.

Note: The world rice price is for Thailand milled white rice (5 percent broken) in MUR per metric tonne. The international wheat price is for No.2 hard wheat (Kansas) in MUR/Bu. The small amount of monthly volatility in the domestic rice price is caused by different types of rice being accounted for in the domestic CPI. Only the price of government imported ration rice is administered in Mauritius.

47. Starting in April 2004, petroleum product prices have been regulated through the APM and imported under a sole license by the STC. Through the APM, the pump prices of gasoline and diesel—the bulk of petroleum products consumed—are adjusted to reflect the most recent contractual conditions negotiated by open tender. Retail prices of petroleum products are adjusted on a more frequent, quarterly basis, but have too been lagging behind world petroleum prices (Figure 3).

Figure 3.Domestic pump price and world petroleum price, 2004–07

Source: Mauritius Central Statistics Office, WEO, and Staff estimates.

Note: Pump price index is a CPI weighted nominal price index for gasoline and diesel; the world market petroleum price index is the MUR average of 3 international market crude oil prices (Dated Brent, West Texas Intermediate, and Dubai Fateh) per

48. In its 2006/07 budget the government announced a series of changes to administered prices. While price controls were not abandoned, existing subsidies on food items were reduced, fees increased, and petroleum prices raised. The savings on reduced rice and flour subsidies were used to finance the primary school feeding program, a targeted income support program, and the Empowerment (re-training) program for workers formerly employed in the sugar sector. Excise duties on tobacco were raised by 20 percent, and duties on imported alcohol were brought in line with those for imported products (also reflecting higher taxation for higher alcohol content).20 Finally, excise taxation was applied to PET bottles which are used in the soft drinks industry. The government further announced its intention to abandon traditional price fixation in favor of an appropriate competition framework.

Inflation Developments in the Wake of Price Liberalization

49. Following the mid-2006 liberalization, major price increases were observed for a large number of goods including food, soft drinks, cigarettes, and alcoholic beverages (Table 2). Other commodities affected were clothing, electricity, and taxi fares. The price of government imported flour and kerosene almost doubled, while that of items such as bread, rice, diesel oil, and gasoline rose by around 50 percent. After the initial shock, CPI inflation eased in the first half of 2007, with end-June inflation rates coming down for most product items and overall inflation falling by 2 percentage points.

Table 2.Yearly inflation rates (end-of-period) for specific items, 2004–07
Dec–05Dec–06Jun–07Dec–05Dec–06Jun–07
Individual product items
Bread0.056.823.1Electricity0.09.40.0
Government imported rice54.3Gasoline11.140.03.8
Government imported flour0.066.245.0Diesel oil14.852.5-0.5
Flour preparations-2.924.113.5Motor oil16.610.59.3
Cigarettes5.417.84.8Component CPI
Rum3.917.93.2Food and non-alcoholic beverages6.213.719.9
Whisky22.620.09.9Alcoholic beverages and tobacco4.918.23.9
Beer and stout0.725.41.8Transport-3.015.34.1
Beef3.425.124.4Electricity/gas/fuels/housing/water6.110.44.5
Frozen beef12.121.230.8Restaurants and hotels6.520.210.3
Cooking gas16.826.014.5
Korosene131.490.65.2Overall CPI3.911.99.9
Source: Mauritius Central Statistics Office and staff estimates.
Source: Mauritius Central Statistics Office and staff estimates.

50. As a consequence, inflation accelerated (Figure 4). High inflation was reflected not only in the overall CPI, but also in three measures of core inflation, which the Bank of Mauritius (BoM) began compiling in 2006. Core 1 inflation strips out food and alcoholic beverages. Core 2 also removes energy and administered price products. Trim 10 is calculated by symmetrically trimming 5 percent of the distribution of CPI changes (BoM, 2007). According to the CPI, inflation pressures reached a maximum (of 10 to 12 percent) in the second half of 2006.

Figure 4.CPI and core inflation, 2004–07

(yearly, end-of-period)

Source: Mauritius Central Statistics Office.

51. This wave of price liberalization is estimated to have directly contributed around one half of overall inflation three months later (Table 3). There are two ways to measure the direct effect of liberalization on overall inflation: one is to focus on the administered price increases for individual goods, and another is to use the observed CPI subindices for the categories to which the individual goods belong, together with CPI weights.21 First, we consider the administered price increases of all goods subject to liberalization in mid-2006, and find that the post-liberalization (weighted) average price increase was 13.3 percent, corresponding to 2.8 percentage points of CPI inflation. If we look at the inflation rates based on corresponding CPI categories, the average price increase is 13.8 percent, or 3.2 percentage points. This implies that price liberalization added between 48 and 54 percent to the overall inflation rate (representing the increase in the CPI between May and September 2006) of 5.9 percent.22

Table 3.Direct effect of price liberalization on overall inflation1/
WeightPrice changeEstimated effectCPI sub indexCPI changeObserved effec
(%) (I)Date(%) (II)(I) * (II)May–06Sep–06(%) (III)(I) * (III)
Cigarettes4.810–Jun–0617.80.9132.2155.717.80.9
Alcoholic beverages3.810–Jun–0612.10.5133.0151.213.70.5
Vehicles4.2Jun–06–6.1–0.3101.696.6–5.0–0.2
Rice (overall)2.10.3120.6144.219.60.4
of which: Rice (grn imported)0.53–Jul–0654.00.3
Flour (overall)0.30.1140.0190.235.90.1
of which: Flour (grn imported)0.23–Jul–0649.00.1
Bread1.73–Jul–0623.10.4147.9191.029.10.5
Flour based products0.53–Jul–0649.00.2129.2143.911.30.1
PET bottles1.2Jul–064.90.1113.5134.118.10.2
Cooking Gas1.43–Jul–069.10.1130.6142.59.10.1
Gasolene2.73–Jul–0620.00.5157.7189.220.00.5
Diesel0.43–Jul–0615.00.1245.7282.314.90.1
Sum’ weighted average21.413.32.813.83.2
Overall CPI124.3131.75.9
Contribution to overall inflation48%54%
Source: Mauritius Central Statistics Office and staff estimates.

Measured as the change in the CPI before the liberalization (May 2006) and two months after the liberalization (September 2006)

Source: Mauritius Central Statistics Office and staff estimates.

Measured as the change in the CPI before the liberalization (May 2006) and two months after the liberalization (September 2006)

52. Second-round effects may have played a role in driving overall inflation at least through mid-2007. Second-round effects are changes in those CPI subcategories that are indirectly affected in the first round of price adjustments. For example, while the increase in taxi fares in the month subsequent to an adjustment in fuel prices could still be considered a first round, direct effect, adjustments in other prices that will follow the increased taxi fare would be considered of higher order, and can be symptomatic of changes in long–run inflation expectations. Second–round effects are an endogenous response to inflation, and often arise through wage bargaining (for a theoretical contribution, see Hledik, 2003). In the case of Mauritius, public sector wages are only adjusted in response to inflation every five years, the last such adjustment having taken place in May 2003. Estimating second–round effects is difficult in this context without further assumptions. Nevertheless, they are likely to have also played a role in driving up prices.

53. Exchange rate movements have also affected inflation. Since 2004 the Mauritian rupee (MUR) has depreciated against the Euro and, to a smaller extent, against the U.S. dollar. The nominal effective exchange rate (NEER) has depreciated steadily since 2004 at an average annual rate of 5.9 percent, while the real effective exchange rate (REER) depreciated on average by almost 1 percent per annum (Table 4). The BoM has intervened in the foreign exchange market solely to smooth exchange rate fluctuations rather than alter the trend. Consistent with the existence of administered prices, the pass-through of the nominal exchange rate to changes in domestic prices in Mauritius has been found to be limited (with an elasticity of 0.23 over the period 1977 to 2004).23

Table 4.Exchange Rate Developments (annual percentage change), 2004–07
20042005200620071/2004–20072/
NEER-4.3-7.4-7.0-3.2-5.9
REER-2.5-3.9-0.82.3-0.8
US$’MUR1.8-6.9-6.91.7-4.1
EUR’MUR-7.4-7.1-7.6-5.6-6.7
Source: Authors’ estimates.

Until June 2007.

Average annual percentage change.

Source: Authors’ estimates.

Until June 2007.

Average annual percentage change.

C. Relative Price Variability and Inflation

54. Several contributions have formalized the relationship between inflation and relative price variability. For example, Ball and Mankiw analyze firms’ responses to supply shocks in a setting where adjusting firms incur “menu costs” (Ball and Mankiw, 1994, 1995). Menu costs, a form of transaction costs, refer to the cost of updating menus, price lists, brochures, etc. by firms when prices change in an economy. The authors use a one-period theoretical set-up to analyze the relationship between the distribution of unobserved real sectoral shocks (which can be proxied in empirical applications by actual price changes) and inflation in the presence of menu costs. Once these are taken into account, the model predicts that: (1) firms react more to positive shocks than they do to negative ones when the cross-sectional price distribution is skewed to the right; and (2) relative shocks that raise some prices (while lowering others) will induce more upward than downward adjustment. An increase in the cross-sectional dispersion of price changes (variance) is associated with higher inflation because of price rigidity caused by the presence of menu costs. Furthermore, a simultaneous increase in skewness (caused by a few large price increases in some goods accompanied by small price increases or reductions for other goods) will also be associated with inflation. The interplay between the mean, variance, and skewness of the price change distribution are illustrated in Figure 5.

Figure 5.The variance, mean, and skewness of the inflation distribution

Source: Coorey, Mecagni, and Offerdal (1997).

55. The empirical literature has generally identified a positive relationship between relative price variability and inflation rates. 24 Early studies include Okun (1971) and Vining and Elwertowski (1976) who analyze for the first time the link between the standard deviation of relative price changes and inflation using a cross-sectional and time series approach, respectively. More sophisticated econometric analyses have subsequently been undertaken by Engle (1982) who popularized conditional variance approaches to analyzing the link between relative price variability and inflation. Numerous country studies have since then been undertaken.25 The typical finding has been that changes in administered prices affect the general price level and therefore inflation, as in Ball and Mankiw’s models (see, for example, Coorey, Mecagni, and Offerdal, 1996, 1997; Wozniak, 1999; Uzagalieva, 2003).

56. We find that in Mauritius changes in administered prices have contributed substantially to the (cross-sectional) variance of relative price changes. Figure 6 shows the results of a variance decomposition into within- and between-group components: the variation arising due to price changes for products whose prices are free (within-group), that arising due to price changes for products whose prices are administered (within-group), and that caused by the interaction between free and administered prices (between-group). We find that around 40 percent of the time changes in administered prices were responsible for at least a quarter of the total monthly variance. In 2006, four consecutive petroleum product price adjustments in January, April, July, and October contributed 50 to 60 percent to the total cross-sectional variance in those months. Furthermore, the variation in administered prices driven by sharp changes exceeded that in free prices (in 7 of 59 months) and did so by a large margin. Thirtyeight percent of the cross-sectional variance observed during these 7 peaks together is accounted for by variation in administered prices compared to 25 percent by free prices.26 It follows that administered price changes account for a large share of the overall cross-sectional variance, especially during episodes of sharp price adjustment.

Figure 6.Decomposition of Price Variation

Source: Staff estimates.

57. The (Theil) variance and skewness are positively correlated with inflation over time (Figure 7).27 Based on moving-average (monthly) data, the correlation coefficients are 0.88 (0.61) between variance and inflation, and 0.69 (0.31) between skewness and inflation; and are highly statistically significant. Figure 6 also shows that skewness has been positive, and notably higher during fiscal year 2006/07. This suggests that large price spikes for several goods have co-existed with downward price rigidity for other goods. The variance and skewness of the price distribution are likely to have driven the inflationary process over the period, an issue we investigate econometrically in the next section.

Figure 7.Inflation, variance, and skewness

Source: Mauritius Central Statistics Office and Staff estimates.

D. An Empirical Analysis of the Inflation Process

58. We undertake an empirical analysis of the factors affecting inflation in Mauritius, focusing on basic relationships between inflation, monetary policy variables, and the nominal exchange rate. All data are drawn from the IMF International Financial Statistics, World Economic Outlook, and various issues of the BoM Monthly Bulletin. The sample contains monthly data for two periods: June 2003 to December 2006, and June 2003 to June 2007.28

59. The model includes the traditional regressors such as monetary variables and nominal exchange rate. First, we explain the inflationary process using monetary variables (shown in Appendix Figure 1): nominal interest rates and the growth rate of money supply (M2 net of offshore bank deposits). For interest rates, we consider the key policy rate (Lombard) and two de facto intervention rates (the interbank money-market rate and the average T-bill rate).29 Dynamics in the effect of broad money growth and the interest rate on inflation are accounted for with lags on these variables. For broad money growth, the most robustly significant lag is at 12 months. For the interest rates, we also consider a one year lag.30 Second, we account for the effect of the nominal exchange rate by including the rate of change of the NEER. This helps assess the impact of the exchange rate pass-through to changes in domestic prices.

60. In addition to standard variables, the model accounts for relative price variability through two indicators: the variance and skewness of the inflation distribution. The general specification is as follows:

where π represents the yearly end-of-period inflation rate, Var and Sk are the cross-sectional variance and skewness of the price change distribution (expressed as 12 month moving averages), it-12 is the interest rate lagged 12 months, (ΔM2)t-12 represents the rate of growth of broad money lagged 12 months, and Δ NEER represents the rate of change in the NEER. Unit root tests suggest that the series are difference-stationary (Appendix Table 1).

61. Strong unconditional relationships are apparent for inflation and the explanatory variables considered (Appendix Table 2). The matrix of unconditional correlation coefficients suggests a strong a positive relation between inflation and relative price variability (variance and skewness). Although the correlation coefficients between inflation and the various interest rates are not statistically significant, they have the expected sign. Moreover, a nominal effective exchange rate appreciation is correlated with a reduction of inflation.

62. The estimation is performed through OLS in a simple cointegrating framework (Table 5). To preserve degrees of freedom, the models discussed do not include seasonal dummies. As a robustness check, models with the seasonal dummies were estimated and the results held up (Appendix Table 3). For each model, we report the results of diagnostic residual tests—including serial correlation, normality, and unit roots—to ensure that statistical inference is valid.31

63. The main messages that emerge from the empirical analysis can be summarized as follows:

  • a. Relative price adjustment has a significant impact on inflation, as illustrated by the cross-sectional variance and skewness of price changes being strongly and positively correlated with inflation.32
  • b. Interest rates generally have significant and negative coefficients in the model, and work with a lag of 12 months. The Lombard rate, however, loses its effectiveness in the longer sample, whereas the de facto rates of intervention—the interbank and T-bill rates—tend to reduce inflation in both samples considered.
  • c. Broad money growth is associated with higher inflation with a lag of around 12 months.
  • d. An appreciation of the nominal effective exchange rate has a dampening effect on inflation, as illustrated by the negative and significant coefficient. A 1 percentage point NEER appreciation is associated with a 0.2 percentage point reduction in inflation.
  • e. These findings suggest that monetary control, including by allowing nominal appreciation, can help reduce inflation.

64. The results are generally stronger for the period before the abolishment of the Lombard rate (June 2003–December 2006) than the sample extended to July 2007. A possible explanation is that structural shifts in the model which may have occurred during 2007 are affecting the results. For example, in early 2007, the BoM announced several measures to improve the transparency and effectiveness of monetary policy, including the set-up of a Monetary Policy Committee (MPC) which held its first meeting in April 2007, and the initiation of conceptual work on the measurement of core inflation.

Table 5.The Determinants of Inflation: OLS Estimates(Dependent variable: Yearly end-of-period rate of change in the CPI)
Sample period:June 2003 – December 2006June 2003 – June 2007
Variance0.4610.4580.4030.5170.5290.452
(14.30)***(12.41)***(8.15)***(6.54)***(7.07)***(6.69)***
Skewness0.9240.8340.9460.7450.7100.857
(8.48)***(9.39)***(6.91)***(3.92)***(3.91)***(4.68)***
Lombard, t-12-0.632-0.283
(4.21)1(0.61)
Interbank rate, t-12-0.193-0.223
(4.48)***(1.97)***
T-bill rate, t-12-0.259-0.275
(2.61)***(2.04)***
ΔM2, t-120.4260.4250.3170.2170.2110.186
(8.12)***(8.11)***(4.54)***(2.24)***(2.21)***(1.72)***
ΔNEER-0.184-0.179-0.191-0.170-0.167-0.167
(9.46)***(9.72)***(9.04)***(3.94)***(4.22)***(4.15)***
Residual diagnostic tests
Serial correlation
Durbin Watson1.9862.0171.9801.9571.9081.909
Normality
Jarque-Bera p-value0.53860.34840.7660.08240.11950.0827
Unit root tests
Phillips-Perron p-value0.00000.00000.00000.02570.00810.0056
ADF p-value0.00000.00000.00000.03460.01140.0078
Observations414141474747
R-squared0.970.980.960.740.820.81
Source: Staff estimates.

Notes: t statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%. The standard errors are robust to heteroskedasticity of unknown form. Seasonal dummies are not included.

Source: Staff estimates.

Notes: t statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%. The standard errors are robust to heteroskedasticity of unknown form. Seasonal dummies are not included.

E. Conclusions

65. In this paper we have analyzed some aspects of inflation in Mauritius, focusing on an episode of sharp administered price changes in fiscal year 2006/07. Using monthly data for the period June 2002–July 2007 we have found that marked rises in administered prices of foods and fuels initiated in June 2006 led to increased relative price variability. This was reflected in the higher variance and positive skewness of the cross-sectional inflation distributions. Our estimate of the direct effect of this wave of price liberalization on inflation three months later is around 50 percent of total.

66. An empirical model of the determinants of inflation was estimated to understand the factors that drove inflation over the period. We find that nominal exchange rate appreciation dampens inflation. Furthermore, broad money growth and interest rates (the key policy Lombard rate, the interbank money-market, and the T-bill rate) affect inflation with a lag of approximately one year. The effectiveness of the Lombard rate is less evident in the second half of 2006, before a number of steps were taken to improve monetary policy, including the set-up of the Monetary Policy Committee and conceptual work on core inflation.

67. A key result is that relative price variability—to which sharp administered price changes are an important contributor—is strongly and positively correlated with inflation. Discrete price adjustments—common to administrative price regimes—change the shape of the individual price distribution. Depending on the relative magnitude of the individual price changes, the skewness and variance of the inflation distribution increase, driving up the overall inflation rate. The results suggest that the large discrete price changes associated with the administered price regime in Mauritius affect inflation through higher relative price variability. Frequent price adjustments could reduce the contribution of administered prices to overall price variability and the inflationary pressures associated with it.

68. The results underscore the importance of accounting for relative price variability in both inflation estimation and forecasting exercises. The BoM will need to incorporate the impact of expected price liberalizations in its inflation monitoring calculus. This can be achieved by coordinating the schedule of price liberalizations with the government, so as not to hamper the effectiveness of monetary policy in an environment of administered prices.

69. The analysis could be expanded by looking at a longer time span (especially to more recent periods after the monetary policy change), including additional explanatory factors, and using more sophisticated modeling. A longer time span would be necessary to better estimate monetary transmission lags and enrich the model with additional explanatory factors, such as world commodity prices. Furthermore, an analysis of level (cointegrating) relationships between the variables of interest and a vector error correction representation would be useful to assess short- and long-run dynamics. Impulse response functions in a vector autoregressive framework would also help investigate the projected inflation path for different monetary policy scenarios.

References

    BallL. andMankiwN. G.1994“Assymetric Price Adjustment and Economic Fluctuations,”Economic JournalVol.104No.423 pp. 247--61.

    BallL. andMankiwN. G.1995“Relative Price Changes as Aggregate Supply Shocks”,Quarterly Journal of EconomicsVol.110No.1 pp. 161-93.

    Bank of MauritiusAnnual Reportsvarious issues (Port Louis: Bank of Mauritius).

    Bank of Mauritius2007“Methodological note on core inflation” (Port Louis: Bank of Mauritius).

    BlinderA.S.1991“Why Are Prices Sticky? Preliminary Results from an Interview Study”American Economic ReviewVol.81 pp. 231-36.

    • Search Google Scholar
    • Export Citation

    CecchettiS.1985“Staggered Contracts and the Frequency of Price Adjustment,”Journal of EconomicsVol.100 (Supplement 1985) pp. 935-59.

    • Search Google Scholar
    • Export Citation

    Central Statistical Office2006External Trade—4th quarter 2005 (Port Louis, Mauritius).

    ClementsK. W. andP.Nguyen1981“Inflation and Relative Prices: A System-wide Approach,”Economic LettersVol.7No.2 pp. 131-37.

    CooreyS.M.MecagniE.Offerdal1996“Disinflation in Transition Economies: The Role of Relative Price Adjustment,”IMF Working Paper 96/138(Washington:International Monetary Fund).

    • Crossref
    • Search Google Scholar
    • Export Citation

    CooreyS.1997“Designing Disinflation Programs in Transition Economies: The Implications of Relative Price Adjustments” IMF Paper on Policy Analysis and Assessment PPAA/97/1 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    DebelleG. andOLamont1997) “Relative Price Variability and Inflation: Evidence from US Cities,”Journal of Political EconomyVol.105No.1 pp. 132-25.

    • Crossref
    • Search Google Scholar
    • Export Citation

    EngleR.F. (1982) “Autoregressive Conditional Heteroskedasticity with Estimates of the Variance of UK Inflation,”EconometricaVol.50No.4 pp. 987-1008.

    • Crossref
    • Search Google Scholar
    • Export Citation

    GhoshM. andJ.Whalley2004“Are Price Controls Necessarily Bad? The Case of Rice in Vietnam,”Journal of Development EconomicsVol.73No.1 pp. 215-32.

    • Crossref
    • Search Google Scholar
    • Export Citation

    GoodhartC.A.E.2001“Monetary Transmission Lags and the Formulation of the Policy Decision on Interest Rates,”Federal Reserve Bank of St. Louis Review (July/August) pp. 165-86.

    • Search Google Scholar
    • Export Citation

    HledikT.2003“Modelling the Second-Round Effects of Supply-Side Shocks on Inflation,”Czech National Bank Research Department Working Paper No. 2003/12 (Prague: Czech National Bank).

    • Search Google Scholar
    • Export Citation

    LachS. andD.Tsiddon1992“The Behavior of Prices and Inflation: An Empirical Analysis of Disaggregated Price Data,”Journal of Political EconomyVol.100No.2 pp. 349-89.

    • Crossref
    • Search Google Scholar
    • Export Citation

    Ministry of Industry Small & Medium EnterprisesCommerce & Cooperatives (Commerce Division) 2006) “White Paper on Price Control” (Port Louis Mauritius).

    • Search Google Scholar
    • Export Citation

    OkunA.1971“The Mirage of Steady Inflation,”Brookings Papers on Economic ActivityVol.2 pp. 435-98.

    ParksR.W.1978“Inflation and Relative Price Variability”Journal of Political EconomyVol.86 pp. 79-95.

    ParsleyD.1996“Inflation and Relative Price Variability in the Short Run and Long Run: Evidence from the United States,”Journal of Money Credit and BankingVol.28No.3 pp. 232-41.

    • Crossref
    • Search Google Scholar
    • Export Citation

    SchumacherL.2006“Inflation and Monetary Policy in Mauritius,”Mauritius: Selected Issues and Statistical Appendix IMF Country ReportNo. 06/224 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    UzagalievaA.2003“Price Liberalization Inflation and Regulated Prices in the Kyrgyz Republic: An Econometric Analysis” photocopy (Prague: Center for Economic Research and Graduate Education).

    • Search Google Scholar
    • Export Citation

    ViningD. R. andT. C.Elwertowski1976“The Relationship between Relative Prices and the General Price Level,”American Economic ReviewVol.66No.4 pp. 699-708.

    • Search Google Scholar
    • Export Citation

    Wozniakp.1999“Relative Prices and Inflation in Poland, 1989–1997: The Special Role of Administered Price Increases,”World Bank Policy Research Working PaperNo. 1879 (Washington:The World Bank).

    • Search Google Scholar
    • Export Citation
Appendix

Figure 1.Monetary Policy Variables: Interest Rates and Broad Money Growth, 2003–07

Source: Bank of Mauritius Monthly Bulletin; Mauritius Central Statistics Office, IMFInternational Financial Statistics; and Staff estimates.

Table 1.Unit root tests
Series:InflationVarianceSkewnessLombardInterbankT-billM2NEER
Sample: June 2003 – December 2006 (T=42)
Augmented Dickey-Fuller
level0.699-0.054-0.972-1.272-1.866-1.647-2.575-0.705
p-value0.9900.9540.7630.6420.3480.4590.0980.846
1st difference-4.615-4.670-5.343-4.223-7.270-3.421-8.195-3.974
p-value0.0000.0000.0000.0010.0000.0100.0000.002
Phillips-Perron
level0.159-0.201-1.181-1.357-1.751-1.667-2.554-1.401
p-value0.9700.9380.6820.6030.4050.4480.1030.582
1st difference-4.672-4.526-5.288-4.513-7.429-3.441-8.437-3.893
p-value0.0000.0000.0000.0000.0000.0100.0000.002
Sample: June 2003 – June 2007 (T=46)
Augmented Dickey-Fuller
level-0.759-0.576-1.301-0.869-2.062-1.764-2.632-1.938
p-value0.8310.8760.6290.7980.2600.3990.0870.314
1st difference-5.255-4.931-5.875-5.643-7.943-4.347-8.349-4.908
p-value0.0000.0000.0000.0000.0000.0000.0000.000
Phillips-Perron
level-0.996-0.730-1.464-1.064-1.927-1.790-2.604-2.307
p-value0.7550.8390.5510.7290.3190.3850.0920.170
1st difference-5.256-4.787-5.839-5.706-8.187-4.357-8.379-4.816
p-value0.0000.0000.0000.0000.0000.0000.0000.000
Source: Staff estimates.Note: Deterministic intercept and trend are included. Both tests’ null hypothesis is that of a unit root. In both cases, the MacKinnon approximate pvalue is reported.
Table 2.Correlation matrix
Sample: June 2003–December 2006
InflationVarianceSkewnessLombard rateInterbank rateT-bill rateΔ M2Δ NEER
Inflation1
Variance0.8532*
Skewness0.6290*0.5776*1
Lombard rate-0.0010.09470.4817*1
Interbank rate-0.2112-0.06910.24860.9046*1
T-bill rate-0.3541-0.30690.34680.7780*0.7845*1
Δ M20.5386*0.302-0.0958-0.2656-0.31-0.6019*1
Δ NEER-0.5899*-0.2316-0.18750.03430.13750.1864-0.4717*1
Sample: June 2003–June 2007
InflationVarianceSkewnessLombard rateInterbank rateT-bill rateΔ M2Δ NEER
Inflation*
Variance0.8762*1
Skewness0.6941*0.6620*1
Lombard rate0.27480.31850.5905*1
Interbank rate-0.0990.0240.30060.8296*1
T-bill rate-0.3074-0.27530.31660.6416*0.7644*1
Δ M20.5907*0.4262*0.11380.0661-0.1526-0.4961*1
Δ NEER-0.5747*-0.2603-0.1795-0.11330.05680.1719-0.5307*1

Note: * indicates significance at the 1 percent level. Interest rates and broad money growth are lagged 12 periods.

Note: * indicates significance at the 1 percent level. Interest rates and broad money growth are lagged 12 periods.

Source: Staff estimates.
Table 3.The Determinants of Inflation: OLS Estimates (Robustness check)(Dependent variable: Yearly end-of-period rate of change in the CPI)
Sample →June 2003 – December 2006Jun 2003 – June 2007
Model →1231’2’3’
Variance0.4580.4510.3420.5020.5180.452
(13.56)***(14.07)***(6.36)***(6.36)***(7^9)***(5.45)***
Skewness0.9360.8531.1170.7260.7310.840
(7.54)***(9.26)***(7.43)***(3.75)***(4.26)***(3.97)***
Lombard, t-12-0.666-0.157
(3.82)***(0.30)
Interbank rate, t-12-0.219-0.222
(4.74)***(1.70)***
T-bill rate, t-12-0.418-0.241
(3.84)***(1.23)
ΔM2, t-120.4260.4090.2000.2050.2050.175
(8.13)***(8.29)***(2.99)***(1.77)***(1.82)***(1.29)
ΔNEER-0.189-0.188-0.212-0.194-0.188-0.190
(8.61)***(10.51)***(15.74)***(4.47)***(4.51)***(4.36)1
Residual diagnostic tests
Serial correlation
Durbin Watson2.01282.11752.01912.00231.96671.9871
Normality
Jarque-Bera p-value0.35940.37780.01070.78610.09090.0812
Unit root tests
Phillips-Perron p-value0.00000.00000.00000.01450.00370.0049
ADF p-value0.00000.00000.00000.01720.00520.0061
Observations414141474747
R-squared0.980.990.980.770.850.83

Notes: t statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%. The standard errors are robust to heteroskedasticity of unknown form. Seasonal dummies are included.

Notes: t statistics in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%. The standard errors are robust to heteroskedasticity of unknown form. Seasonal dummies are included.

Source: Staff estimates.
15Prepared by Fabian Bornhorst and Camelia Minoiu.
16For early contributions that use disaggregated goods data in this type of analysis, see Parks (1978), Lach and Tsiddon (1992), Parsely (1996), and Debelle and Lamont (1997).
17In July 2007, the CSO revised its CPI methodology using weights for goods and services derived from the 2006/07 Household Budget Survey. For this reason our analysis ends in June 2007.
18The relative merits of different policy responses to the current global conditions are not the object of this study. Rather, we seek to understand the role of administered prices and that of monetary policy and the nominal exchange rate in driving the inflation process in Mauritius.
19All of Mauritius’ cement requirements are imported. In 2005, cement imports represented around 1 percent of total imports (CSO, 2006).
20A partially offsetting measure was the reduction of customs duties for alcoholic beverages and cigarettes.
21Neither of the two methods for computing the average rise in prices is without shortcomings. In the first case, we use actual price increases, but the weights may correspond to broader categories of goods. This is the case with PET bottles which are used in the soft drinks industry: while we observe the administered price increase (4.9 percent), the weight of PET bottles in the CPI is not known. Instead, we use the weight of soft drinks in the CPI. In the second case, neither the price increase nor the weight is “correct”, but they are good approximations. In our example, the weight and the CPI inflation rate for soft drinks is imputed to determine the contribution of PET bottle price increase to the overall inflation rate.
22The contribution of the June 2006 price liberalizations to September inflation (yearly, end of period) was 45 percent. We do not extend the analysis beyond September 2006 because October witnessed another wave of price adjustments for petroleum products.
24For exceptions and evidence consistent with a negative relation between relative price variability and inflation, see, e.g., Cecchetti (1985), Blinder (1991), and Lach and Tsiddon (1992).
25See, e.g., Ghosh and Whalley (2004) for Vietnam, and Clements, Jung, and Gupta (2003) for Indonesia.
26The 7 months over the sample period when the variation in administered prices exceeded that in free prices (and the main goods whose prices were subject to change) are: February 2004 (bread, flour, cooking gas), April 2004 (gasoline and Diesel oil), July 2004 (gasoline and Diesel oil); November 2005 (cooking gas, kerosene, and Motor oil); April (gasoline and Diesel oil); July 2006 (see Table 2); and January 2007 (kerosene, gasoline, Diesel oil, and Motor oil).
27In our analysis, the variance and skewness are weighted by the CPI basket weights. For formulas, see, e.g., Appendix I in Coorey, Mecagni, and Offerdal (1996).
28Our sample period starts in June 2003 to avoid the use of linking coefficients in deriving comparable end-period inflation figures prior to that date, as the CPI basket was changed in June 2002 based on the 2001/02 Household Budget Survey. The second sample period (June 2003 – June 2007) is also considered because it overlaps with a period of monetary policy transition, notably from the use of the Lombard rate and the repo rate for intervention.
29The Lombard rate is a standing facility introduced in 1999 as a lender of last resort for commercial banks to meet unexpected liquidity shortfalls. Until 2002, the T-bill rates moved in tandem with the Lombard rate, suggesting that the latter was an effective signal of monetary policy. As this relationship gradually broke down, the Lombard rate was replaced in December 2006 by the repo rate, which targets the overnight interbank money market rate.
30See Goodhart (2001) for a discussion of the expected length of monetary policy transmission lags.
31We emply the Cochrane-Orcutt transformation to correct for serial correlation; the residuals therefore pass the serial correlation test. Similarly, the Jarque-Bera test indicates that there is no evidence against the null hypothesis of normality. Finally, the Philipps-Perron and the Augmented Dickey-Fuller (ADF) tests reject the hypothesis of unit roots in the residuals.
32We have also considered models in which each measure of relative price variation has been included in the model alone, but the main result—i.e., that each variable is strongly and positively correlated with inflation—holds up despite the high correlation between the two.

Other Resources Citing This Publication