Journal Issue
Share
Article

Mongolia: Selected Issues and Statistical Appendix

Author(s):
International Monetary Fund
Published Date:
January 2007
Share
  • ShareShare
Show Summary Details

I. Determinants of Inflation in Mongolia1

A. Introduction

1. Despite robust demand growth, inflation in Mongolia has been generally moderate in recent years. Inflation, as measured by the end-of-period consumer price index (CPI)2 averaged 7.4 percent during 2000-05 and is estimated at 7.0 percent in 2006. The moderate inflation in the 2000s’ is in a sharp contrast with the 1990s. In the early 1990s, when Mongolia recorded triple digit inflation in 1993-94 and an average inflation rate of nearly 20 percent during 1995-2000. The moderation of inflation in recent years has taken place despite rapid monetary expansion. During 2000-05, broad money expanded by 32.4 percent and reserve money increased by 16.8 percent, rates much higher than inflation.

2. These somewhat surprising developments reflect the variety of factors—both real and monetary—that are found to influence inflation in Mongolia. It is widely agreed that the relative inelasticity of food supply, especially meat products in Mongolia, is a key contributor to the inflationary process, regardless of money supply developments.3 Import prices, especially for petroleum products, are also potential sources of inflationary pressures, given the large share of imported goods and raw materials in Mongolia’s consumer basket take. Changes in world commodity prices (e.g., copper and gold) can also indirectly lead to inflation changes in both directions. That is, a surge in foreign exchange inflows from commodity price hikes brings about liquidity expansion, thereby inflationary pressures, while it could partly offset inflationary pressures, given that exchange rate appreciation arises due to the capital inflows. Loose government spending (e.g., wages, pensions, and social transfers) may cause potential inflationary impacts as well, although those factors are not the main focus in this chapter.

3. The extent to which low inflation can be sustained under rapid monetary expansion and high mineral prices remains uncertain. This chapter analyzes the main determinants of inflation in Mongolia using empirical tests based on a structural model approach and VAR (Vector Autoregression Model), with a view to assessing whether inflation is predominantly affected by commodity prices or by money supply developments. To that end, simulation and impulse response analyses are used to estimate the components of the inflation dynamics under various exogenous shocks. Although inflation is strongly influenced by exogenous shocks to meat, copper, and oil import prices, we find that monetary expansion and changes in exchange rate are also critical determinants of inflation.

B. Recent CPI and Monetary Aggregates Developments

4. Inflation was relatively volatile in 2005 due to a severe shortage of meat supply, reaching its recent peak of 17.6 percent (y/y) in June 2005. As the supply shock was reversed from the third quarter of 2005, inflationary pressures abated quickly, with the CPI increasing by only 1.8 percent (y/y) in July 2006, the lowest in four years. However, with the waning influence of the supply shock, inflation picked up to 6.8 percent (y/y) at end-November 2006. The exchange rate appreciation due to the strengthening of the external position, has also contributed to suppress inflationary pressures. However, core inflation as of November was 6.7 percent (y/y), a relatively high increase in comparison to the overall CPI. This may imply that inflationary pressures driven by the demand side of the economy is significant this year.

Inflation and Broad Money

Inflation and Broad Money(In percent)
2000200120022003200420052006
CPI 1/8.17.91.74.710.69.26.82)
Core Inflation 1/-4.31.35.28.03.56.72)
Broad Money17.528.041.949.720.337.338.34)
Togrog/U.S.$ 3/-2.2-0.5-2.0-3.8-3.2-1.0+4.92)5)

Year on year.

End-November.

End of period, (+)/(-) indicates an appreciation/depreciation.

End-October.

Since end-2005.

Year on year.

End-November.

End of period, (+)/(-) indicates an appreciation/depreciation.

End-October.

Since end-2005.

Meat and Oil Prices, and Exchange Rate

(12-month percent change)

Sources: Mongolian authorities; and IMF, International Financial Statistics.

5. The limited inflationary response to rapidly increasing money supply largely reflects a still strong monetization process, which is still underway. In other words, as income rises, the private sector expands with large increase in deposit into and borrowing from banking system. Total deposit in the banking system in relation to nominal GDP almost doubled during 2000 and 2006, while the money multiplier steadily increased from 1.9 in 2000 to 4.8 in 2006. Since 2000, the pace of credit expansion to private sector seems to outpace that of deposit increases in the banking system.

Indicators of Remonetization(In percent)
20002001200220032004200520061/
Broad Money/GDP25.429.637.847.743.946.747.1
Money multiplier 2/1.92.32.73.53.64.14.8
Deposit/GDP15.519.928.138.836.540.640.9
Loan/GDP6.812.419.030.533.536.437.3

End-September.

Broad money/Reserve money.

End-September.

Broad money/Reserve money.

6. The degree of remonetization seems more pronounced in Mongolia than in other Commonwealth of Independent States (CIS) countries. At end-2005, credit to the private sector in Mongolia as share of GDP reached its highest level of 37.5 percent, compared with 36.3 percent in Kazakhstan, 33.5 percent in Ukraine, and much lower levels in other countries. Furthermore, the pace of the credit expansion to the private sector is the fastest in Mongolia during the period 2000-05, increasing by almost fivefold in five years.

Credit to Private Sector in CIS Countries

C. Empirical Results

Structural Model

7. To identify the dominant factors affecting inflation, six endogenous variables (CPI, broad money, reserve money, exchange rate, export and import) are estimated by simple OLS regressions in a structural model. Quarterly data are used during the sample period of 1993 Q1-2006 Q1. To consider a strong seasonality in the Mongolian economy, seasonal dummy variables are included (Box I.1).

8. The coefficients of the explanatory variables in the six estimation equations show expected signs in most cases (see Annex). Inflation is positively related to increases in broad money, increases in copper and oil prices and the depreciation of togrog. The statistical significance of the coefficients, however, is not robust except for the exchange rate. Broad money is positively affected by an increase in reserve money and the second quarter seasonal dummy variable. However, broad money is negatively affected by the depreciation of the exchange rate. Reserve money supply is positively related to the increase in lending interest rates, reflecting high money demand in the money market.4 The exchange rate is influenced by the supply and demand of foreign exchange, which is mostly envisaged in exports and imports, respectively.5 The exports are positively affected by increase in copper price and exchange rate depreciation, while the imports are insignificantly influenced by world oil prices.

Box I.1.Empirical Method Using the Structural Model

Data

  • Estimation Period: 1993 Q1–2006 Q1
  • Frequency: Quarterly
  • Number of observations: 53
  • All variables are log-transformed.
  • Seasonal dummy variables are included to reflect high seasonality.

Structure of the Model

  • Six endogenous variables, including CPI, broad money(BM), reserve money(RM), nominal exchange rate(ER), export(EX) and import(IM) are estimated using OLS. The composition of the structural model is as follows:
  • CPI = f(Constant, BM, ER, OIL(-1), COPP, CPI{-1})
  • BM = f(Constant, RM, ER, DUM2, BM{-1})
  • RM = f(Constant LR, D2, RM{-1})
  • ER = f(Constant, EX, IM, ER{-1})
  • EX = f(Constant, COPP, GOLDP, ER, DUM4, EX{-1})
  • IM = f(Constant, OIL, EX, CPI, IM{-1})
Endogenous VariablesExogenous Variables
CPIConsumer Price IndexOILOil price (Brent)
BMBroad MoneyCOPPCopper price
RMReserve MoneyGOLDPGold price
ERExchange Rate (period average)LRLending Rates
EXExportDUM22nd quarter dummy variable
IMImportDUM44th quarter dummy variable
  • After applying the RMSE test for verifying the stability of the structural model, a policy simulation is applied, in which a certain shock is generated to see the dynamic impacts of each variable on inflation within the structural model.
  • The simulation period is 2001 Q1–2006 Q1.

9. The simulation results of this structural model sufficiently fit the statistical criteria of RMSE6for each estimation equation. RMSEs in each endogenous variable are less than 5 percent, which is in general regarded as a benchmark. The results of the RMSE are as follows:

Results of RMSE
Endogenous variablesRMSE (percent)
CPI1.4184
Broad Money0.5426
Reserve Money0.4823
Exchange Rate0.9622
Exports4.3878
Imports4.2135

10. The empirical results of the policy simulation indicate that inflation in Mongolia is positively affected by increases in copper and oil prices, money growth, and the depreciation of exchange rate. Increases in copper prices have a larger impact on inflation than increases in oil prices. It is estimated that a 10 percent increase in copper prices as a permanent shock causes a 1.29 percent increase in price level (annual average), while a 10 percent increase in oil prices results in a 0.36 percent increase in price level. The impact of gold prices on inflation, however, seems negligible. It is confirmed that the money growth also has significant impact on inflation. It is estimated that a 5 percent increase in reserve and broad money increases the price level (annual average) by 0.27 percent and 0.28 percent, respectively.

Result of Policy Simulations on Inflation(In percent)
Permanent ShocksImpact on CPI
1st year2nd year3rd year4th year5th yearAnnual

average
10 % increase in copper price0.831.251.391.471.511.29
10 % increase in gold price-0.05-0.04-0.04-0.03-0.03-0.04
10 % increase in oil price0.170.360.410.430.440.36
5 % increase in reserve money0.060.190.300.380.440.27
5 % increase in broad money0.180.270.300.320.330.28
1 % appreciation of Togrog-0.70-0.94-0.96-0.95-0.95-0.90

Impacts of Selected Shocks on CPI

(In percent)

VAR Model

11. The togrog exchange rate seems to have the largest influence on inflation, implying that the recent appreciation of the togrog may be the main factor behind low inflation in recent months. It is estimated that a 1 percent appreciation of the togrog causes a 0.90 percent decrease in inflation (annual average). It is also noticeable that the impact of the exchange rate shock on inflation occurs with a short time lag, while other shocks seem to have a more gradual impact.

12. The impulse response analyses in the VAR model, which shows a dynamic process of a specific variable to various shocks, confirms broadly the results of the structural model (Box I.2). A one percent standard deviation shock on variables, such as copper prices, oil prices, money supply and the exchange rate affects inflation significantly. Increases in copper prices affect inflation with about a three-month lag, while increases in oil prices have a more rapid impact on inflation. The length of the response, however, is more lasting for copper price changes than oil price changes. The nominal exchange rate depreciation shock has an instant and persistent impact on inflation, confirming the results of the structural model. The money growth shock also seems crucial in affecting inflation. It appears that reserve money growth affects the inflation with about a two-quarter lag, compared with a three-month lag for the increase in broad money.

Box I.2.Empirical Method Using VAR Model

Data

  • Estimation Period: 1995 Dec.- 2006 May
  • Frequency: Monthly
  • Number of observations: 126
  • All variables are log-transformed.

Model Specification

  • It consists of 6 variables, including CPI, exchange rate, copper prices, oil prices, reserve money and broad money, based on the empirical results in the above structural model.
  • - The order of the variables (copper prices, oil prices, exchange rate, reserve money, Broad money and CPI) are determined by the exogenity of the variables.
  • - The lags of the VAR model are set as 2 by the AIC.

Impulse Response of CPI to Each Shock

13. Variance decomposition analysis, indicates that the exchange rate is the predominant factor in explaining the inflation. Changes in the exchange rate explain more than 20 percent of inflation and the impact seems to be persistent. Copper price changes show the importance in CPI changes, as is in the simulation results of the structural model. Reserve money growth also provides explanatory power, with more than 14 percent of the change in inflation related to a reserve money shock one year before.

Variance Decomposition
LagsCPICOPPOILERRMBM
1100.00.00.00.00.00.0
298.20.20.10.20.31.0
392.00.80.62.73.10.9
479.81.02.48.77.50.7
660.61.55.122.19.51.2
851.36.65.026.48.81.9
1044.812.34.524.611.72.2
1240.815.64.622.714.02.3
1439.317.34.722.014.32.4
1638.518.54.522.013.82.6

D. Concluding Remarks

14. Inflation in Mongolia is largely affected by food supply constraints, especially meat products, and mineral commodity prices, mostly copper and oil import prices. However, empirical results also indicates that the increase in money supply and changes in the exchange rate have very significant impacts on inflation. In particular, recent low and stable inflation is mostly attributable to the exchange rate appreciation, despite recent commodity price hikes and sharp increases in money supply.

15. Recent inflation trends, however, might not be sustainable unless monetary policy is implemented in a more prudent way to cushion the effects of high mineral prices and the consequent economic boom. In the pursuit of lowering inflation in the medium-term to around 5 percent, as called for under 2007 Monetary Policy Guidelines, the monetary authorities may need to lower reserve money growth further, thereby that of broad money. Inflationary pressures from excessive money supply generally materialize with some time lags and become more difficult to control when the exchange rate starts depreciating, coupled with the impacts of loose fiscal policy, or adverse weather conditions. Accordingly, monetary policy should be pre-emptive to prevent renewed inflation, in line with central bank’s ultimate goal.

Annex: Estimation Results of the Structural Model

adjusted R2 = 0.983, D.W.= 1.98

adjusted R2 = 0.993, D.W.= 1.20

adjusted R2 = 0.983, D.W.= 2.27

adjusted R2 = 0.977, D.W.= 1.34

adjusted R2 = 0.674, D.W.= 1.60

adjusted R2 = 0.674, D.W.= 1.60

* Figures in parenthesis indicate t-values.

References

    AndersonJ. and D. A.Citrin1995“The Behavior of Inflation and Velocity,” in Policy Experiences and Issues in the Baltic Russia and Other Countries of the Former Soviet Unioned. ByD. A.Citrin and A. K.LahiriIMF Occasional PaperNo. 133(Washington:International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    BalkeN. and K.Emery1994“Understanding the Price Puzzle,”Economic and Financial Policy Review (4th Quarter)Federal Reserve Bank of Dallas.

    • Search Google Scholar
    • Export Citation

    BarroR. J. and H. I.Grossman1974“Suppressed Inflation and Supply Multiplier,”Review of Economic StudiesVol. 41No. 1pp. 87104.

    BrissmissS. and N.Magginas2004“Forward-Looking Information in VAR Models and Price Puzzle,”Bank of Greece Working PaperNo.10(February).

    • Search Google Scholar
    • Export Citation

    DeGrauwe Paul and MagdalenaPolan2005“Is Inflation Always and Everywhere a Monetary Phenomenon?,”in Scandinavian Journal of EconomicsVol. 107No. 2pp. 23959.

    • Crossref
    • Search Google Scholar
    • Export Citation

    KhanS. M. and A.Schimmelpfennig2006“Inflation in Pakistan: Money or Wheat?,”IMF Working Paper 06/132(Washington:International Monetary Fund).

    • Crossref
    • Search Google Scholar
    • Export Citation

    KuijsA.1999“Determinants of Inflation, Exchange Rate and Output in Nigeria,”IMF Working Paper 99/160.

    LiuO. and O.Adedeji2000“Determinants of Inflation in the Islamic Republic of Iran: A Macroeconomic Analysis,”IMF Working Paper 00/127(Washington:International Monetary Fund).

    • Crossref
    • Search Google Scholar
    • Export Citation

    MohantyM.S. and MarcKlau2001“What determines Inflation in Emerging Market Economies?”Bank for International Settlement PapersNo. 8.

    • Search Google Scholar
    • Export Citation

    MunozS.2005“High Inflation and Money Demand,” in Zimbabwe: Selected Issues and Statistical Appendix IMF Country ReportNo. 05/359(Washington:International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    OomesNienke and FranziskaOhnsorge“Money Demand and Inflation in Dollarized Economies: The Case of Russia,”IMF Working Paper 05/144(Washington:International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    PortesR. and D.Winter1978“The Demand for Money and For Consumption Goods in Centrally Planned Economies,”Review of Economics and StatisticsVol. 60No. 1

    • Search Google Scholar
    • Export Citation
1Prepared by SeungHo Lee.
2Since April 2006, Mongolia’s NSO (National Statistical Office) has expanded the number of items in the CPI basket from 239 to 287 and changed the base year to December 2005=100.
3Mongolia’s consumer basket includes meat and meat products (12.5 percent), dairy products (5.2 percent), and fruits and vegetables (9.0 percent), which are excluded in Fund staffs estimation of core inflation. Even though core inflation is a proper alternative to assess the monetary impact on inflation, in this chapter the overall CPI was used as an inflation proxy because of data limitations and its importance as a key monetary policy target.
4Central Bank Bill rates instead of lending rates could be a good alternative as an explanatory variable for reserve money. However, because of the lack of interest rate signaling channel of CBBs as a policy instrument, lending rates are used to capture the magnitude of money demand in the economy.
5Because of the inaccessibility of quarterly data in the sample period, foreign remittances were not included in the estimation.
6RMSE = SQRT((1/T)*Σ(Ys-Ya)/Ya)2* 100, where T: number of observations in simulation period, Ys: simulated value, Ya: actual value.

Other Resources Citing This Publication