Information about Asia and the Pacific Asia y el Pacífico
Chapter

V Determinants of Inflation

Author(s):
Krishna Srinivasan, Erich Spitäller, M. Braulke, Christian Mulder, Hisanobu Shishido, Kenneth M. Miranda, John Dodsworth, and Keon Lee
Published Date:
March 1996
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

Vietnam has had remarkable success in bringing inflation under control. In the wake of policies that the country has pursued, the increase in consumer prices declined from 394 percent in 1988 to 5 percent in 1993. Thereafter, however, despite continuing commitments to consolidate the gains that had been made, inflation again accelerated. By the end of 1994, inflation stood at 14 1/2 percent, and there are indications that its pace increased further, to about 20 percent, in the first half of 1995. The debate about the appropriate policy response has raised questions about the factors involved in inflation and whether the recent acceleration is likely to be temporary or protracted.

The objective of this section is twofold: (1) to examine the influence of money on inflation, with regard to both its earlier trend decline and its reacceleration; and (2) to address other determinants of inflation that, in addition to money, appear to have affected its behavior. The discussion focuses solely on the inflationary repercussions of the recent sharp increases in rice prices in the wake of shocks to domestic supply and higher international prices.

Results of the analysis indicate a significant relationship between money and inflation. An increase in liquidity leads to higher inflation with a lag of one to five months, with the maximum impact involving a one- to two-month lag. This confirms that the early introduction and consistent implementation of tight financial policies was a key element of Vietnam’s success in bringing inflation under control.

Results of the analysis also confirm that changes in the relative prices of rice are positively related to inflation in overall consumer prices in the short run. This phenomenon was particularly noticeable in the reacceleration of inflation in late 1994. The argument is that the inflationary impact of the surge in rice prices has not been fully offset by a weakening of other prices, owing to price rigidities and inertia of consumption behavior. The consequence of this would be a rise in inflation above trend, associated with an increase in velocity. The inflation late last year was thus sparked initially by rice prices and subsequently intensified under the impact of monetary expansion.

Money and Inflation

Money and inflation have exhibited broadly similar trends, at least since the general liberalization of prices and the exchange rate (Chart 5.1). To examine more closely the link between money and inflation, we posit the following relationship:

Chart 5.1.Money and Inflation

(Year-on-year; changes in percent)

Source: General Statistical Office.

where the change in money stock in period t a function of the difference between real money demand in period t (RMD) and money stock in period t - 1. The above equation can be simplified and expressed in terms of rates of change:

For lambda equal to unity, we have P* = M* - RMD*, which relates changes in prices to differential growth in the money stock and real money demand.

We assume here, in the absence of monthly GDP data, that real money demand follows a rising trend, reflecting the general expansion of the economy and a general increase in the attractiveness of the dong as an asset since 1990. Allowing for concurrent and lagged influences of changes in money on prices, we thus have

where -c0 is a constant term reflecting the trend growth in real money demand, and L is a lag operator; the sum of lagged coefficients should approximate unity.

We estimate equation (3) for the period December 1990-June 1995 using a traditional Almon polynomial lag model with a Prais-Winsten correction for autocorrelation.1 We incorporate eight lags and assume that the lag coefficients can be approximated by a third-order polynomial. Experiments with specifications that include lags of less than eight are also carried out for purposes of comparison. To examine the effect on inflation of a change in consumer spending patterns during the Tet festival season, we include dummy variables in the regression exercise. Hence, the precise form of the estimating equation is

where Pt* and Mt* represent month-to-month percentage changes in consumer prices and the money stock (contemporaneous and with lags), respectively. Tet1, Tet2, and Tet3 represent the Tet festival dummies, which take the value of one for the month of December, January, and February and zero for all other months. To carry out the empirical analysis, we use monthly data on consumer prices and money stocks, including both dong liquidity and foreign currency deposits.

Results (Table 5.1) indicate that an increase in liquidity leads to higher inflation with a lag of one to five months, with contemporaneous money growth having no impact and money growth with a one- to two-month lag having the maximum impact. The constant term representing real money demand has the expected negative sign. As indicated by the Tet dummy variables, an increase in consumer spending during January and February significantly affects inflation. The results suggest that the pursuit of tight monetary policy during the transition to a market economy has played a decisive role in Vietnam’s success in curbing inflation.

Table 5.1.Inflation and Money: Regression Results, December 1990-June 1995
Coefficient and t-ratio (within parentheses)
VariableIIIIII
Constant−1.028(1.7)−0.825(1.4)−0.629(0.8)
M0.090(1.4)0.091(1.4)0.093(1.4)
M10.113(2.7)0.132(3.2)0.140(3.2)
M20.116(2.7)0.135(3.1)0.139(3.1)
M30.105(2.7)0.113(2.9)0.110(2.9)
M40.087(2.5)0.082(2.2)0.072(1.8)
M50.069(1.9)0.054(1.4)0.046(1.1)
M60.058(1.5)0.045(1.1)0.050(1.0)
M70.060(1.6)0.068(1.3)
M80.083(1.6)
Tet10.451(1.0)0.512(1.0)0.495(0.7)
Tet21.603(2.7)1.590(2.7)1.503(2.5)
Tet33.023(5.7)2.975(5.6)2.891(5.3)
Adjusted R squared0.600.590.58
Durbin-Watson12.202.202.10
Sum of lag coefficients20.780.720.65
Note: Dependent variable: Month-to-month percentage change in consumer price index. Independent variables: Month-to-month percentage change in total liquidity (contemporaneous and with lags).

A Durbin-Watson statistic close to 2.0 indicates the absence of serial correlation.

The sum of lag coefficients is an estimate of the long-run multiplier and will approximate to unity as lag length increases.

Note: Dependent variable: Month-to-month percentage change in consumer price index. Independent variables: Month-to-month percentage change in total liquidity (contemporaneous and with lags).

A Durbin-Watson statistic close to 2.0 indicates the absence of serial correlation.

The sum of lag coefficients is an estimate of the long-run multiplier and will approximate to unity as lag length increases.

Despite the reasonably high explanatory power of the model, the fitted path for inflation deviates from the actual path of inflation at certain times. In particular, we find that actual inflation is well above fitted values in the latter part of 1994 and the early months of 1995. To further analyze this issue, we estimated equation (4) using observations through the third quarter of 1994. The results from this estimation were used to make projections for the fourth quarter of 1994 and the first five months of 1995. It can be seen (Table 5.2) that our model projects a lower rate of inflation than the actual rate for the same period.

Table 5.2.Inflation and Money: Actual Versus Projected Inflation
TimePeriodActual InflationProjected Inflation1
1994October1.30.1
November1.7−0.1
December1.10.5
1995January3.81.1
February3.42.7
March0.2−0.1
April1.10.1
May1.80.3
Note: Dependent variable: Month-to-month percentage change in consumer price index, December 1990-September 1994.Independent variables: Month-to-month percentage change in total liquidity (contemporaneous and with lags), December 1990-September 1994.

Projections based on estimation of equation (4) for the period December 1990-September 1994.

Note: Dependent variable: Month-to-month percentage change in consumer price index, December 1990-September 1994.Independent variables: Month-to-month percentage change in total liquidity (contemporaneous and with lags), December 1990-September 1994.

Projections based on estimation of equation (4) for the period December 1990-September 1994.

Box 5.1.Money and Prices: A Long-Term Relationship

A long-term relationship between prices and money could be established using cointegration techniques. To examine a cointegrating relationship between prices and money, the stationarity of these two individual series needs to be examined. Stationary series have a tendency to return to their original value after a random shock, and the mean and variance of such series do not change with the passage of time (see Eken and others, 1995). Stationarity can be achieved through (1) differencing the data; or (2) formulating a regression model with nonstationary variables, a linear combination of which produces stationary error terms.

Tests for stationarity (that is, Unit Root tests; see table below) indicate that logarithmic values of both prices and money, when taken as a difference over the previous month’s values, are stationary. Furthermore, residuals of a least-squares estimation of the logarithmic value of prices on the logarithmic value of money, the cointegrating equation, are found to be stationary. Thus, money and prices tend to move together over the long run and revert to an equilibrium relationship in response to shocks (see Chart 5.2).

Unit Root Tests for Stationarity

Dicky-FullerCritical Value
VariableStatistic5 percent10 percent
LPrice−3.168−3.497−4.142
LMoney−1.928−3.495−4.138
Δ LPrice−5.871**−3.497−4.142
Δ LMoney−5.472**−3.499−4.146
Residual−1.955*−1.947−2.607
Note: ** and * indicate statistical significance at the 5 percent and 10 percent levels, respectively. A means first differencing.
Note: ** and * indicate statistical significance at the 5 percent and 10 percent levels, respectively. A means first differencing.

Chart 5.2.Cointegrating Relationship Between Money and Prices

Source: General Statistical Office; and IMF staff estimates.

It is evident from the above discussion that, despite a significant relationship between money and prices, there have been spurts in prices that cannot be explained by changes in money. In the last quarter of 1994, inflation rose rapidly despite relatively tight liquidity growth in the second and third quarters of the year. This acceleration of inflation was associated with a sharp increase in rice prices—the topic we discuss next.

Rice Prices and Inflation

The Vietnamese authorities have attributed the worsening of inflation from the latter part of 1994 to soaring rice prices, pushed up by natural calamities and rising international prices. In August 1994, the worst rains in seven years flooded eight northern provinces, destroying large amounts of crops. Estimates by the Ministry of Agriculture put output losses in the Red River Delta at up to 1 million tons of rice and other crops. According to the ministry, the price effects of those output disruptions were exacerbated by hoarding and adverse expectations. In addition, there were said to be repercussions on domestic rice prices from increases in the unrecorded cross-border exports of rice to China, which was similarly affected by the weather. At the Chinese border, prices for plain rice were approaching D 4,000 a kilogram by the end of 1994, double their level in the north of Vietnam. By contrast, in the south, despite heavy weather-related losses, output in the Mekong Delta proved strong, boosting national rice output to an estimated level of 23.5 million tons, 3.0 percent higher than in 1993. Nevertheless, market imperfections prevented the alleviation of the output shock in the north. Moreover, strong increases in the international prices of fertilizer and rice coincided with the domestic output shocks (Table 5.3).

Table 5.3.International and Domestic Prices of Fertilizer and Rice

(Annual average changes in percent)1

19901991199219931994
International prices (U.S. dollars)
Fertilizer14.79.6−18.4−23.938.5
Rice (Bangkok)−10.38.8−8.1−6.833.6
Domestic prices (U.S. dollars)
Rice export unit value−15.418.8−5.5−1.99.3
Fertilizer
Nam Ha10.110.0−6.3−20.115.4
Quangnam-Da Nang18.613.6−14.9−24.315.4
An Giang31.97.0−14.0−21.220.3
Rice retail
Hanoi3.435.8−20.0−5.623.2
Nam Ha11.431.5−23.4−5.622.2
Quangnam-Da Nang26.211.4−14.92.710.6
Ho Chi Minh City21.517.6−11.57.95.7
An Giang33.412.0−15.74.8−4.0
Exchange rate (dong-U.S. dollar)31.180.617.5−3.63.1
Sources: State Pricing Committee, General Statistical Office; State Bank of Vietnam; and IMF, International Financial Statistics.

The main rice-producing areas are Nam Ha in the north, Quangnam-Da Nang in central Vietnam, and An Giang in the south. The main urban areas are Hanoi and Ho Chi Minh City.

Sources: State Pricing Committee, General Statistical Office; State Bank of Vietnam; and IMF, International Financial Statistics.

The main rice-producing areas are Nam Ha in the north, Quangnam-Da Nang in central Vietnam, and An Giang in the south. The main urban areas are Hanoi and Ho Chi Minh City.

The influence of the domestic and international shocks on the behavior of Vietnamese rice prices, most notably in the north (Nam Ha Province and Hanoi), is also reflected in the prices of consumer staples and in total consumer prices2 (Charts 5.3 and 5.4 and Tables A15 and A16).

Chart 5.3.Consumer Prices of Staples by Region

(In percent; December 1993 = 100)

Source: General Statistical Office.

Chart 5.4.Consumer Prices by Component

(Cumulative changes in percent)

Source: General Statistical Office.

One prominent feature worth emphasizing is the steep increase in relative staples prices and the apparently associated acceleration of inflation. From a theoretical point of view, inflation and movements in relative prices are not necessarily related in any systematic manner. Indeed, no such sweeping claim is made. The claim that is made is that, in Vietnam, the increase in staples prices was not fully offset by a weakening of other consumer prices, in the presence of price rigidities and a persistent consumption pattern, and this involved an increase in overall prices.

Relative Prices and Inflation

To examine the significance of the relationship between inflation and the price of staples relative to other goods in the context of the polynomial lag model, equation (4) is reestimated, including the relative price of staples (RPS) and other goods together with the other explanatory variables. It is clear from the results (Table 5.4) that, along with an increase in the explanatory power of the model, the relative price variable is highly significant in explaining inflation. An increase in the price of staples translates into a shift away from savings toward higher expenditure, which even with a constant money stock leads to an increase in inflation through an increase in the velocity of money.

Table 5.4.Inflation, Money, and Relative Prices: Regression Results
Coefficient and t-ratio (in parentheses)
VariableIII
Constant−1.028(17)−1.240(1.0)
M0.090(1.4)0.106(1.5)
M10.113(2.7)0.118(2.9)
M20.116(2.7)0.120(2.9)
M30.105(2.7)0.114(2.9)
M40.087(2.5)0.104(2.9)
M50.069(1.9)0.093(2.5)
M60.058(1.5)0.084(2.1)
M70.060(1.6)0.081(2.2)
M80.083(1.6)0.087(1.8)
Tet10.451(1.0)0.503(1.2)
Tet21.603(2.7)1.202(2.3)
Tet33.023(5.7)3.185(7.3)
RPS0.214(3.8)
Adjusted R squared0.600.64
Durbin-Watson12.202.10
Sum of lag coefficients20.780.86
Note: Dependent variable: Month-to-month percentage change in consumer price index.Independent variables: Month-to-month percentage change in total liquidity (contemporaneous and with lags) and relative price of staples.

A Durbin-Watson statistic close to 2.0 indicates the absence of serial correlation.

The sum of lag coefficients is an estimate of the long-run multiplier and will approximate to unity as lag length increases.

Note: Dependent variable: Month-to-month percentage change in consumer price index.Independent variables: Month-to-month percentage change in total liquidity (contemporaneous and with lags) and relative price of staples.

A Durbin-Watson statistic close to 2.0 indicates the absence of serial correlation.

The sum of lag coefficients is an estimate of the long-run multiplier and will approximate to unity as lag length increases.

To further explore whether supply shocks transmitted through movements in staples prices affect inflation, some additional experiments are carried out. Equation (4) is estimated regressing monthly rates of change in consumer prices, excluding staples, on total liquidity growth. The results indicate that money explains changes in the consumer price index (CPI), excluding staples, better than changes in aggregate CPI (Table A17). Similarly, money explains changes in CPI for “other food” and CPI for “consumer goods” better than changes in aggregate CPI; and it explains changes in staples prices poorly (Table A18). Furthermore, short-run dynamics using an error correction model also show the significant impact RPS has on prices (Box 5.2).

Box 5.2.Inflation and Staples Prices: An Error Correction Model

We analyze a dynamic version of the long-run relationship between money and prices examined in Box 5.1 using the concept of error correction; the following equation is estimated:

where Δ represents the first difference operator, RPS the price of staples relative to all other prices, Ti the Tet dummies, and Ut a disturbance term. ECt-1 is the error correction term defined as the difference between the fitted value of price according to the cointegrating equation of Box 5.1 and its actual value. This term represents the short-run response necessary to move the system back toward equilibrium. This model is estimated using standard ordinary least-squares techniques with six-month lags (ρ = 6).

The results below show that the model performs well; estimated coefficients have the expected signs and the model’s explanatory power is reasonably high. In particular, changes in the price of staples relative to other goods have a positive and significant impact on inflation in the short run. The significance of the error correction term indicates that any deviation in the long-run relationship between prices and money is only short-lived and that over time these variables revert to an equilibrium relationship. This reinforces the conclusion of Box 5.1.

Adjusted R squared = 0.72 F(18,27) = 6.432 DW = 1.8; t-statistics in parentheses.

The results in this section are consistent with the assumption that supply factors influence inflation through changes in the prices of staples and, specifically, that changes in the relative prices of staples have, in addition to money, been a significant factor in inflation.

The short-run effect on inflation of an increase in the relative price of staples should diminish, and inflation should again be more closely related to money as shocks subside, rice supply and distribution strengthens, cost increases become more muted, and demand pressures from neighboring countries decline. In turn, the prospects for monetary control and more lasting inflation control depend importantly on any pressures seeking monetary accommodation. Some such pressure may be related to market imperfections. For partial evidence on this score, selective trends in the Vietnamese rice market are explored below.

Selective Trends in Rice Market

Some evidence of imperfections in the Vietnamese rice market may be gleaned from the markup behavior of different rice prices across different regions. The presumption is that the persistence of any trend increase in the markup might give rise to pressures for monetary accommodation.

In the main rice producing areas, margins of farm-gate prices over fertilizer costs during 1989-94 were on average highest in the north (Nam Ha, 21.7 percent) followed by central Vietnam (Quangnam-Da Nang, 14.3 percent) and the south (An Giang, 5 percent). By 1994, corresponding figures for the year as a whole were about 40 percent, 42 percent, and 12 percent, down from peaks earlier in the year.

By comparison, the main rice producing areas registered retail price margins over farmgate prices on the average order of 55 percent in the north, 62 percent in the center of the country, and 71 percent in the south. With trends generally declining, corresponding 1994 margins were close to 52 percent in the north and center and 51 percent in the south.

This evidence does not point to any systematic pressures for monetary accommodation. To the ex-tent that relative paddy prices continue to rise and relative rice retail prices continue to decline, developments should be consistent with monetary control and a deceleration of inflation.

Furthermore, there are indications of a potential for large efficiency gains, arising from a north-south dichotomy, which, if realized, would tend to reduce any pressure for monetary accommodation. A north-south dichotomy, already apparent in the dispersion of the recent increases in rice prices, applies also to cost and price levels in the rice market, as shown in Table 5.5. Over the years 1989-94, fertilizer prices, farmgate prices, and rice retail prices in the main rice producing areas in the north (Nam Ha) were, respectively, about 5 percent, 24 percent, and 15 percent higher than in the south (An Giang). In the urban centers, the level of rice retail prices in Hanoi was about 3 percent higher than in Ho Chi Minh City.

Table 5.5.Rice Costs and Price Levels, 1989-94(Deviation from mean, in percent)
HanoiNam HaQuangnam-Da NangHo Chi Minh CityAn Giang
Fertilizer1.931.51−0.42−3.68−3.14
Farmgate12.696.55−1.540.31−16.68
Rice retail11.133.26−2.117.93−11.96
Source: State Pricing Committee, General Statistical Office.Note: Thirteen provinces are covered.
Source: State Pricing Committee, General Statistical Office.Note: Thirteen provinces are covered.

Those differences in cost and price levels are associated with important north-south differences in the way rice markets are organized and operate. In the south, involvement of the private sector pervades trading and distribution at different levels, ensuring relatively smooth functioning of markets and greater competition, among both private companies and state-owned companies. The activity of private companies ranges from importing fertilizer and distributing it to farmers to purchasing rice paddy from the farmers and selling the milled rice to domestic retailers and exporters. At the same time, state-owned and private companies also collaborate in that state companies regularly use private companies to carry out certain transactions, with corresponding gains in efficiency.

In the north, private sector activity in the market is not yet as firmly established or as well organized, and gains from efficiency and competition are largely nonexistent. In these circumstances, it would appear that modeling the trade and distribution system in the north after that in the south should pay off, reducing the gap in cost and price levels between the two regions, with gains in real incomes and on the inflation front.

This potential for narrowing the north-south dichotomy, when realized, would tend to dampen any pressure for monetary accommodation, including through a trend reduction in price markups.

Conclusions

The findings on the determinants of inflation in Vietnam may be summarized as follows: (1) A monetary explanation broadly traces inflation in consumer prices over the past five years or so, since the shift from pervasive price and exchange rate controls to general price and exchange rate liberalization. (2) As for the reacceleration of inflation from the latter part of 1994, following years of sharply declining inflation in the wake of successful financial stabilization, changes in the relative prices of rice and, hence, staples played an important role. This confirms the view that the shocks related to natural calamities and foreign price increases in 1994 boosted the relative prices of staples from the second half of the year, fueling overall inflation. (3) There is no evidence of trends in rice (staples) prices that would seek monetary accommodation. Hence, with the inflationary effects of shocks subsiding, in the absence of new shocks, and with expansion of liquidity under control, it should be possible for inflation to resume its downward trend.

1See Box 5.1 for an elaboration of the long-term relationship between prices and money.
2Rice, including paddy and three kinds of retail rice—namely plain rice, which is the largest component, glutinous rice, and fragrant rice—accounts for 75-80 percent of consumer staples, according to the General Statistical Office of Vietnam. Fertilizer costs are the single largest cost component in paddy production and the best documented; labor costs are a “buffer,” according to the Ministry of Agriculture.

    Other Resources Citing This Publication