I. Introduction

The economic reforms introduced in China since 1978 have greatly increased the role of monetary policy in macroeconomic management. ^1/ Prior to the reforms, the banking system played a passive role, supplying credit to support the output targets contained in the central plan. Over the past 12 years, however, decision making has been decentralized and central planning scaled back. As the importance of market mechanisms has grown, the financial system has played an increasing part in allocating resources in the economy, and the monetary authorities have been given a correspondingly greater responsibility for controlling aggregate demand. ^2/ The expansion in the role of monetary policy as an instrument for macroeconomic management has added to the potential importance of monetary aggregates as intermediate targets and indicators of policy. For a monetary aggregate to be useful in formulating and monitoring monetary policy, it is, of course, necessary that the demand for it be a stable function of a small number of macroeconomic variables. With this in mind, the present paper attempts to estimate the demand for a range of monetary aggregates on the basis of quarterly data from the post-reform period.

An important issue in considering money demand in China concerns whether there is significant repressed inflation in the economy—that is, whether agents are forced to hold substantially more real money balances than they would like to—because goods are rationed at fixed prices. ^3/ Some observers have suggested that the decline in the income velocity of money during the 1980s (Chart 1) reflects such a phenomenon, and there is some econometric evidence that supports this view—see Section III. The prevalence of forced saving would clearly have important implications for attempts to estimate money demand, since real balances would be largely determined by the nominal money supply and official pricing policy rather than by notional demand, as is usually assumed.

China Money, Inflation, And Velocity, 1984–88

Citation: IMF Working Papers 1990, 042; 10.5089/9781451975543.001.A001

Source: Data provided by the Chinese authorities.1/ General retail price index2/ Gross national product divided by annual overage money balances.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

China Money, Inflation, And Velocity, 1984–88

Citation: IMF Working Papers 1990, 042; 10.5089/9781451975543.001.A001

Source: Data provided by the Chinese authorities.1/ General retail price index2/ Gross national product divided by annual overage money balances.

• Download Figure
• Download figure as PowerPoint slide

China Money, Inflation, And Velocity, 1984–88

Citation: IMF Working Papers 1990, 042; 10.5089/9781451975543.001.A001

Source: Data provided by the Chinese authorities.1/ General retail price index2/ Gross national product divided by annual overage money balances.

• Download Figure
• Download figure as PowerPoint slide

It is easy to understand how repressed inflation could arise in a full command economy in which nearly all prices are fixed by the government and black markets are effectively banned. In China, however, prices have been substantially liberalized, and during the sample period perhaps only about one half of transactions took place at prices that were set administratively. In particular, for many commodities China operates a two-track price system under which quota production is distributed at state-set prices while production above quota is sold at market prices. In addition, the prices of a number of commodities have been fully liberalized, although the prices of the entire output of some products remain regulated. It therefore seems possible that incipient increases in real balances above their desired level were Largely eliminated through their impact on those prices that were market determined. Indeed, it appears from casual observation chat market prices, and also the price Level as a whole, are strongly influenced by movements in the money supply. The decline in real money balances during 1988 as inflation picked up sharply is also striking (Chart 2). ^4/ While this paper does not attempt to estimate directly the importance of repressed inflation, the extent to which it is possible to identify a well-determined money demand function without explicit allowance for forced financial saving can be viewed as providing indirect evidence on the question. It has to be borne in mind, however, that under a partially liberalized system a demand-for-money relationship may exist even though the price level is Less than the level that would emerge if prices were fully liberalized. ^5/

China Money Balances And Inflation, 1984–88

Citation: IMF Working Papers 1990, 042; 10.5089/9781451975543.001.A001

Sources: People’s Bank of China; and staff estimates.1/ Deflated by the general retail price index.2/ Twelve-month rate of increase in the general retail price index.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

China Money Balances And Inflation, 1984–88

Citation: IMF Working Papers 1990, 042; 10.5089/9781451975543.001.A001

Sources: People’s Bank of China; and staff estimates.1/ Deflated by the general retail price index.2/ Twelve-month rate of increase in the general retail price index.

• Download Figure
• Download figure as PowerPoint slide

China Money Balances And Inflation, 1984–88

Citation: IMF Working Papers 1990, 042; 10.5089/9781451975543.001.A001

Sources: People’s Bank of China; and staff estimates.1/ Deflated by the general retail price index.2/ Twelve-month rate of increase in the general retail price index.

• Download Figure
• Download figure as PowerPoint slide

A number of features of China’s economic system have a bearing on the choice of monetary aggregates to be estimated. Traditionally, currency in circulation was the main monetary aggregate monitored by the authorities, largely because it constituted the only available medium of exchange. While the reforms have increased the attention paid to broader aggregates, cash still has a special role as a medium of exchange as checkable deposits have yet to be introduced. Partly for this reason, and also because of the attention still paid to developments in currency by policymakers, the paper estimates the demand for currency as well as for broader aggregates. ^6/ A second consideration is that even though enterprises have been given greater independence and responsibility for their financial performance under the reforms, many continue to face soft budget constraints. ^7/ This makes them potentially less sensitive than households to the costs and returns of holding money, ^8/ With this possibility in mind, demand functions of the money balances of enterprises and of households were estimated separately.

The remainder of this paper is organized as follows: Section II reviews the Literature on money demand in China; the methodology employed is discussed in Section III; Section IV reports the results of the estimation of the demand for various monetary aggregates; and concluding comments are presented in Section V.

II. Review of the Literature

Recent studies of the demand for money in China, which include papers by Chow (1987), Feltenstein and Ha (1988), and Portes and Santorum (1987), reach different conclusions about the significance of repressed inflation and also about the dynamics of adjustment to Long-run equilibrium. Chow estimates a simple partial adjustment model of the demand for money in which long-run demand depends only on national income (the simple quantity theory), using annual data For the period 1952–83 and with currency as the monetary aggregate. He found that this formulation provides “a reasonable first approximation” to explaining money demand in China. ^9/ Money demand models for two aggregates, currency and currency plus household savings deposits, were estimated by Portes and Santorum with similar annual data to that employed by Chow. Their results suggest that currency behaved in a manner more consistent with conventional (real) partial adjustment models than the broader aggregate for which a more general lag structure could not be rejected. Portes and Santorum also argue that, at least for their sample period, disequilibrium in the goods market needs to be taken explicitly into account (see below).

Feltenstein and Ha, using quarterly data for part of the post reform period, also present results that suggest that repressed inflation is important in understanding money demand in China. They estimate a modified partial adjustment model in which the demand for real money balances (and expected inflation) is assumed to be based not on the observed price level, but on the price level that would prevail in the absence of price controls—the so-called virtual price level. The latter, in turn, is hypothesized to be a function of the observed price level and the ratio of currency to retail sales (the inverse of velocity). In the estimated model, the coefficient on the ratio of currency to retail sales was significantly different from zero, suggesting that repressed inflation was a factor affecting the demand for money. The difficulty with this approach, however, is the circularity involved in using velocity itself to explain the behavior of the demand for money.

There is also a literature on household savings behavior in China that overlaps substantially with the studies on money demand mentioned above. It includes papers by Feltenstein, Lebow and van Wijnbergen (1986), Qian (1988) and Portes and Santorum (1987). ^10/ These studies, like those on money demand, differ in their conclusions about the extent of forced saving. Qian is unable to reject the hypothesis that the high savings rate in the post-reform period is the result of a shift in economic structure and/or household behavior, and hence is inclined to the view that it may not represent an increasing amount of repressed inflation. Feltenstein, Lebow and van Wijnbergen, on the other hand, using an approach similar to that of Feltenstein and Ha, conclude that unless repressed inflation is directly taken into account, false conclusions about savings behavior will be drawn. ^11/ Portes and Santorum estimate the explicit disequilibrium consumption model with annual data. They find that, since 1954, there have been both periods of generalized excess demand and excess supply, and, in particular, that the period 1980–83 was characterized by excess demand. Unfortunately, their sample period ends in 1983, about the time that substantial price liberalization was undertaken, and their results, therefore, have little to say about the potential for and extent of repressed inflation during 1983–88, the period covered by this paper.

III. Methodology

In estimating the demand for money, we have followed an approach based on methodology closely associated with David Hendry. ^12/ As applied in this paper, it has two distinctive features. The first is that, unlike in previous work on money demand in China, the data are used to determine the structure of the dynamic model through a process of testing down from general to specific formulations. This approach seems appropriate since theory provides little guide to the short-run dynamics of money demand. The second feature is that the dynamic model is estimated in “error correction” form. Such a formulation has the advantage that it helps to avoid the spurious regressions problem that can arise in estimating models with nonstationary variables and also serves to reduce multicollinearity in right-hand side variables.

The implementation of this approach first involves testing down from a general autoregressive distributed lag model of the form,

where M is the log of real money balances, Y is the log of a scale variable, and Z is an opportunity cost variable (or vector), through the sequential deletion of insignificant lags on the basis of F tests. ^13/ The resulting dynamic model may then be rewritten in error correction form. For example, if the tested-down model is given by

this may be reformulated as

The term in square brackets is the deviation of real balances in period t-1 from their long-run equilibrium value; the coefficient (1 - δ₁) captures the responsiveness of real balances to such deviations, and must be positive (i.e. δ₁ < 1) for stability. The coefficients β₀ and γ₀ represent the immediate effect on real balances of changes in income and opportunity cost, respectively. ^14/ The final step in the process is to estimate (3) using the deviations from long-run real balances calculated from the tested-down Lag model as the error correction variable in square brackets. ^15/

While it can be argued that this approach is no less ad hoc than estimation of the traditional partial adjustment model, it nevertheless has the clear advantage that it allows for a much broader range of dynamic behavior. ^16/ Indeed, the partial adjustment model is nested within the error correction model, and the restrictions implied can be tested. For instance, the error correction model (3) can be rewritten as

where M^* can be calculated from the estimated long-run model. It follows that the partial adjustment model implies the restriction β₁ = γ₁ = 0.

IV. Data

Money demand models are estimated for six monetary aggregates: currency in circulation, household money balances, enterprise money balances, total deposits, narrow money, and broad money. All equations are estimated with quarterly data for the period 1983 Ql to 1988 Q3. Household money balances are defined as currency plus time and savings deposits. ^17/ Since data on household currency holdings are available only annually, this component of quarterly household money was derived from total currency on the basis of the annual average proportion of total currency held by households. In deriving a series for enterprise money (currency plus deposits), the currency component was calculated using the same procedure. Narrow money comprises currency, household and enterprise sight deposits, and the sight deposits of quasi government institutions. Broad money is defined as narrow money plus household time and savings deposits.

The scale variable in most of the estimated models is a quarterly real national income series constructed from annual data (see Appendix I). However, a quarterly series for industrial production was used as the scale variable in the estimation of enterprise money demand. A third scale variable, household cash income, was also used in the estimation of one model of household money demand. ^18/ for broader monetary aggregates, it would have been desirable to include a broad definition of wealth in the estimated equations (broad money accounts for the bulk of financial wealth), but the needed data are not available.

Although interest rates were used more actively during 1988 and early 1989 than in the past, they were more or Less constant for the bulk of the sample period. For this reason, various measures of inflation were the only opportunity cost/rate of return variables included in estimated equations.

The price data employed in the study relate to general retail prices (a weighted average of controlled, regulated, and market prices) and to market prices. The only available series for both variables are in the form of 12-month percentage changes, however, and price indices were constructed by assuming that the price level increased smoothly during a base year (1982) in line with the twelve month growth rate for December of that year.

Given the small number of observations, formal tests for the order of integration of the time series involved were not conducted. However, inspection of plots of real balances (for each aggregate), real income and inflation suggested that the first differences of their logs were stationary.

V. Empirical Results ^19/

In applying the methodology described in the previous section, the autoregressive Lag model for each aggregate was estimated initially with four lags for all variables. A general finding was that the inflation rate of the general retail price index performed better than the inflation rate of market prices as a proxy for inflation expectations, and hence only models with general retail price inflation included are reported. ^20/

1. Household money balances

The estimated error correction model for household money balances is

$\mathrm{\Delta M}\mathrm{=}\mathrm{-}\underset{\mathrm{\left(}\mathrm{0.10}\mathrm{\right)}}{\mathrm{0.001}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{3.82}\mathrm{\right)}}{\mathrm{0.781}}{\overline{\mathrm{u}}}_{\mathrm{t}\mathrm{-}\mathrm{1}}\mathrm{+}\underset{\mathrm{\left(}\mathrm{3.99}\mathrm{\right)}}{\mathrm{0.910}}{\mathrm{\Delta Y}}_{\mathrm{t}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{5.57}\mathrm{\right)}}{\mathrm{1.583}}{\mathrm{\Delta \pi }}_{\mathrm{t}}^{\mathrm{*}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{3.85}\mathrm{\right)}}{\mathrm{0.114}\mathrm{D}}$

R2 = 0.74			Ċ = 0.026
$\mathrm{AR}(4,11)\underset{1}{4}$	=	0. 53 (3.36)	$\mathrm{AR}(1,14)\underset{1}{1}$	=	0.08 (4.60)
NORM (2)	=	0.82 (5.99)	ARCH (2)	=	0.54 (5.99)
HET (1, 18)	=	1.50 (4.41)	CHOW (4, 11)	=	0.37 (3.36)
PREF (12, 13)	=	0.09 (8.74)

View Table

R2 = 0.74			Ċ = 0.026
$\mathrm{AR}(4,11)\underset{1}{4}$	=	0. 53 (3.36)	$\mathrm{AR}(1,14)\underset{1}{1}$	=	0.08 (4.60)
NORM (2)	=	0.82 (5.99)	ARCH (2)	=	0.54 (5.99)
HET (1, 18)	=	1.50 (4.41)	CHOW (4, 11)	=	0.37 (3.36)
PREF (12, 13)	=	0.09 (8.74)

where û_t-1 is the deviation of real household balances from long-run equilibrium calculated from the tested-down version of the general Lag model. ^21/ The (expected) inflation term, Π^*, is the average of current (quarterly) inflation, inflation lagged one period and inflation lagged four periods; ^22/ Y is the log of real national income; ^23/ and D is a dummy variable for the second quarter of 1985, a period when monetary growth was sharply reduced primarily through the tightening of direct credit controls.

The estimated equation satisfies the diagnostic tests, including a Chow test for parameter stability and a Chow predictive failure test (with the sample split at the second quarter of 1985 for both tests). ^24/ The coefficient on the error correction term, û_t-1. is significant at the 1 percent level, which provides evidence that real balances, real income, and inflation are cointegrated. The underlying equilibrium relationship from which ù is derived implies that the Long-run income elasticity is close to 2 (Table 1), indicating an increasing role for money in the economy as the reforms progressed during the sample period; ^25/ income elasticities greater than one are obviously not sustainable in the very long run. ^26/ The long-run coefficient on the inflation variable (whose dimensions are percent per quarter) indicates that a 1 percentage point increase in the annual inflation rate would reduce real household money balances by about one half percentage point.

Table 1.

China: Money Demand—Estimated Long-Run Coefficients ^1/

^1/The coefficients are calculated from the estimated autoregressive lag models for each monetary aggregate.

^2/The scale variable is industrial production in the case of enterprise money and national income for all other aggregates.

^3/For all aggregates, the reported coefficient captures the effect of a permanent increase in quarterly general retail price inflation (in percent per quarter).

Table 1.

China: Money Demand—Estimated Long-Run Coefficients ^1/

	Log of scale variable 2/	Inflation variable 3/
Household money	1.97	-2.06
Enterprise money	1.48	-5.32
Currency in circulation	1.66	-1.01
Savings deposits	1.69	-4.70
Narrow money	1.34	-2.38
Broad money	1.53	-3.67

^1/The coefficients are calculated from the estimated autoregressive lag models for each monetary aggregate.

^2/The scale variable is industrial production in the case of enterprise money and national income for all other aggregates.

^3/For all aggregates, the reported coefficient captures the effect of a permanent increase in quarterly general retail price inflation (in percent per quarter).

View Table

Table 1.

China: Money Demand—Estimated Long-Run Coefficients ^1/

	Log of scale variable 2/	Inflation variable 3/
Household money	1.97	-2.06
Enterprise money	1.48	-5.32
Currency in circulation	1.66	-1.01
Savings deposits	1.69	-4.70
Narrow money	1.34	-2.38
Broad money	1.53	-3.67

^1/The coefficients are calculated from the estimated autoregressive lag models for each monetary aggregate.

^2/The scale variable is industrial production in the case of enterprise money and national income for all other aggregates.

^3/For all aggregates, the reported coefficient captures the effect of a permanent increase in quarterly general retail price inflation (in percent per quarter).

Turning to the dynamics of household money demand, the results imply that a 1 percent increase in national income initially raises real household balances by just under 1 percent, while a 1 percentage point increase in Π ^* (at an annual rate) lowers real balances on impact by 0.4 percent. ^27/ The remaining adjustment to the long-run equilibrium appears to occur rapidly, with almost 30 percent of the deviation between actual and equilibrium money balances (lagged one quarter) eliminated in a quarter. ^28/ A final point with regard to the dynamic structure is that the restrictions implied by the partial adjustment model are decisively rejected when the error correction model is estimated in a form similar to equation (4).

2. Enterprise money

The estimated error correction model for enterprise money is

$\begin{array}{l}\mathrm{\Delta M}\mathrm{=}\underset{\mathrm{\left(}\mathrm{0.67}\mathrm{\right)}}{\mathrm{0.10}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{4.70}\mathrm{\right)}}{\mathrm{0.335}}{\stackrel{\mathrm{‐}}{\mathrm{u}}}_{\mathrm{t}\mathrm{-}\mathrm{1}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{3.37}\mathrm{\right)}}{\mathrm{1.438}}\mathrm{\Delta }{\mathrm{Y}}_{\mathrm{t}}^{\mathrm{*}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{2.92}\mathrm{\right)}}{\mathrm{0.553}}\mathrm{\Delta \Delta }{\mathrm{Y}}_{\mathrm{t}}^{\mathrm{*}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{4.74}\mathrm{\right)}}{\mathrm{1.467}}{\mathrm{\Delta \pi }}_{\mathrm{t}}\\ \underset{\mathrm{\left(}\mathrm{3.27}\mathrm{\right)}}{\begin{array}{c}\mathrm{-}\mathrm{0.098}\end{array}}\mathrm{D}\mathrm{+}\underset{\mathrm{\left(}\mathrm{6.36}\mathrm{\right)}}{\mathrm{0.189}}{\mathrm{D}}_{\mathrm{1}}\mathrm{+}\underset{\mathrm{\left(}\mathrm{6.19}\mathrm{\right)}}{\mathrm{0.128}}{\mathrm{D}}_{\mathrm{2}}\end{array}$

R2 = 0.929			Ċ = 0.025
$\mathrm{AR}{\left(4,13\right)}_1^4$	=	0.77 (3.18)	$\mathrm{AR}{\left(1,16\right)}_1^1$	=	0.08 (4.49)
Norm (2)	=	1.16 (5.99)	ARCH (2)	=	2.64 (5.99)
HET (1, 23)	=	0.59 (4.28)	CHOW (6, 13)	=	0.64 (2.92)
PRE (13, 6)	=	0.21 (3.56)

View Table

R2 = 0.929			Ċ = 0.025
$\mathrm{AR}{\left(4,13\right)}_1^4$	=	0.77 (3.18)	$\mathrm{AR}{\left(1,16\right)}_1^1$	=	0.08 (4.49)
Norm (2)	=	1.16 (5.99)	ARCH (2)	=	2.64 (5.99)
HET (1, 23)	=	0.59 (4.28)	CHOW (6, 13)	=	0.64 (2.92)
PRE (13, 6)	=	0.21 (3.56)

The inflation term, Π, is the current inflation rate of the general retail price index (the general lag model indicated that longer lags did not influence enterprise money demand); and Y is the log of industrial production. The model also includes a fourth quarter seasonal dummy (D₁) and a dummy variable for the fourth quarter of 1984 (D₂), in addition to the dummy variable for the second quarter of 1985 discussed earlier. The dummy variable D₂ is included because there was a surge in credit in late 1984 that resulted in a sharp, but temporary, rise in enterprise real money balances. ^29/

With regard to the dynamics of the model, the presence of the accelerator term, ΔΔY, implies that the impact effect of a change in income is smaller when income growth is rising. The role of current inflation rather than the lag structure found relevant for household money demand suggests either that the inflation expectations of enterprises react faster to changes in inflation than those of households, or that enterprise money demand responds sooner to changes in expected inflation. The Latter is perhaps the more likely explanation since enterprise money balances consist almost entirely of currency and sight deposits, whereas a substantial proportion of household money consists of time deposits for which there are penalties for early withdrawal. It is also interesting to note that the long-run effect of a change in inflation on enterprise money balances is substantially larger than for households. At the same time, the speed of response to deviations from long-run real balances (38 percent per quarter) is substantially slower than in the case of household money balances. To the extent that enterprise money demand is determined largely by transactions requirements, an adjustment that is faster than for households might have been expected.

Finally, the restrictions imposed by the partial adjustment model on the error correction equation were tested and rejected at the 1 percent level.

3. Currency in circulation

The estimated model for currency has a more complex dynamic structure than the models for the other aggregates. The fitted equation is

$\begin{array}{l}{\mathrm{\Delta M}}_{\mathrm{t}}\mathrm{=}\underset{\mathrm{\left(}\mathrm{0.15}\mathrm{\right)}}{\mathrm{0.002}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{2.71}\mathrm{\right)}}{\mathrm{0.639}}{\hat{\mathrm{u}}}_{\mathrm{t}\mathrm{-}\mathrm{1}}\mathrm{+}\underset{\mathrm{\left(}\mathrm{4.28}\mathrm{\right)}}{\mathrm{1.176}}{\mathrm{\Delta Y}}_{\mathrm{t}}\mathrm{+}\underset{\mathrm{\left(}\mathrm{2.15}\mathrm{\right)}}{\mathrm{2.438}}\mathrm{\Delta }{\mathrm{\Pi }}_{\mathrm{t}\mathrm{-}\mathrm{1}}\\ \mathrm{-}\underset{\mathrm{\left(}\mathrm{2.78}\mathrm{\right)}}{\mathrm{2.161}}\mathrm{\Delta }{\mathrm{\Pi }}_{\mathrm{t}\mathrm{-}\mathrm{2}}\mathrm{+}\underset{\mathrm{\left(}\mathrm{2.92}\mathrm{\right)}}{\mathrm{0.140}}{\mathrm{D}}_{\mathrm{1}}\end{array}$

R2 = 0.81			Ĝ = 0.043
$\mathrm{A}\mathrm{R}{\mathrm{\left(}\mathrm{4}\mathrm{,}\mathrm{10}\mathrm{\right)}}_1^4$	=	1-00 (3.48)	$\mathrm{A}\mathrm{R}{\mathrm{\left(}\mathrm{1}\mathrm{,}\mathrm{14}\mathrm{\right)}}_1^1$	=	0.00 (4.60)
NORM (2)	=	0.19 (5.99)	ARCH (2)	=	0.21 (5.99)
HET (1)	=	0.168 (4.41)	CHOW (5, 11)	=	1.44
PRE (13, 3)	=	0.60 (8.73)

View Table

R2 = 0.81			Ĝ = 0.043
$\mathrm{A}\mathrm{R}{\mathrm{\left(}\mathrm{4}\mathrm{,}\mathrm{10}\mathrm{\right)}}_1^4$	=	1-00 (3.48)	$\mathrm{A}\mathrm{R}{\mathrm{\left(}\mathrm{1}\mathrm{,}\mathrm{14}\mathrm{\right)}}_1^1$	=	0.00 (4.60)
NORM (2)	=	0.19 (5.99)	ARCH (2)	=	0.21 (5.99)
HET (1)	=	0.168 (4.41)	CHOW (5, 11)	=	1.44
PRE (13, 3)	=	0.60 (8.73)

where Π is a three quarter average of general retail price inflation. The model implies that real currency holdings at first rise in response to an increase in inflation, before declining in later periods to less than their initial value. ^30/ The subsequent reduction in demand is brought about both through the error correction term (adjustment is quite rapid), and also through the term, Π_t-2. ^31/

The dynamic response of the demand for currency to changes in inflation indicated by the model has an intuitively plausible, “cash-in-advance”, interpretation in the Chinese context. When inflation rises with nominal interest rates held unchanged, the long-run objective of agents is to shift from both currency and deposits into consumer durables and other storable commodities. However, as noted earlier, there are no checking accounts in China and the availability of some goods has been subject to uncertainty. In these circumstances, it seems likely that agents would need to increase their cash holdings initially to put themselves in a position to purchase goods when they became available. This response pattern was particularly apparent during mid-1988 when currency holdings rose sharply as a result of large withdrawals from savings deposits, whose real return had become substantially negative with rising inflation.

It seems unlikely, however, that the relationship between currency demand and inflation is symmetric. If inflation falls, the long-run real demand for both currency and deposits would increase, but there would seem to be little reason for currency holdings to be reduced in the short term as agents attempt to shift from real goods to various forms of money. ^32/ The properties of the estimated model probably reflect the fact that inflation was rising for most of the sample period.

As in the case of household and enterprise money balances, the restrictions implied by the partial adjustment model are rejected.

4. Savings and time deposits

Given the different dynamic behavior of currency to household and enterprise money, a model for total deposits (defined as broad money minus currency) was also estimated, The fitted error correction model

$\begin{array}{l}\mathrm{\Delta }\mathrm{M}\mathrm{=}\mathrm{-}\underset{\mathrm{\left(}\mathrm{0.89}\mathrm{\right)}}{\mathrm{0.001}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{4.25}\mathrm{\right)}}{\mathrm{0.480}}{\hat{\mathrm{u}}}_{\mathrm{t}\mathrm{-}\mathrm{1}}\mathrm{+}\underset{\mathrm{\left(}\mathrm{5.37}\mathrm{\right)}}{\mathrm{0.803}}\mathrm{\Delta }{\mathrm{Y}}_{\mathrm{t}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{6.27}\mathrm{\right)}}{\mathrm{0.992}}\mathrm{\Delta }{\mathrm{\Pi }}_{\mathrm{t}}\\ \mathrm{-}\underset{\mathrm{\left(}\mathrm{2.11}\mathrm{\right)}}{\mathrm{0.431}}\mathrm{\Delta }\mathrm{\Delta }{\mathrm{\Pi }}_{\mathrm{t}}\mathrm{-}\underset{\mathrm{\left(}\mathrm{2.82}\mathrm{\right)}}{\mathrm{0.057}}\mathrm{D}\mathrm{+}\underset{\mathrm{\left(}\mathrm{5.51}\mathrm{\right)}}{\mathrm{0.112}}\mathrm{D}\mathrm{1}\end{array}$

R2 = 0.90			Ĉ = 0.018
$\mathrm{A}\mathrm{R}\mathrm{\left(}\mathrm{4}\mathrm{,}\mathrm{12}\mathrm{\right)}{}_1^4$	=	1.228 (3.26)	$\mathrm{A}\mathrm{R}{\mathrm{\left(}\mathrm{1}\mathrm{,}\mathrm{15}\mathrm{\right)}}_1^1$	=	2.52 (4.54)
NORM (2)	=	0.80 (5.99)	ARCH (2)	=	2.35 (5.99)
HET (1, 21)	=	0.10 (4.32)	CHOW (5, 12)	=	0.99 (3.11)
PRE (12, 5)	=	0.37 (4.68)

View Table

R2 = 0.90			Ĉ = 0.018
$\mathrm{A}\mathrm{R}\mathrm{\left(}\mathrm{4}\mathrm{,}\mathrm{12}\mathrm{\right)}{}_1^4$	=	1.228 (3.26)	$\mathrm{A}\mathrm{R}{\mathrm{\left(}\mathrm{1}\mathrm{,}\mathrm{15}\mathrm{\right)}}_1^1$	=	2.52 (4.54)
NORM (2)	=	0.80 (5.99)	ARCH (2)	=	2.35 (5.99)
HET (1, 21)	=	0.10 (4.32)	CHOW (5, 12)	=	0.99 (3.11)
PRE (12, 5)	=	0.37 (4.68)

is similar to those for enterprise and household money, although it differs in that it includes an acceleration term in inflation. The underlying long-run model indicates a greater responsiveness to inflation than either household or enterprise money. It seems likely that this reflects the relatively low estimated inflation elasticity of the currency component of the latter two aggregates. The restrictions implied by the partial adjustment model are rejected at the 1 percent level.

VI. Concluding Observations

The results presented above suggest the existence, for a range of monetary aggregates, of relatively stable demand relationships that might be helpful in guiding monetary policy. The extent of the testing for stability, however, was limited by the small number of observations. It is noteworthy that, for most aggregates, the standard real partial adjustment model is decisively rejected, indicating that a more complex dynamic structure is required to characterize the data adequately. In all cases, the long-run income elasticity is greater than unity, which probably reflects monetization of the economy under the reforms as the coordination of economic activity has increasingly taken place through markets.

A possible alternative explanation of the high income elasticities is that a rising degree of repressed inflation has been correlated with real income growth. However, another feature of the results is that real balances appear to be sensitive to expected inflation as proxied by the inflation rate of the general retail price index. This suggests that agents are able to adjust the level of their real balances to changes in opportunity costs, and hence tends to undermine the argument that a substantial amount of money holdings is involuntary. ^33/ At the same time, the question of how expectations were formed remains murky, as the lag pattern on the inflation variable differed across equations. Indeed, the lag patterns probably reflected a combination of expectation formation and the dynamics of adjustment of money demand to changes in expectations.

A particular finding is that an increase in inflation raises the real demand for currency on impact, although it lowers currency demand in the long run; for all other aggregates the impact effect is negative. The dynamics of currency demand probably reflect the absence of checking accounts and the uncertainties about the availability of some consumer goods. In these circumstances, to reduce money holdings (currency and deposits) by increasing purchases of commodities, it is likely that agents initially would need to increase their cash holdings so as to be in a position to purchase goods when they become available. It is also likely that this effect is not symmetric; there is no obvious reason why cash holdings should fall in the short-run when the opportunity cost of holding money declines. The dynamic properties of the estimated model probably reflect the fact that inflation was rising for most of the sample period. The estimation of a model in which an asymetric response is allowed for might make an interesting research project.

Although the estimated models passed a limited range of stability tests, the nature of money demand in China is likely to evolve as the economic system, particularly the financial sector, develops. For instance, the introduction of checking accounts would be likely to affect the dynamics of the demand for currency. In addition, the dynamic structure of the models almost certainly reflects a combination of both the dynamics of adjustment to long-run equilibrium and the structure of expectations formation mechanisms; the latter are likely to evolve if the processes generating the variables in question change (the Lucas critique). It should also be remembered that many of the series employed were constructed, and the results obtained may reflect, to some extent, the idiosyncrasies of the data.

An issue that remains to be addressed is the impact of changes in interest rates on the demand for various aggregates; expected inflation is the only opportunity cost variable in the estimated equations because of the very Limited variability of interest rates during the sample period. It is likely that a reduction in deposit interest rates would have a similar effect on the demand for deposits to that of an increase in inflation. The consequences for the demand for currency in the present circumstances, however, would probably be more complex. In particular, the long-run demand for currency would be likely to rise, but possibly by less than the short-run increase, as agents temporarily increased their currency holdings as a transitional stage in the substitution from deposits into commodities. The response of currency to an increase in deposit rates, however, would be unlikely to involve a larger short-run than long-run decline in demand, and hence might not be symmetric. Since interest rate policy in China has become more active over the past two years, it may soon be possible to investigate this question empirically.