Chapter III.3 The Monetary Overhang: an Analytical and Empirical Study

International Monetary Fund
Published Date:
December 1991
  • ShareShare
Show Summary Details


Official price indexes in the USSR have been remarkably stable over the last 30 years; between 1960 and 1980 the official retail price index remained almost unchanged and, even in the 1980s, official recorded inflation barely exceeded 1 percent per year. It is widely recognized, however, that inflationary pressures, particularly throughout the latter period, were substantially higher. On the one hand, official price indices almost certainly underestimated actual price increases, a phenomenon known in the Western literature as hidden inflation.1 On the other hand, insofar as price controls prevented actual prices from reaching their equilibrium level, inflation remained artificially contained (repressed inflation). This constrained nominal consumption and led to involuntary savings. Since most of the involuntary wealth accumulation was in the form of liquid assets, it is described as a “monetary overhang.” The monetary overhang, and the underlying repressed inflation that it represents, has potentially serious implications for the equilibrium of the Soviet economic system in the event of price liberalization. Section 2 presents some simple analytical considerations regarding the definition and implications of repressed inflation and the monetary overhang; section 3 provides quantitative measures of the overhang, while section 4 discusses possible ways to tackle the problem in the near future. Finally, section 5 draws the main conclusions.


In order to define the concept of repressed inflation, and the strictly related concept of a monetary overhang, consider initially a one good and one asset (money) economy. The good can be traded only on the official market by the Government (there is no black market). Real production in period t is equal to yt; its market value is ptyt=Yt. Repressed inflation can be defined as a situation in which, at price pt, demand exceeds supply, so that the removal of price controls would lead to a price increase.2 The excess of demand over supply implies some rationing;3 if all demand is for consumption purposes (disregarding for the moment the existence of investment), the counterpart to a scarcity of goods is involuntary saving of households and therefore an involuntary increase in nominal wealth.4 The wealth overhang can be defined as the difference between the nominal amount of wealth held by agents at the end of period t and the desired amount that would have been held in the absence of current and past rationing.

In the debate on repressed inflation in centrally planned economies (CPEs), it is usually assumed that all undesired accumulation of wealth is held as monetary balances; therefore, the terms monetary overhang and wealth overhang are often used synonymously. The two concepts, however, coincide only in a special case (which is only approximately met even in CPEs), i.e., when money is the only available store of wealth, or, equivalently, when, despite the existence of other stores of wealth, their supply and their price is fixed (so that additional resources can only be invested in money). In general, the concepts of wealth and monetary overhang differ; while “wealth overhang” refers to the stock of accumulated involuntary saving, monetary overhang is the difference between the amount of money held by agents and the amount that would have been held in the absence of current and past rationing (in the goods and financial markets). Clearly the existence of a monetary overhang can occur even in the absence of a wealth overhang; indeed, total consumption (and total wealth) could be in equilibrium, but, given the structure and regulation of financial and real asset markets, the composition of wealth might be biased towards money. Due to their different policy implications, both concepts are relevant and will be explored empirically. More precisely, in the empirical analysis of section 3, two questions will be addressed separately: first, to what extent has household consumption been rationed in the USSR and hence given rise to an involuntary accumulation of wealth. Second, to what extent is the composition of wealth of both households and enterprises biased towards financial and monetary assets.

In the presence of many goods, and of price controls, markets in which shortages and surpluses prevail can co-exist. How to define the concept of aggregate excess demand—from which that of the “overhang” is derived—under these conditions has been a subject of considerable controversy. On one side is the so-called “disequilibrium school.”5 According to this approach, the step from a one good economy to an economy with many goods is not a difficult one; if, at the current price vector, aggregate demand for goods and services (the sum of demands in all markets) is greater than aggregate supply (the sum of supplies in all markets), then the economy is in excess demand and, with sticky prices, it is subject to repressed inflation (i.e. the absolute price level is too low to clear the market). If, instead, aggregate demand and aggregate supply are in equilibrium, the contemporaneous existence of excess supply and excess demand in different markets requires an adjustment in relative prices, but does not imply the need to adjust the price level. Certainly, in the presence of both shortages and surpluses, the measurement of excess demand may present some problems; however these problems can be tackled econometrically.6

On the other side is the so-called “shortage economy” school, which argues that the concept of aggregate disequilibrium, legitimate in market economies, becomes ill-defined in economies characterized by “chronic” shortages. Two specific points are raised: first, agents who are unable to buy the goods they desire almost always end up buying substitutes; very rarely do their savings rise involuntarily.7 Second, the volume of voluntary savings may be directly affected by the existence of chronic shortages at the micro level;8 the precautionary reserve of purchasing power (stored mainly in monetary assets) increases, artificially inflating the saving rate.9

The debate between these two points of view has been fruitful. From a theoretical standpoint, there seems to be no doubt that, in principle, the concept of aggregate disequilibrium can be extended from market economies to CPEs and that the distinction between distortion of the price level and distortion of relative prices is an important one. In this respect, the first point raised by the “shortage” school has to be interpreted properly. If forced substitution is widespread, consumers’ utility is clearly affected, as their desired expenditure basket is distorted. Yet, it would be correct to say that there is no overhang; total expenditure may still, ex post, equal planned expenditure. The second point raised by the shortage school is, however, more important, as it implies that the desired accumulation of wealth (or, equivalently, the propensity to save) is not independent of current and expected market conditions and, more specifically, of the existence of shortages. Therefore, even if empirical evidence showed the absence of forced saving in a system in which shortages at the micro level are pervasive, it might still be possible that the (credible) announcement of a permanent change in regime regarding the supply of goods could increase desired consumption (with potentially inflationary effects on the price level).

One additional conceptual point could be mentioned. When both shortages and surpluses are present simultaneously, the size of forced saving may overstate the amount of macroeconomic disequilibrium. Consider the case of two markets—market A in which, at the current fixed prices, excess supply is present, and market B in which there is excess demand. Buyers would like to purchase 50 “rubles” worth of A and 50 “rubles” worth of B. But the equivalent of 70 rubles are offered of A and only 30 of B. Applying the min condition, transactions occur at 50 for A and at 30 for B. Aggregate expenditure is 80 and forced saving is 20. Yet, ex ante, aggregate demand and aggregate supply are equal (each sums to 100 “rubles”), implying no macroeconomic disequilibrium. This discrepancy disappears, however, when the involuntary accumulation of inventories (equal to 20 “rubles” of A in the previous example) is brought into the picture. This suggests that the existence of a macro-disequilibrium between demand and supply should, in principle, be measured taking into account not only the involuntary accumulation of wealth by buyers, but also the undesired accumulation of unsold products due to distortions in relative prices. It is for this reason that the ratio between monetary holdings of the population, which in some cases may be a good indicator of excess saving, and the stock of unsold consumer goods is sometimes used as an indicator of disequilibrium in CPEs.10

To take the analysis further, assume that, in addition to the official markets, parallel markets are also present. These may include legal and illegal (black) markets, in which suppliers can sell a portion of their total production at unregulated prices. In principle, parallel markets may eliminate macroeconomic rationing. If these markets are large enough (and in particular if they are accessible to everybody and can supply most goods), buyers will address to them all unused purchasing power left from shopping at state stores (less of course, desired saving). Prices on the parallel market will tend to be higher than those on the state market; and, in general, the two price levels will be respectively higher and lower than the single price level prevailing if controls were removed. In this case, relative prices would continue to be distorted but the average price level may be close to the unregulated equilibrium. Total expenditure may equal desired expenditure, involuntary accumulation of wealth may be zero, and the money market would be in equilibrium since the real amount of money (deflated by a weighted average of official and parallel market prices) would be equal to the desired demand.11

Several factors can, however, prevent the relevant level of prices from equilibrating the money market. Parallel markets may not be extensive enough: they may cover a limited spectrum of goods, may be geographically concentrated in some areas, and trade in these markets may be discouraged by legal sanctions. It has been shown that the existence of markets in which prices are free does not prevent the accumulation of excess saving if sales in these markets are an imperfect substitute for sales on the official market.12

Even if parallel markets are extensive, buyers may still be reluctant to purchase on these markets if they expect the price system to be liberalized (which would lead to a decline in parallel market prices). Thus, purchases on the parallel markets may be inadequate to prevent the buildup of a monetary overhang—albeit one that is “voluntarily held.” Note that the same result is obtained when the availability of goods in the official market is expected to increase in the near future or when the probability of obtaining these goods in the official market increases with queuing.13

The extent to which the existence of parallel markets in the USSR has prevented a monetary overhang from emerging is ultimately an empirical question. Information on this subject is inadequate, but the size of the second economy may be fairly large and, as a consequence, official data on disposable income and expenditure are likely to underestimate their actual values.14 Some information on price behavior in parallel markets is available, but, used in isolation, may provide misleading information on the degree of macroeconomic imbalance, for the reasons given earlier.


a. The repressed consumption debate in the USSR

Few would argue that parallel market activity was extensive enough in the late 1980s to absorb all of the notional demand for consumption goods that could not be satisfied on official markets. It therefore seems reasonable to assume that households experienced some degree of involuntary saving in this period. There is less agreement, however, on the existence of involuntary saving before the most recent period, an issue which is important for establishing the current size of the monetary and wealth overhangs. The debate has been based on both theoretical and empirical arguments.15

Supporters of the thesis that repressed inflation and involuntary accumulation of wealth have always characterized the Soviet economy, as well as CPEs in general, stress the well known phenomena of shortages at the micro level and limits on the ability of households to respond to shortages of consumer goods by reducing labor supply.16 The need for large voluntary savings in a society in which the state takes care of most of the basic “life cycle” requirements of the population has also been questioned.17

Three sets of indicators of repressed inflation have been proposed. First, there are indicators of shortages at the micro level,18 ranging from the number of people in queues for officially allocated apartments, to measures of the time spent searching for goods, or the physical length of queues—unfortunately these indicators are based on data that are not readily available, and none appears to have been computed for the USSR.19 A second set of indicators has been based on the ratio between prices in the collective farm (kolkhoz) markets and state prices for the same products, which shows a continuous increase throughout the last 30 years, albeit at a much faster rate at the end of the 1980s and in 1990 (Chart 1, upper panel). There are two problems, however, with this indicator. First, it has been suggested that this ratio should be weighted for the share of expenditure on the kolkhoz markets, with respect to total consumers’ expenditure.20 If this is done, the trend over the longer period disappears (Chart 1, lower panel), although a sharp increase in the most recent period remains. Second, as already mentioned, while the relative kolkhoz price may point to rising disequilibrium in state markets, it is not necessarily a good indicator of macroeconomic disequilibrium.


(Relative to state prices)

Sources: Dirkens (1981); Nuti (1989); Goskomstat (1989); Alexeev, Gaddy and Leitzel (1990).

1. Unweighted ratio multiplied for the share of output sold in the free market (1960 = 100). (Holzman’s rationing indicator).

A third group of indicators, which has been applied frequently in the Soviet case, is based on ratios between financial (or monetary) holdings of the population and some “scale variable” (usually households’ consumption or disposable income). These ratios exhibit spectacularly rising trends in recent years (Chart 2), strongly suggesting involuntary increases in saving. A clear upward trend is also apparent in the ratio between total wealth (defined as financial wealth plus houses and other real assets of the population) and the same scale variables.21

Chart 2.USSR : HOUSEHOLDS’ WEALTH, 1955-90

Sources: See statistical note.

In spite of this evidence, the view that CPEs have been necessarily characterized by repressed inflation has been challenged by many authors. In addition to the claim discussed above that the indices of shortages at the micro level may only reflect microeconomic imbalances, but not an aggregate demand problem, it has been argued that: (a) there is some (albeit weak) evidence that labor supply is flexible in CPEs (and specifically in the USSR),22 which might provide households with a way to reduce imbalances even in the presence of an initial disequilibrium; and (b) there are several reasons to have expected households’ wealth to increase faster than consumption and income. An increase in the ratio of savings to income could have been connected to the general rise in the standard of living of the population (as saving is zero if income is at subsistence levels);23 to a deterioration in the provision of public services and social security payments; and to increased expenditure on durables (given the absence of consumer credit, the purchase of durables requires previous accumulation of savings).24 In addition, it has been observed that the initial level of wealth in the USSR was extremely low; total wealth represented only 70 percent of annual consumption in 1955, and financial wealth (for which a more precise measurement is possible) only 19 percent. By contrast, the wealth to consumption ratio in Western countries ranges from 400 to 600 percent and, for financial wealth, is close to 200 percent in many countries (Table III.3.1).25 Even after 35 years of continuous increase, the comparable ratios in the USSR remained far below these levels. Given such a low starting point, it is not surprising that a steady rise in the wealth to income ratio should have occurred.26 Finally, an increase in this ratio is also consistent with equilibrium consumption in the presence of a decelerating growth rate of real disposable income,27 as was the case in the 1960s and, especially, the 1970s.

Critics of the repressed inflation hypothesis have argued that the ratio of saving to disposable income is a more revealing indicator of underlying behavior than the wealth-consumption ratio. The saving rate in the USSR does not appear particularly high, compared to Western countries (Table III.3.1). This may reflect the state’s provision of “life cycle” services and cannot per se rule out the existence of rationing. However, the remarkable stability of the Soviet savings rate up to the late 1980s—with the exception of a step increase in the mid-1960s (Chart 3)—has been used to suggest that if rationing ever existed it remained relatively stable.28

Table III.3.1.USSR: International Comparison of Wealth Ratios and Saving Rates

Net Wealth 1

Net Wealth 1

Saving 2

Saving 2
United States19814.232.249.716.17
France1983 & 19884.39(1983)1.27 (1988)7.57
United Kingdom19886.
(Average G-7)8.967.56 3
USSR19851.091.095.70 25.64 2
USSR19901.40 41.23 4

As a ratio to households’ consumption.

In percentage of households’ disposable income. All saving rates refer to the 1980-87 average with the exception of Italy (1983-87) and the USSR (1980-85).

Excluding France.


As a ratio to households’ consumption.

In percentage of households’ disposable income. All saving rates refer to the 1980-87 average with the exception of Italy (1983-87) and the USSR (1980-85).

Excluding France.


Chart 3.USSR : HOUSEHOLDS’ SAVINGS, 1955-901

Sources: See statistical note.

1. In percent of disposable income.

All these arguments have been used to reject, a priori, the presence of involuntary saving in CPEs and several attempts have been made to reinforce the point empirically by estimating household consumption functions. Some studies have estimated consumption behavior under the hypothesis of equilibrium and have evaluated this hypothesis based on the performance of the estimated equation (in terms of fit and of the behavior of the residuals).29 In all cases, households’ disposable income (or its moving average) has been used as the main explanatory variable of consumption.30 In some of these studies, indicators of possible disequilibrium (such as the ratio of kolkhoz to state prices or the rate of change in the CPI index) were introduced as explanatory variables, but did not produce satisfactory results.31 These studies seem to indicate that consumption behavior in the USSR can be explained mainly in terms of disposable income with the exception of a structural break which occurred in the middle of the 1960s.32 It is significant, however, that they focused on the period prior to 1980.

b. Econometric estimates of the consumption function

The preceding discussion suggests a strategy for the empirical analysis of consumption trends in the USSR and for the measurement of households’ wealth overhang. As a working hypothesis, consumption behavior can be analyzed in terms of standard consumption theory. The life cycle approach, modelling consumption as a function of both human and nonhuman wealth, appears to be particularly appropriate as it takes into account explicitly the effect on consumption of (desired and undesired) accumulation of saving.

Under these assumptions, equations were estimated on a data set which excluded the second half of the 1980s, the period during which any existing overhang is likely to have increased substantially. The existence of possible forced saving over the estimation period (up to 1985) was taken into consideration by introducing, as an indicator of rationing, the ratio between kolkhoz and state prices. It was therefore implicitly assumed that the kolkhoz market is extended enough to provide a reliable indicator of excess demand, but is not sufficient (especially because of the limited number of products supplied) to eliminate excess demand at the macro level.33 Any effects of shortages during the sample period, other than those captured by the rationing indicator, were assumed to remain approximately stable and therefore to leave unaffected the properties of the estimated equation.34

The equation was then used to project consumption in the second half of the 1980s, and forced saving could be measured as the difference between actual and projected saving plus the estimate of rationing derived from the inclusion in the equation of the rationing indicator plus a judgmental estimate of nonmeasurable “chronic” shortages.

The following model specification (in static form) was used as the base for the estimation:

where C is households’ consumption (including of consumer durables)35; W is non-human wealth36 and H is human wealth37; X is a vector of variables influencing the steady state ratio between consumption and total wealth; IR is the index of rationing computed on the basis of kolkhoz prices; and a0, a1, and a2 are parameters and b is a parameter vector. Note that a2IR should approximate the percentage amount of rationing (possibly in excess of an unobservable fixed amount, included in the constant).

Vector X includes three variables: the real interest rate on deposits, computed as the nominal interest rate on bank deposits minus an estimate of actual (as opposed to official) inflation (see the statistical notes); the “dependency ratio”, defined as the ratio of the nonworking population (children below 16 years of age and pensioners) to the remaining population; and the “benefit ratio,” defined as total benefits extended by the social consumption fund to each nonworking member of the population, divided by the wage rate.38 The expected sign is negative for the real interest rate and positive for the benefit ratio and the dependency ratio.

Equation (1) assumes that human and nonhuman wealth have the same effect on consumption; a less restrictive specification would be:

which is equivalent to (1) if a’1=a1. As a check of the model specification, the equations were estimated without imposing the equality constraint between the coefficient on log(W) and the coefficient on log(1+H/W). As to the dynamic specification, it is postulated that consumption adjusts slowly to changes in wealth; in particular, the following “quasi-error-correction mechanism” was adopted:

where Y is disposable income and D is the first difference operator.39

Tables III.3.2-III.3.4 report the OLS estimates of equation (3) (on annual data for the period 1965-85).40 Three sets of results are presented. Table III.3.2 refers to equations in which nonhuman wealth is included in its broader definition (net financial assets plus houses and rural properties). On account of possible errors in the measurement of nonfinancial wealth, Table III.3.3 reports estimates for which only net financial wealth was used. Finally, Table III.3.4 reports results referring to a simpler specification based only on disposable income. Together with coefficient estimates, t statistics and the usual goodness of fit indicators, the tables show the results of a number of “diagnostic tests” on the normality of residuals, on the absence of autocorrelation and heteroskedasticity, and on the within-sample stability of the equation (see also Table III.3.5). The latter test is of particular relevance in this context because the measurement of the overhang as an “anomalous” increase in saving requires an equation that shows stability over the estimation period.41

Table III.3.2.USSR: Estimates of Households’ Consumption (1)(OLS; annual data: 1965-85; dependent variable DlogC)
VariablesEquation AEquation BEquation CEquation DEquation E
log (W/C)-10.410.410.400.480.39
log (1+H/W)-10.230.410.40-0.38-0.08
Dependency ratio-0.0021-0.0024
Benefit ratio-0.550.700.80
Real interest rate0.0550.29
Adjusted R20.880.890.890.930.92
Standard error0.00800.00780.00760.00610.0065
Heteroskedasticity 158.8%69.3%78.8%17.1%60.7%
Heteroskedasticity 216.2%45.1%37.1%5.6%0.45%
Chow test19.0%50.0%58.5%58.1%95.3%
Variance ratio test20.3%11.0%20.3%26.7%3.9%
Generalized Chow test15.3%38.3%60.5%91.0%87.4%
Harvey’s Psi (forward)-3.860.05-0.78-2.11-0.32
Harvey’s Psi (backward)-0.26-0.561.130.42-0.72
F test on a1 = a-176.4%
Table III.3.3.USSR: Estimates of Households’ Consumption (2)(OLS; annual data: 1965-85; dependent variable DlogC)
VariablesEquation AEquation BEquation CEquation D
log (WFIN/C)-10.260.420.370.30
log (1+H/WFIN)-10.190.430.370.27
Dependency ratio-0.0011-0.0010
Benefit ratio-0.58-0.32
Real interest rate0.015-0.015
IR 10.09
Adjusted R20.870.890.890.86
Standard error0.00830.00780.00770.0087
Heteroskedasticity 177.6%64.3%83.7%97.9%
Heteroskedasticity 23.1%37.6%44.8%10.5%
Chow test23.5%47.0%64.5%20.9%
Variance ratio test22.1%9.6%25.9%2.8%
Generalized Chow test32.3%32.2%70.3%5.9%
Harvey’s Psi (forward)-2.580.17-1.01-1.22
Harvey’s Psi (backward)0.44-0.791.341.32
F test on a1 = a-143.6%
Table III.3.4.USSR: Estimates of Households’ Consumption (3)(OLS; annual data: 1965-85; dependent variable DlogC)
VariablesEquation AEquation BEquation C
log (Y/C)-10.380.380.36
Dependency ratio-0.0010-0.0008
Benefit ratio-0.12-0.09
Real interest rate-0.04
Adjusted R20.870.890.86
Standard error0.00840.00780.0085
Heteroskedasticity 195.4%77.9%85.4%
Heteroskedasticity 223.7%40.4%27.6%
Chow test15.4%61.2%20.6%
Variance ratio test15.7%22.1%26.3%
Generalized Chow test7.0%64.1%67.1%
Harvey’s Psi (forward)-1.83-0.86-0.57
Harvey’s Psi (backward)0.181.351.95
Table III.3.5.USSR: Estimates of Households’ ConsumptionDefinitions
Y = households’ disposable income (per capita in real terms)
W = total nonhuman wealth (per capita in real terms)
C = households’ consumption (per capita in real terms)
H = human wealth (per capita in real terms)
WFIN = net financial wealth (per capita in real terms)
IR1 = Holzman’s disequilibrium indicator (Chart 1, lower panel)
IR2 = relative kolkhoz prices (Chart 1, top panel)
Normality = Lagrange multiplier test of normality of residuals (Jarque and Bera (1980))
DW = Durbin-Watson test
MLM = Lagrange multiplier test of autocorrelation of residuals modified for small samples (Harvey (1981))
Ljung-Box = portmanteau test of autocorrelation of residuals (Ljung and Box (1978))
Heteroskedasticity 1 = test of autoregressive conditional heteroskedasticity (Engle (1982))
Heteroskedasticity 2 = test of linear dependence between residuals and regressors (Breusch and Pagan (1979))
Chow test = test of stability of equation parameters against the hypothesis of structural break in 1975
Variance ratio test = test of stability of residual variance against the hypothesis of structural break in 1975 (Phillips and McCabe (1983))
Generalized Chow test = Wald test of stability of equation parameters against the hypothesis of structural break in 1975 in the presence of instability of residual variance (Honda and Ohtani (1984))
Harvey’s Psi (forward) = stability test based on recursive estimates (starting from the beginning of the sample period) (Harvey (1981))
Harvey’s Psi (backward) = stability test based on recursive estimates (starting from the end of the sample period) (Harvey (1981))
Note: t statistics are reported in parentheses below the coefficient estimates. For all the above tests, with the exception of the DW and of Harvey’s Psi test, the table reports the percentage of the appropriate test distribution lying on the right of the computed test statistic under the null hypothesis of absence of misspecification. The lower bound of the DW statistic (at the 5 percent level) is 0.637 in the equation with eight regressors and 1.026 in the equations with four regressors (including the constant). Harvey’s Psi test has a t distribution with T-K-1 degrees of freedom.

The specification search process begins, for each of the three formulations, by estimating an equation containing all available regressors. Starting from Table III.3.2, the more general specification (equation A, which includes Holzman’s indicator of rationing) shows a remarkably good fit, with an adjusted R2 close to 0.9 (this is particularly high considering that the estimations are on percentage changes) and a standard error of around 0.8 percent. The normality and residual autocorrelation tests are easily passed, but one of the two heteroskedasticity tests shows unsatisfactory results. Traces of instability are also apparent; Harvey’s Psi test, which is distributed as a Student’s t, falls well beyond the 5 percent critical value. As to the coefficient estimates, the human and nonhuman wealth effects and the impact effect of disposable income have the correct sign; the sign of all other variables is opposite to the expected sign. Most t statistics, including the one on the variable containing human wealth, are low. In summary, while the overall performance of the equation appears to be adequate, there are signs of mispecification; or, possibly, the limited number of observations available does not allow the identification of all the expected effects according to economic theory.

In equation B, all variables that had the “wrong” sign and were not significant at the 5 percent level in equation A were removed. Specification B improves substantially over specification A. While there is some deterioration in the autocorrelation tests (which however remain much above their critical values), the heteroskedasticity and stability tests improve (the only exception being the variance ratio test). Note also that the t statistics on the coefficients on both human and nonhuman wealth increase well above the critical level. Finally, these coefficients, even if unrestricted, converge to a similar value (0.41), as predicted by the theory. The equality restriction is imposed in specification C and is easily accepted (see the last row of the table). Note that the equation’s standard error is further lowered and that the variance ratio test improves. In summary, specification C seems to be entirely adequate in terms of goodness of fit, diagnostic tests and consistency of parameter values with what could be expected from economic theory. Actual and fitted values for this equation are shown in Chart 4.42 Note that equation C implies that Soviet households maintain in the long run a broadly stable ratio between consumption and total (human and nonhuman) wealth; the fact that the coefficient on the change in disposable income is lower than unity implies that, in the short run, the saving ratio is correlated to the growth rate of disposable income, possibly because households try to offset partially the effect on consumption of changes in income that are perceived to be temporary.


(Equation C of Table III.3.2)

(Percentage growth rates)

In equation D the unweighted ratio between kolkhoz and state prices replaces Holzman’s index as an indicator of rationing. Specification D has a slightly better fit, but shows signs of instability; and, while the rationing index and the benefit ratio have the expected sign, the interest rate level, the dependency ratio and the term including human wealth have the wrong signs. Some of these problems persist in the more parsimonious specification E (where the dependency ratio and the interest rate have been dropped).43

The specification using net financial wealth rather than total nonhuman wealth as a regressor (Table III.3.3) does not substantially alter the empirical results just described. The properties of the equation remain approximately unchanged and the specification search leads again to a parsimonious representation (equation C) in which, in the long run, the ratio between consumption and (human and nonhuman) wealth appears to be fairly stable. Note, however, that, contrary to equation D in Table III.3.2, the relative kolkhoz price has an extremely low coefficient and t statistic, suggesting that the correlation found before may have been spurious.

Finally, Table III.3.4 presents estimates of a more traditional specification in which only disposable income influences consumption expenditure. The performance of the equation remains satisfactory in terms of fit and diagnostic tests. Again, the relative kolkhoz price appears in specification C with a very low coefficient and t statistics.

In conclusion, the behavior of consumption between 1965 and 1985 can be adequately described by the evolution of total (human and nonhuman) wealth; the fact that a simpler specification in terms of disposable income (on which the computation of human wealth is based) closely approximates the results obtained by the more complete specification is probably a reflection of the fact that, in the USSR, households’ wealth is predominantly “human”.44

c. Estimates of households’ accumulated excess savings

(1) The wealth overhang

The equations described above can be used to compute the size of the wealth overhang accumulated in the second half of the 1980s. A preliminary step is to show that the estimated consumption equations, very stable until 1985, present strong instability in the subsequent period. For this purpose, equation C in Tables III.3.2 and III.3.3 and equation B in Table III.3.4 were re-estimated over the 1965-90 period; the stability tests decisively rejected the hypothesis of structural stability.45

The calculation of the wealth overhang of households was then based on the difference between actual wealth accumulation and the amount of wealth that would have been accumulated in the absence of rationing on the basis of the estimated econometric equations. Table III.3.6 presents two sets of projections; the first is based on a simulation of equations C in Tables III.3.2 and III.3.3 and of equation B in Table III.3.4 for the period 1986-90; the second is based on a simulation of the same equations for the period 1983-90, as a check of the hypothesis that rationing (or rationing above a structural stable level) started in 1983.46

Table III.3.6.USSR: Estimates of the Wealth Overhang of Households in 1990(In billions of rubles)
Based on



Estimate of



Equation C (Table III.3.2)10059159
Equation C (Table III.3.3)10359162
Equation B (Table III.3.4)10459163
Equation C (Table III.3.2)12448172
Equation C (Table III.3.3)13148179
Equation B (Table III.3.4)13548183

The first column of Table III.3.6 shows the estimates of the accumulated wealth overhang during the simulation period derived from the model simulations; they range from a minimum of around rub 95 billion to a maximum of rub 130 billion. As mentioned, these estimates do not include the “unobservable” chronic overhang which may have affected consumption even before 1986 (1983). Taking a figure of rub 59 billion (derived by the Soviet authorities from sample surveys of households’ behavior) as the estimate of the “chronic” overhang at the end of 1985, and rub 48 billion for the end of 1982,47 we obtain, in the last column of the table, estimates for the overhang in 1990 ranging between rub 159 billion and rub 183 billion with an average of rub 170 billion (around one third of households’ financial wealth). Using the same procedure, we obtain an average estimate for the overhang at the end of 1989 equal to rub 128 billion. Thus, during 1990 the monetary overhang appears to have increased by around rub 42 billion (over 55 percent of households’ saving in that year).48

Three observations must be made with reference to these figures. First, as mentioned in section 2, the measurement of consumption may be underestimated (and that of the overhang may be overestimated) by lack of adequate information on incomes generated on parallel markets.49 In addition, the estimate of desired consumption, and hence that of the overhang, may be affected by lack of consideration of how expected future price increases may affect the desired intertemporal distribution of households’ resources.50

Second, the amount of the overhang should be compared to the stock of unsold consumer goods. The latter appears to have declined sharply during the last twenty years; the estimated stock was rub 62 billion in 1970, rub 31 billion in 1985, and rub 15 billion at the end of 1989—a very modest proportion of the overhang, even at official prices. This finding suggests that, in the presence of price liberalization, the overhang would not lead simply to a change in relative prices (e.g., a decline in the price of the unsold inventories of consumer goods and an increase in the price of the goods for which shortages are observed) but would bring about an increase in the average price level.

Third, it must be recalled that the reported estimate of the overhang refers to the cumulative sum of involuntary saving incurred in past years. Under normal circumstances (i.e., assuming that behavior continues to conform to the estimated equations), Soviet households are unlikely to try to spend all the undesired accumulation of wealth in a single period (say one year). According to the life cycle hypothesis, they should allocate the expenditure of the undesired stock of wealth over time. Of course, a quantitative estimate of how rapidly the overhang would be released in the presence of price liberalization would be extremely difficult to make; it would depend on the type of consumption foregone in the past,51 on price expectations and on the extent to which price liberalization was seen as temporary or permanent.

(2) The monetary overhang

Table III.3.7, describing the trends in the composition of households’ wealth and savings, sheds some light on the question of what proportion of the estimated wealth overhang might have been accumulated in monetary assets. The table shows a steady increase in the financial component of total wealth, almost entirely matched by a decline in the share of housing. Although the latter may be influenced to some extent by miscalculations in the valuation of houses at current prices, the increased role played by financial wealth is confirmed by the data on the allocation of saving flows. Investment in real assets fell from around one-third of total saving during the late 1950s and early 1960s to less than 10 percent in subsequent years, mainly as a consequence of the decline in the share of private housing investment.52

Table III.3.7.USSR: Composition of Households’ Wealth and Saving(Percentage shares)
Wealth 1
Net financial wealth32.740.457.170.779.983.387.2
Other real assets4.
Saving 2
Net financial wealth66.765.884.989.893.090.491.9
Other real assets4.
Sources: See statistical note.

Composition of wealth at the end of the period.

Average composition of saving during the period.

Sources: See statistical note.

Composition of wealth at the end of the period.

Average composition of saving during the period.

Clearly, the composition of households’ wealth is currently biased towards financial assets, which account for around 90 percent of both total wealth and saving.53 In addition, financial wealth is almost entirely composed of monetary assets; throughout the period under consideration, M1 components (cash and demand deposits) represented around two thirds of total financial assets; and M2 components (including M1, time deposits and lottery bonds54) covered a share slightly above 90 percent.

The main conclusion that can be drawn from these data is that, given the limited share taken by the real component of saving in the last 20-25 years, almost all of the involuntary accumulation of saving is likely to have been in the form of financial, and specifically monetary, assets.

d. Estimates of the monetary overhang of enterprises

Enterprises’ monetary holdings are likely to be determined basically by a transactions demand which, in theory, could be estimated. The approach used for households could then be applied, equating the overhang with the monetary accumulation that is not explained by the estimated equation. Two factors prevent the practical application of this procedure. First, the available information on enterprise money only extends back to 1980, which precludes econometric estimation techniques. Second, the reforms enacted in the second half of the 1980s significantly altered the nature of enterprise money, with potentially complex repercussions for the underlying demand function. On the one hand, the reforms increased the degree of control enterprises had over their resources; previously, their bank deposits had been held jointly with ministries and state organizations. The reform also raised the fungibility of different bank accounts, and therefore their liquidity.55 On the other hand, the demand for money may have been increased by a tightening of enterprise credit, which accompanied the reforms and partially hardened the “soft” budget constraint of enterprises.56 In addition, the increasing disruption in the distribution and trade system in the late 1980s may have increased enterprises’ demand for liquidity, as they needed to be prepared to buy inputs as and when they became available. It is difficult to do other than assume that all these factors may have offset one another, although this is little more than a working hypothesis.

With this caveat, some indication of the involuntary accumulation of enterprise money can be obtained from both cross-sectional and time series empirical evidence. It is possible, in the first place, to compare the level of enterprise money in the USSR with that prevailing in Western European industrialized countries.57 Enterprise holdings of M1 (cash and demand deposits) in relation to both NMP and GDP appears to be higher in the USSR than in the Federal Republic of Germany, France, Italy and the United Kingdom, although not dramatically higher (Table III.3.8). Corresponding ratios for M2 (M1 plus time deposits58) were lower than in the FRG, although higher than in the other countries. The ratio computed for total financial assets (which is also the M2 ratio in the USSR) fell close to the average in the five countries. Based on this evidence, it could be argued that Soviet enterprises are too liquid, given that the real yield of enterprise deposits in other countries exceeds that in the USSR.

Table III.3.8.USSR: International Comparison of Enterprise Liquidity 1
M1 2M2 3Total




(in percentage of value

added at factor cost) 4
(in percentage of GDP)
United Kingdom6.319.019.84.613.814.4

Data refer to 1988 for all countries except for the USSR, for which projections for 1990 have been used.

Currency and demand deposits.

M1 plus time deposits and another short-term assets.

Excluding value added of public administration. For the USSR, NMP net of turnover tax but with depreciation added.

Data refer to 1988 for all countries except for the USSR, for which projections for 1990 have been used.

Currency and demand deposits.

M1 plus time deposits and another short-term assets.

Excluding value added of public administration. For the USSR, NMP net of turnover tax but with depreciation added.

Consider now the time series evidence, based on a plotting of the ratios of M1 and M2 to NMP (Chart 5). In 1987, both ratios started rising rapidly, doubling in four years. Notwithstanding some of the factors mentioned earlier, it is difficult to believe that this could have reflected equilibrium behavior. It may not necessarily reflect shortages of goods either, but may simply have been a short-run response to sharply increased profits during this period. In the long run, however, given the current level of interest rates, the high level of liquidity is likely to lead to increased expenditure.


(Percentage ratios)

1. In relation to NMP.

1. In relation to GDP.

How large is the excess accumulation of money by enterprises? One, admittedly rough, approximation could be derived by extrapolating the growth of liquidity that would have maintained the ratios of liquidity to NMP at their 1986 value. If this is done, the excess liquidity held in 1990 is rub 79 billion for M1 and rub 99 billion for M2. If, instead, we assume, as “normal” liquidity growth, the increase that would have kept the liquidity ratios close to the trend increase observed between 1980 and 1986, we obtain an estimate of the overhang equal to rub 65 billion for M1 (around 40 percent of the total) and to rub 85 billion for M2 (42 percent of the total). The corresponding figures for the end of 1989 are rub 48 billion for M1 and rub 53 billion for M2. Thus, the enterprise overhang is estimated to have increased in 1990 by rub 17 billion for M1 and by rub 32 billion for M2. Similar values are obtained by using ratios with respect to GDP instead of NMP.


The apparent substantial (and rapidly rising) involuntary accumulation of monetary holdings for both households and firms suggests large and growing shortages. At the same time, it would be a cause for concern on the eve of a possible price liberalization, and dealing with it must be an integral part of any stabilization strategy. Certainly, all factors that could possibly determine a further increase in the overhang should be promptly removed: the expansion of monetary income has to be kept in line with the development of production, for both households and firms, which requires a tightening of fiscal, monetary, and possibly incomes policies. Even if this flow problem were solved, however, the stock of excess liquidity would remain. This section considers different possible ways to reduce that stock.

a. Price increases

One obvious way to eliminate the overhang is to liberalize prices and allow inflation to erode the excess purchasing power. Two problems have to be considered in this context: first, what is the type and magnitude of the price adjustment that would eliminate the overhang; and second, under what conditions could the adjustment be limited to a once-and-for-all increase in the price level, without triggering an inflationary spiral.

Let W and WF be the actual amounts of total and financial wealth, respectively, held by Soviet households in 1990, and Wd and WFd the corresponding desired amounts. Under the assumptions that: (a) the desired amount of wealth in real terms and actual nominal wealth are constant59; and (b) the nominal value of the non-financial component of wealth (W-WF) (mainly houses) increases in line with prices, the price level which would have equated actual and desired wealth in real terms is obtained by solving for P in the equation:

where OV is the overhang. This implies an increase in the price level of 45.3 percent with respect to 1990.60

Note that the price increase, at least in the short run, is not the only adjustment necessary for the elimination of the overhang. The overhang was created by an excess of income relative to the available supply of consumer goods; this disequilibrium fueled the increase in the stock of the overhang and has to be eliminated if the source of the overhang is to be eliminated permanently. This will probably imply a temporary drop in real households’ disposable income with respect to its 1990 level.

A 45 percent price increase may be considered a reasonable cost to pay for price liberalization. Several factors, however, could push prices higher than this. First, some relative prices (e.g., energy prices) will have to increase much more than 45 percent. In the presence of downward price rigidity or of minimum price increases connected to wage indexation (a likely occurrence),61 this may imply an average price increase above 45 percent. Second, a 45 percent price increase would be insufficient to absorb the enterprise overhang, which is estimated to equal 75 percent of this sector’s desired money holdings;62 moreover enterprise profits will probably rise as consumers’ expenditure increases. Enterprises will therefore have an incentive to spend the undesired money accumulation, bidding up goods prices and, possibly, wages. Expectations would be likely to play a crucial role; in the current situation of political uncertainty, with fears of falling production, and a complete lack of experience in market mechanisms, initial price increases may easily be interpreted as signaling the beginning of an inflationary spiral. These expectations may be self-fulfilling as they could lead to increased hoarding of goods, highly negative expected yields on financial assets and, therefore, to a stronger increase in aggregate demand. Only through tight monetary and fiscal policies could this process be halted.

In conclusion, while in theory the absorption of the monetary overhang purely through price liberalization may involve moderate price increases, the practical difficulties in avoiding an inflationary spiral should not be underestimated. Consideration must be given to non-inflationary ways of eliminating the overhang.

b. Increased availability of goods

An increase in the supply of consumer goods has been one of the cornerstones of all recently proposed reform programs in the USSR. To the extent, however, that this involved shifting resources from the production of investment goods to consumer goods, it could reduce the overhang of households but at the risk of aggravating shortages in the investment goods market. Unable to increase capital investment (or facing higher prices for investment goods), firms might be induced to use their large liquidity holdings to bid for more labor, thus pushing up wages.63 A solution might be to cut down the production of goods which do not supply the market (primarily military investment); or to increase overall productivity. In both cases, the supply responses would take time to materialize. A more rapid increase in the supply of consumer goods might be possible through stepped-up imports, but only if the external financing constraint would permit.

c. Monetary reform

The cancellation or write-down of outstanding monetary assets, i.e., monetary reform, has also been considered as a way to reduce the overhang. Here the problem is one of credibility. Some distinction can be made, however, between the deposits of households and enterprises.

For households, the distribution of bank deposits appears to be fairly concentrated in the USSR (and rather stable over time; see Chart 6): over 40 percent of bank deposits are accounted for by only 10 percent of bank accounts and over 60 percent by only 20 percent. At the end of 1988, around 35 percent of the entire stock of deposits consisted of accounts larger than rub 5,000 (3.5 times the value of per capita annual consumption in that year).64 The distribution of bank deposits per capita (or household) is probably even more concentrated because of the widespread practice of breaking larger accounts into a number of smaller accounts, for fear of administrative measures against bank deposits.65


The high concentration of bank deposits implies, in principle, that a monetary reform that converted larger deposits at an unfavorable rate would affect a limited share of the population. Using end-1988 figures, for example, the conversion of deposits below rub 5,000 at par and of all deposits above rub 5,000 at, say, a 2:3 ratio (two new rubles for three old rubles) would have reduced the outstanding stock of deposits by around rub 60 billion and would have involved less than 20 percent of the stock of bank deposits. Assuming no change in bank distribution, the corresponding estimate for end-1990 would be around rub 80 billion.

There is, however, a political problem in implementing a reform of this kind; the last monetary reform in the USSR was implemented under Stalin (in 1947) and the practice of multiple accounts seems to indicate that the memory of that reform still lingers.66 Instilling popular confidence in the inviolability of property rights is an important element in making a successful transition to a market economy. To begin such a transition with an arbitrary confiscation of assets (however illusory their purchasing power may have been) risks undermining the credibility of the entire strategy. This is not to say that some limited reform (possibly in the form of a temporary freeze), or some type of tax on wealth, should be entirely ruled out; but its contribution to the absorption of the households’ overhang should, in most circumstances, be limited.

Administrative measures may possibly be politically more realistic with respect to enterprise deposits. Clearly, the advantage here would be that the “shareholder” is the state itself; the disadvantage is that a measure of confiscation (or conversion at very unfavorable rates) of enterprise liquidity would be against the free market spirit of the reforms and it may in particular discourage the profit motivation of enterprises. One risk is the possible disruptive effect on production of a sharp decline in the liquidity of firms; in order to reduce this risk, the targeted cut in liquidity should remain somewhat below the estimated monetary overhang.67

d. Sale of state property

The sale of state property represents, in theory, an effective way to reduce the monetary overhang. Since the composition of households’ wealth in the USSR is severely biased towards monetary assets, it is likely that the potential demand for real forms of wealth is high. It is also likely that the increased availability of real stores of wealth could increase the desired amount of wealth (or, in other words, the propensity to save), thus bringing about not only a decline in the liquidity of the overhang, but also a decline in its absolute size.

Housing is obviously the first candidate for privatization. The pent-up demand of Soviet households for houses is well known.68 Given the rather concentrated distribution of bank deposits, however, only a small proportion of households would have enough resources to purchase a house at nonsubsidized prices (or even to make a downpayment). Of course, credit could be extended to sustain the purchase of houses; in this case the initial amount of the overhang would not be immediately affected, but the gross saving rate in future periods might increase, as households had to face the repayment of the debt.69

The amount of resources that could be raised by the sale of houses, and land, will crucially depend on two factors. First, the Government must be able to offer a credible guarantee of property rights, which would require the introduction of legislation specifying in detail the procedures for the registration of property and for its transfer. Second, rents must be substantially raised (see Chapter V.9). On account of all these factors, any estimate of the amount of money that the Government could raise by selling property is extremely uncertain. The value of the stock of houses owned by the Government is officially put at about rub 800-900 billion at the end of 1989. The privatization of even a small proportion of this stock could offset a substantial share of the overhang.

As for the sale of enterprise shares on assets, again the authorities must develop a system of credible property rights guarantees.

e. Increase in interest rates and financial innovation

Until 1990, the yield on bank deposits in the USSR had remained fixed in nominal terms for more then 25 years. In real terms, it followed a slightly declining trend, becoming generally negative at the end of the 1980s, thus exacerbating the overhang. Despite increases toward the end of 1990, real interest rates remained negative. Raising the return on bank deposits and in general improving the conditions offered on financial assets is an effective way, under normal conditions, to increase saving rates and the demand for financial assets.

However, increasing deposit rates would require that loan rates on bank assets be raised accordingly. While this increase would certainly be beneficial in reducing the enterprise overhang, it could have a negative impact on the fiscal deficit as it would increase the cost of servicing the public debt. In addition, a rise in interest rates generally has (opposing) income and substitution effects on current saving rates. Given the relatively low total financial assets to income ratio (the income effect increases with this ratio), the substitution effect is likely to prevail and saving would therefore probably increase, though this has yet to be confirmed empirically in the case of the USSR. At the same time, efforts should be taken to develop a sufficiently wide market for bonds; in this context the issue of medium-term indexed bonds may represent the only way to support a bond market.70


During the second half of the 1980s, continued price controls, coupled with an increasingly inadequate supply of goods relative to nominal household and enterprise incomes, stimulated the accumulation of a substantial amount of unspent purchasing power held in the form of monetary assets. It is estimated that in 1990 this “monetary overhang” reached rub 170 billion for households and rub 85 billion for enterprises. These amounts represent fairly substantial shares of the monetary holdings of the two sectors (33 percent for households and 43 percent for enterprises). These estimates should be interpreted with some caution since they are based on several assumptions, the most important of which is that agents’ underlying behavior has remained broadly stable over time. Should, for example, price expectations or lack of confidence in government policies destabilize established patterns of saving and expenditure, the amount of undesired money could easily and rapidly change.

While the stock of the overhang is significant, its high growth rate is particularly alarming. The total overhang is estimated to have increased in 1990 by over rub 75 billion (or by 41 percent). Clearly, a lasting solution to the problem requires the implementation of fiscal, monetary and incomes policies capable of checking the current income streams fueling the overhang. However, in the context of any stabilization effort, appropriate policies will also have to be implemented in order to reduce the current excess stock of money. Even in the absence of new liquidity injections, and without other measures to reduce the monetary surplus, the price increase required to remove the overhang could be fairly substantial (40-50 percent). The real risk is, however, that the first round of price increases would start an inflationary spiral. This, of course, is more likely in the presence of indexation mechanisms (of both labor incomes and interest rates), which may be difficult to avoid. In summary, relying on price increases to eliminate the overhang may involve a lengthy and potentially disruptive adjustment process. Other, noninflationary, ways to deal with the problem should also be considered.

Four such types of solution have been discussed. The sale of government assets is one option. While the overhang is large compared to the wealth of the population, it is small with respect to the value of state property, so that the sale of even a small part of the latter could easily absorb the overhang. The difficulties are in rapidly creating the legal and economic conditions that make the purchase of state properties attractive. Increased goods supply is another obvious solution and should be implemented through a cut in the production of goods which are not demanded by the market (mainly military hardware). Massive conversion in the composition of production has occurred in many economies in a relatively short time, but the same process may take longer in the USSR, given the current state of disarray in the economy. It is also possible that, given realistic exchange rates, a large portion of the overhang might be absorbed by a relatively modest expenditure of foreign exchange (if used to import goods in strong demand). Monetary reform is a third possibility; the highly concentrated distribution of bank deposits would imply that a relatively large share of households’ excess money could be absorbed by penalizing a relatively small proportion of the population. However, this approach has the drawback that it would tend to work against the attempt to establish credibility and the population’s confidence in the inviolability of property rights. More feasible could be the adoption of measures to reduce administratively, or to freeze, the excessive liquidity of state enterprises, possibly in the context of their conversion into joint-stock companies. Finally, an increase in the demand for financial assets, preferably less liquid than bank deposits, could be facilitated by an increase in interest rates and by the introduction of new financial instruments (such as indexed bonds).

All these instruments present advantages and drawbacks, and do not necessarily affect all population groups equally. In these circumstances, it would seem unrealistic to rely solely on any one instrument. Rather, recourse to a broad spectrum of measures is likely to prove the most effective strategy.


  • Financial assets and saving: all data on financial assets between 1964 and 1989 have been provided by the Gosbank; data on bank deposits for the previous period can be found in Hutchings (1983), who collected them from Soviet official publications; data on currency and other financial assets of households have been estimated by imposing a constant ratio with respect to bank deposits (at the 1964 level). This may underestimate the actual amount of cash before 1964 if the declining trend in the ratio between cash and deposits observable after 1964 had started before that date. In the absence of capital gains on financial assets, net financial saving has been equated to the change in the nominal stock of financial assets net of the change in households’ credit, also available from the above sources.

  • Households’ disposable income: data on disposable income for the 1980s have been provided by Goskomstat; the behavior of the series is similar for that period to the one published in CIA (1989), which has been used to extend the series back to 1965; information on the previous period has been derived from Pickersgill (1983).

  • Real households’ investment: this is made of two components: houses and other real investment (both considered net of amortization). As to the first component, Goskomstat provided data from 1970 to 1989; data for the previous period have been derived from Smith (1973). The second component (including mainly livestock and other property of the rural population) has also been provided by Goskomstat from 1970 to 1989; for the previous period this series, for which the magnitudes are rather small, has been kept constant with respect to financial saving.

  • Households’ wealth: total wealth has been derived as the sum of three components: net financial wealth, houses and the stock of other real wealth. For financial wealth see Note 1. The value of houses owned by the population, net of depreciation and at current (official) prices, has been provided by Goskomstat from 1965 to 1989. The housing investment series has been used to derive the stock of houses for the previous period. The value of the third component has also been provided by Goskomstat but only for 1970-1989 and not at current prices, but as a cumulative sum of previous investments. For the pre-1970 period the corresponding investment series has also been used to derive the stock of this wealth component. A “discrepancy”, or “capital gain” series has also been derived by subtracting total (real and financial) investment from the change in wealth. Given the procedure followed above, this residual series represented entirely net capital gains on houses and was fairly small.

  • Households’ saving and consumption: households’ saving was derived as the sum of the three saving components (net financial saving, investment in houses, investment in other real assets). Consumption was derived residually as the difference between disposable income and net saving.

  • Other series: the inflation series was derived from CIA(1989) and from Pickersgill (1983); data on population are derived from Narkhoz, various issues; most recent estimates of the composition of the population by age have been published by Kingkade (1987a); data from Howard (1976) have been used for the previous period. Nominal interest rates have been provided by the Gosbank; real interest rates have been derived by using the above mentioned inflation series. Data on monthly wages and the expenditure of the social consumption fund are published regularly in The USSR in Figures. Data on the ratio between kolkhoz prices and official prices for the same products have been collected in Nove (1986); data published in the Narkhoz 1988 (1989) and in Alexeev, Gaddy and Leitzel (1990) have been used for the most recent period. Data on the stock of unsold consumer goods have been provided by the Ministry of Finance.

  • All data for 1990 have been estimated, based on preliminary information.


Hidden inflation can occur for several reasons: “Official price lists may lag behind actual prices… Quality may deteriorate at constant prices. State goods may be sold at prices higher than state prices by dishonest retailers…unrecorded open inflation can arise from quantity weights in official price indices understating the relative weights of goods whose market prices (whether or not equal to the official price) rise relatively faster.” Nuti (1989), p. 110.

Barro and Grossman (1974), p. 87; and Portes (1989). Nuti (1989) points out that, strictly speaking, the term “repressed inflation” should not be used in a situation in which the equilibrium price is simply higher than the actual price, but in a situation in which the difference between the equilibrium and actual price level rises. This point, while correct, seems to be purely semantic and the more common definition is adhered to here.

Following the literature on disequilibrium markets, “rationing” should be interpreted here purely as a situation where demand exceeds supply at current prices and does not imply a specific allocation mechanism (such as allocation through coupons).

There will also be an increase in real wealth at current prices, but not at the price level prevailing when equilibrium is restored.

This is epitomized by the work of Portes and Winter (1980), which has extended to CPEs the concept of macroeconomic disequilibrium proposed for market economies by Barro and Grossman (1974).

Burkett (1988). The main problem is the difficulty of applying at the macroeconomic level the “min condition” on which the econometric analysis of markets in disequilibrium is based. The “min condition” implies that, in the presence of disequilibrium, the observed quantity is the minimum of demand and supply. While this is an acceptable (although not necessarily correct) assumption in specific markets, at the macro level, in the presence of both shortages and surpluses, the observed quantity is always lower than the sum of all demands and of all supplies, if at the micro level the min condition prevails.

In this case “it is not clear what should be called aggregate excess demand on the macro level. Should it be only money which cannot be spent at all? But this is found only with a generally very intense shortage.… Or should we interpret the notion widely and include all money that has not been spent according to initial demand? It seems that the term aggregate excess demand is not an operational category in an economy with chronic shortages.” (Kornai (1982), p. 477). Also see Kornai (1980).

“In a shortage economy an alert buyer purchases not when he wants to consume a good but when the good is available” (Kornai (1982), p. 457).

Note that this component of saving can only be considered as “voluntary” given the state of chronic shortage affecting the economy. Should the state of the economy change, e.g., due to imminent price liberalization, the component of wealth previously accumulated for this motive would immediately become part of the overhang. Note also that this component should not increase over time, in relation to total consumption, unless shortages become more extensive or more severe.

Note, however, that to net out excess involuntary saving with the involuntary accumulation of stocks would require evaluating the latter at the correct (and lower) relative price, and not at the official price at which the goods remained unsold.

According to Nuti (1989), this is indeed the most common case in CPEs. In this respect, he argues that “By and large excess money holdings in the economy as a whole (i.e., with reference to the two-tier market) can be deemed to be small regardless of reliance on econometric evidence, except for the presence of lags and other minor factors which might slow down the adjustment of money balances to the actual two-tier price level. However, the small size or even absence of excess money holdings (and therefore of repressed inflation) is consistent with large-scale imbalance in the state sector, since consumers who—given the two-tier market—voluntarily hold liquid assets, do not refrain voluntarily from converting most or all of them into goods in state shops where they are quantity-rationed.” (p. 142) On this point see also Grossman (1977).

As long as rationing is not implemented through coupons, but through queuing (so that the probability of being able to buy the desired amount of goods at the official prices is, for each individual, different from zero) it is possible that buyers prefer to delay purchases, increase queuing, and keep higher than desired money balances, rather than buying at the higher parallel market prices.

The most important legal parallel markets in the USSR have been the kolkhoz markets and the private market for housing services (mainly housing construction). Grossman (1977) estimates that in the middle of the 1970s “the contribution of private value added to household consumption must have been at least 15 percent, and, in regard to household food consumption, perhaps around 25 percent”. Note, however, that official data on expenditure and consumption partially take into account economic activity outside the state sector, although they entirely exclude incomes from illegal activities.

The following discussion is meant to provide a framework for the analysis of basic empirical evidence on consumption behavior in the USSR; it does not provide a complete survey of the debate on the existence of repressed inflation in CPEs. For surveys of this debate, see Davis and Charemza (1989), Portes (1989), and van Brabant (1990).

As explained below, the possibility of adjusting labor supply in response to goods market disequilibrium has been one of the main theoretical counter-arguments to the hypothesis of macroeconomic disequilibrium. Note however, that a reduction in labor supply may reduce the supply of consumer goods as well as household incomes, and can therefore have an uncertain effect on repressed inflation. Let:

CD = cNW.

be the demand for consumer goods, where c is the propensity to consume, N is the number of hours supplied and W is the hourly wage rate. The value of the production of consumer goods is given by:

Cs = PNqh.

where P is the price of consumer goods, q is the share of workers employed in the consumer goods industry (so that Nq is the total number of hours worked in the production of consumer goods) and h is average hourly labor productivity. Then the effect on rationing (i.e., on the percentage difference between CD and Cs) of a change in N is given by (cw-hq)/CR, where CR is consumption in real terms and w(=W/P) is the real wage rate. This effect could in principle be positive or negative. Specifically, assuming that the average wage rate equals average labor productivity (w=h), then a decline in N would bring about a reduction of rationing if the propensity to consume were higher than the share of workers employed in the consumer goods industry (a likely condition). However, w is likely to be lower than h (remember that average values are being considered); and therefore no precise conclusion can be drawn with respect to the effect on rationing of changes in labor supply.

Kornai (1980), pp. 456-459, however, recalls that some saving is needed to increment the stock of money held for transaction purposes; moreover, in the absence of consumer credit, the purchase of durables requires prior accumulation of saving; as housing credit is also scarce, a large share of housing investment has to be financed by prior saving; and a larger stock of wealth can help raise living standards above the basic level provided by the state. Finally, the existence of shortages at the micro level increases the precautionary demand for wealth (especially for monetary assets).

Kornai (1982), pp. 263-68, suggests a number of such indicators.

The meaningfulness of these indicators in assessing the existence of macroeconomic disequilibrium is also uncertain. For example, the enlargement of parallel markets and the possibility of speculation may inflate demand on the state market (thus lengthening the queues) even in the absence of macroeconomic disequilibrium.

Indicators of this kind are widely used by Soviet sources as a signal of imbalances. As recalled by Nove (1986), pp.255-256: “Some Western authors deny that the increase in savings bank deposits is abnormal… But many Soviet sources regard some of this saving as forced, frustrated purchasing power.”

The lack of consumer credit only increases saving if the population grows or the proportion of available goods that require prior accumulation of savings increases. In this respect the growth of consumer durables in the USSR in the last thirty years has been impressive; between 1960 and 1987 the number of washing machines per 100 families rose from 4 to 70, that of television sets from 8 to 100, that of cars from 1 to 16, that of refrigerators from 4 to 93 (Goskomstat (1987), p. 222). It must be added, however, that most of the improvement occurred in the 1960s and the 1970s and that little progress has occurred in the 1980s, with the exception of cars.

Data availability problems prevent the comparison of the wealth ratios in the USSR with those in countries having similar per capita incomes.

Given a consumption function of form C=cY, the equilibrium level of the wealth to income ratio (w) is given by: w = s/g, where s=1-c and g is the growth rate of disposable income (Modigliani (1986)). Since w is inversely related to g, in periods of decelerating disposable income growth it is normal to observe an increase in w even if s remains constant.

Note, however, that constant rationing of consumption would imply a rising stock of undesired wealth.

See Pickersgill (1976, 1980 and 1983) for the USSR.

A similar approach is followed by Ofer and Pickersgill (1980) based on the estimation of a consumption function on cross-sectional data.

The use of these disequilibrium indicators follows the analysis of rationed markets pioneered by Fair and Jaffee (1974). Under the hypothesis of excess demand and of the usual min condition, the behavior of consumption is described by: C = CD(.) - R, where C is observed consumption, CD is desired consumption and R is the amount for which demand is rationed. R is not observed; however, if variables can be found which are related to R (Fair and Jaffee use the price change in the rationed market under the hypothesis that the price adjustment is proportional to the size of the disequilibrium), then the above equation (and specifically the parameters of the demand function) can be estimated with ordinary econometric techniques. In their econometric model of the Soviet economy, Green and Higgins (1977) also use disequilibrium indicators to estimate consumption demand.

In her 1980 paper, Pickersgill suggests that this break could be due to repressed inflation. A different approach, unfortunately not extended to the USSR, has applied disequilibrium econometrics techniques to the consumption goods market in CPEs (Portes and Winter (1980), Portes, Quandt and Yeo (1988) and Burkett (1988), for example). These techniques, pioneered by Maddala and Nelson (1974), imply the specification and estimation of both a demand and a supply function for consumer goods; all observations are assumed to be either on the demand or on the supply function. The maximum likelihood principle is used to assign each observation either to the demand or the supply equation and to estimate the two functions. These studies, in which again consumption is expressed mainly as a function of disposable income, tend to conclude that CPEs were not characterized by a state of chronic repressed inflation (with the consequent accumulation of involuntary saving); periods of excess supply and periods of excess demand alternated and, in some cases, excess supply was the dominant regime.

The application of this technique is simpler than the one based explicitly on disequilibrium econometrics. First, estimates can be performed with OLS; second there is no need to specify a supply function, whose form would inevitably be arbitrary in the absence of appropriate models of enterprise behavior in CPEs. Finally, the use of disequilibrium econometrics would have implied the separation of the sample between points on the supply and points on the demand function; this procedure seems inappropriate given the very limited number of observations available.

Of course, if the equation did not fit the data, this assumption would be called into question.

This is an approximation; in theory one would like to include in consumption only the value of the “services” obtained from the current stock of consumer durables; in this case wealth could be defined as inclusive of consumer durables. Lack of adequate data on consumer durables, difficulties in estimating the value of their services and uncertainty on the inclusion, especially in a country such as the USSR, of consumer durables as components of wealth, ruled out such an approach.

As detailed in the statistical notes at the end of the chapter, wealth is defined as the sum of financial wealth (currency, bank deposits, Government bonds and insurance policies net of households’ borrowing), plus houses and other real wealth (mainly livestock and other property held by rural households).

Human wealth is defined by the present discounted value of disposable labor income; this has been computed by adding to a three-period centered moving average of current disposable labor income the discounted expected stream of income in the next 27 years. In this respect it has been assumed, for simplicity, that per capita real income is to grow at the constant annual rate of 2.5 percent (close to the average for the sample period 1965-85, considered in the estimates). The average interest rate on bank deposits was used as discount factor. The 27 year interval has been selected in the following way: the average expected life at birth of the population in the USSR has been close to 69 years throughout the sample period (Kingkade (1987b), p. 11). Assuming an average expected life of 72 years for the population of age 18 (taken as the average starting year of working life), 27 (= (72-18)/2) is the average number of years for which a middle aged worker expects to receive labor income (including pension payments) in the rest of his/her life.

The Social Consumption Fund provides education and health services, grants, pensions and scholarships to the Soviet population. The benefit ratio was also computed using the per capita benefits (i.e., the benefits for each member of the population) without altering substantially the econometric results.

Note that this specification does not correspond completely to an error correction applied on equation 2 as the change in disposable income, instead of the change in total wealth, appears as the “impact variable” in (3). Equation (3) implies that changes in disposable income (and hence in human wealth), possibly in connection with liquidity effects, affect consumption faster than changes in nonhuman wealth.

Limiting the estimation period largely to the Brezhnev era may help in identifying stable behavior in the absence of regime shifts. On the other hand, we are forgoing the opportunity to explain the “jump” in the propensity to save that occurred between the first and second half of the 1960s. It must be added, however, that the reliability of the data declines rapidly the further back one goes; indeed, it cannot be ruled out that the jump in the saving ratio observed in the middle of the 1960s was influenced by statistical measurement problems, particularly the absence of a direct measurement of cash holdings (see the statistical notes). The use of OLS in the estimation of consumption functions has well-known drawbacks. The main problem is the endogeneity of income with respect to consumption demand; in the context of centrally-planned economies, however, this endogeneity should not be taken for granted as total income may be entirely supply determined. More serious may be the consequences of measurement errors affecting disposable income; as consumption is derived residually, errors in the measurement of disposable income may spuriously increase the correlation between income and consumption. Little can be done about this problem, given the difficulties in selecting good instrumental variables, unaffected by measurement errors.

As detailed in Table III.3.5, Tables III.3.2-III.3.4 report for all tests (with the exclusion of the DW test and of Harvey’s Psi tests) the percentage of the appropriate test distribution lying to the right of the computed test statistic, under the null hypothesis of absence of mispecification. The null hypothesis cannot be rejected at the conventional 5 percent level if the value reported in the table is higher than 5 percent. Note that for most of the tests reported in the tables only the asymptotic distribution is known. In light of the limited number of degrees of freedom, it seems to be safe to accept the null hypothesis only when the value reported is “substantially” higher than 5 percent.

Attempts to reintroduce in this equation the variables previously excluded were unsuccessful, as these variables remained insignificant and often had the wrong sign.

The fit of equation E is slightly better than that of equation C; but choosing it as the “preferred equation” would imply that, apart from a temporary shock effect (DlogY), changes in human wealth, and hence in disposable income, have no permanent effect on consumption, a conclusion which is rather implausible.

Even in the United States, human wealth is estimated to be around 12 times nonhuman wealth (Jorgenson and Fraumeni (1989)); it is not surprising that in the USSR this ratio is around twice as high.

In a Chow test for the existence of a structural break in 1986, the F-statistics for the three equations were respectively 6.2, 5.8 and 5.5. The null hypothesis of structural stability can therefore be rejected at the 0.5 percent level. The interpretation of this result is that actual consumption was not at the desired level indicated by previous behavior.

The choice of 1983 as an alternative structural break point is to some extent arbitrary. Note, however, that in 1983 procurement prices were substantially raised without any increase in retail prices, which may have created an imbalance between income distributed and consumers’ expenditure. In addition, it should be pointed out that over the period 1983-85 all estimated equations had negative errors (showing actual consumption below its fitted value). However, the errors fall within the conventional range of two standard errors and therefore, strictly speaking, they do not appear as “anomalous”.

This figure for 1982 has been estimated assuming a constant ratio between the chronic overhang and total household wealth at the level observed for 1985.

The range covered by Soviet estimates of the monetary overhang is fairly wide. According to Goskomstat, the overhang at the end of 1989 was rub 165 billion, a figure close to the one reported by the Soviet press in that period and that would exceed the estimates here for that date by almost 30 percent. Estimates made by the Ministry of Internal Trade for 1990 (rub 200 billion) also are higher. On the other hand, the estimate provided by the Gosbank for the end of 1989 is only rub 105 billion (22 percent below the estimate in this chapter). And, according to Braginski (1990) and Petrakov (1990) the overhang in 1989 was even smaller (rub 70 billion). Thus, the estimates derived here fall well within the range of figures presented by different Soviet sources and are indeed close to the average value of those estimates.

The data on consumption expenditure used here are derived as the difference between disposable income and saving; the latter is measured quite accurately but disposable income takes into account only partially the incomes generated outside the socialized sector.

It may be argued, for example, that expected inflation could increase consumption, by reducing the yield on noninterest-bearing (or fixed interest) money holdings.

For example, past prolonged shortages of food products are unlikely to give rise to an immediate expenditure.

As reported by Smith (1973) “although the downturn in individual home building may possibly be a reflection of deliberate restrictive policies, it is more likely the outcome of an increasingly stringent problem in providing building materials” (p. 413). Only recently has the Soviet Government enacted new legislation to favor “a reversal of the long-term decline of this ‘private sector’ over a 30-year period.” (Andrusz (1990), p. 563.)

By comparison, financial wealth in G-7 countries ranges from 30 to 50 percent of households’ total wealth.

Lottery bonds are Government bonds whose return (averaging 3 percent each year) is paid in the form of lottery winnings; although their maturity is formally rather long, they can be converted into cash upon presentation at the state-owned Savings Bank.

Before the 1987 reform, the funds deposited in different accounts by enterprises could be used only for specific purposes. Thus, certain types of expenditure required the previous accumulation of funds in specific accounts, even if other funds were available in other accounts. As the “liquidity services” of one unit of enterprise deposits increased, the demand for nominal monetary balances may have decreased.

Credit to enterprises declined, in nominal terms, by 5.0 percent in 1987, 6.8 percent in 1988, and 3.8 percent in 1989 (Table K.2, Appendix II-1).

This comparison is admittedly difficult because of the structural differences between the Soviet and Western economies. However, it can provide a broad yardstick against which enterprise liquidity can be evaluated.

Including, for the USSR, the investment funds of enterprises.

This assumption corresponds to the traditional assumption that the demand for real money balances is independent of the price level (though not the inflation rate). While this assumption is easily accepted for transaction balances, it is less so in the case of balances held as a store of wealth. The optimum stock of real wealth held by households in the steady state (i.e., for a constant real growth rate) should grow in line with households’ real disposable income (or its “permanent” component) (Modigliani (1987)). As discussed below, the need to equilibrate the flow of consumer goods to its demand in real terms (i.e., the need to avoid the creation of a new overhang), implies a decline in real disposable income in the short run. Moreover, the fact that, as a result of the move to a market economy, long-term prospects for income growth might improve, can further reduce the need to accumulate wealth. It is therefore possible that the desired amount of wealth would decline in real terms after the price liberalization.

This method of computing the equilibrium price index differs from a procedure, sometimes used in the analysis of inflation processes in CPEs, which instead focuses on the equilibrium condition in the goods market. In that approach, the percentage increase in the price level necessary to eliminate the monetary overhang is derived by dividing the overhang by the initial level of consumption spending. The two approaches can yield quite different results in the event that the initial ratio of consumption to wealth is different from unity. In general, price liberalization is likely to affect nominal money holdings—as the result of both increased consumer expenditure in nominal terms and possible increases in nominal wages as enterprise profits rise—as well as prices.

It is significant that all reform plans so far advanced have included wide recourse to wage indexation as a way to protect the incomes of the population against the effect of price increases.

At the end of 1990, enterprise money holdings (M2) were estimated to have been close to rub 198 billion, while desired money holdings were estimated at around rub 113 billion (see section 3.d); thus the price increase necessary to equal their actual and desired money holdings is around 75 percent (198/113=1.752).

The presidential guidelines, approved in October 1990 (see Chapter II.3), explicitly consider the potential conflict between the need to increase the production of consumer goods and the investment demand of enterprises. They suggest that “limits on enterprises’ use of their own funds will have to be imposed for some time. This can be done through special taxes on investments.” In the guidelines there seems to be no consideration of the effect that a reduced supply of investment goods, or an increase in their cost, might have on the demand for labor and on wages.

Data on the distribution of bank deposits by size were recently published by the Soviet press (see Vechernaia moskva, February 27, 1990, p. 2). Of course, it could be argued that, as the auction price of a Volga GAZ 24-10 automobile is close to rub 100,000, deposits of rub 5,000 and more are not excessive. Here, however, the point is not to show that bank deposits are large (it has indeed already been remarked that per capita total and financial wealth is low in the USSR by international standards), but that they are unevenly distributed.

Even if no adjustment is made for this distortionary effect, wealth concentration in the USSR appears to be not too far from that prevailing in market economies. Data derived from Wolff ((1987), p. 137 and p. 153) show that the percentage of wealth owned by the wealthiest 20 percent of the population in France, Belgium, United Kingdom, Denmark, Sweden, United States and Canada averaged 73 percent (at the middle of the 1970s), ranging from around 60 percent in the United States to 85 percent in the United Kingdom. In the USSR, 20 percent of the largest bank deposits account for 63 percent of total bank deposits of the population (Chart 6).

The conversion rates used in the 1947 reform were: 1:1 for coins and bank deposits below 3,000 old rubles; 2:3 for deposits between 3,000 and 10,000 old rubles, 1:2 for deposits above 10,000 rubles; and 1:10 for bank notes. The reform therefore penalized heavily large deposits and, particularly, cash holdings (excluding coins). A nonconfiscatory monetary reform (involving the introduction of a “heavy” ruble at a 1:10 ratio, with a corresponding change in all prices and incomes) was enacted at the beginning of 1961.

Milder measures than monetary conversion at unfavorable rates (such as conversion of liquid assets into long-term deposits or government bonds) could also be considered. Monetary conversion could also be seen as one of the steps in the process of transformation of state-owned enterprises into joint-stock companies (see McKinnon (1990)). A temporary freeze of around rub 50-60 billion of enterprise deposits has recently been considered by the Soviet Government as a possible way to reduce inflationary pressures.

For a detailed discussion of the problems of the housing sector, see Chapter V.9.

It has been suggested that the sale of houses even at token prices would be beneficial for the public deficit because, in many cases, rents do not even cover maintenance costs. However, the effects on saving could be adverse, to the extent that households might perceive their wealth to have risen, and increase their consumption accordingly.

The Italian experience of the 1970s and 1980s provides a case in point. As inflation reached two digit figures after the first oil shock, the traditional bond market was disrupted and the deficit had to rely almost entirely on monetary base financing, thus exacerbating the inflation problem. A return to a nonmonetary financing of the debt was obtained initially through the issue of short term Treasury Bills and, subsequently, through indexed bonds.

    Other Resources Citing This Publication