a. Policies

The April 1991 retail price reform undertaken by the authorities of the former U.S.S.R. reduced the share of fixed prices (to 55 percent), in favor of regulated prices (15 percent) and “contractual prices between the producer and the retail unit” (30 percent). In addition, certain fixed prices were raised with a view to reducing relative price distortions and subsidies. Indeed, the reform at the producer level three months earlier (discussed below) had virtually necessitated reform at the retail level.

As the dissolution of the union proceeded, the government of the Russian Federation that took office in November 1991 prepared a bold program of price liberalization that was announced in a presidential address on October 28, 1991. The corresponding decree was signed on December 3, 1991, and postponed the date of implementation from December 16 to January 2, 1992, on the request of the other former U.S.S.R. republics.

As of January 2, 1992, about 90 percent of retail prices were in principle free (in value terms, at 1991 relative prices). ^1/ Price controls continued to apply to a number of basic consumer goods and services, including certain types of bread, milk and some milk products, baby foods, salt, sugar, vegetable oil, vodka and other spirits, electricity and fuels, matches, medicines, supplies for the disabled, housing rents, public utilities, and public transportation and communication services. Furthermore, the state distribution sector remained subject to a ceiling on its mark-up ratio which was, however, raised to 25 percent (45 percent for the far northern regions). Administered prices for most of the above-mentioned goods were raised by 3 to 5 times. Some state-controlled prices, however, such as air fares, were increased much more.

On March 7, 1992, a federal government decree allowed local authorities to abolish limits imposed on the prices of bread, milk, kefir, skim yogurt, sugar, salt, vegetable oil and matches. This decision generalized the permission granted to local governments in the course of January and February to adjust some of the regulated prices or to free them, as the budgetary resources required to maintain the controlled prices were not always forthcoming, causing some of the price ceilings to be exceeded. In some cases, however, liberalization was short-lived: e.g., Rossiskaya Gazeta reported around mid-March that the Krasnoyarsk krai had reverted to price controls cum subsidies on bread and milk after one week of free prices. Table 1 shows the extent of remaining controls on food prices by mid-1992.

Another limitation on the freedom to set prices in the first half of 1992 was the obligation for enterprises classified as monopolists by the state anti-monopoly committee to notify the state committee on prices (Goskomtsen) of increases in their prices and to abide by pre-defined limits (varying across branches) on profitability ratios. ^2/

b. Data

Prior to the analysis of the price movements in 1992, a few clarifications are in order regarding the methodology and the meaning of the main price indices (presented in Table 2).

The traditional retail price index (RPI) computed by Goskomstat is essentially a Paasche index, published at a monthly frequency since January 1991. The weights are derived from information on current retail sales and paid services. Hence, the month-on-previous-month percentage changes of the RPI cannot be chained when the weights associated with the individual prices vary. In the early months of 1992 for example, the structure of consumption shifted rapidly, implying that chaining monthly RPI inflation data would distort actual price movements. The direction of the resulting net bias is ambiguous, and depends on the relative strength of substitution and income effects. Nevertheless, in the absence of a monthly fixed-weight index, monthly RPI data have been chained within 1991 for the derivation of real wage levels (see Section III, infra). ^1/ Other problems pertaining to the RPI reflect the fact that seasonal and other very-short-term variations (such as strikes, or unusual weather) influence the current weights, ^2/ and the use of delivery as opposed to sales data in the computation of some of the weights (deliveries are more subject to “lumpiness” than sales). In addition, the collection of accurate retail turnover data is increasingly difficult as distribution channels become more diversified. ^3/

As an intermediate step en route to a standard Laspeyres consumer price index (CPI), Goskomstat introduced a “hybrid CPI”. The latter aggregates 140 sub-indices from the traditional RPI weighted by coefficients derived from a 1990 household budget survey, rather than from retail sales records. Since the “hybrid CPI” is a derivative of the RPI, it shares most of its weaknesses.

During the first semester of 1992, a monthly Laspeyres CPI was developed by Goskomstat, with the technical assistance of the IMF’s Department of Statistics, first for Moscow and then for an aggregation of 27 Russian urban areas. This nationwide index is henceforth referred to as the urban CPI. The reference group is workers’ and employees’ households (pensioners’ and farm workers’ households are excluded). ^1/ The weights are fixed and reflect the structure of consumption of an average urban household in 1991-92 (an adjustment was made to the 1991 weights to account for the change that occurred in the first quarter of 1992). ^2/ The index is computed at a monthly frequency starting in December 1991, with the price registration dates falling between the 11th to the 20th of each month. For comparison purposes, the December 1990 value of the index has also been computed. The underlying basket incorporates 262 items (as opposed to around 1,300 in the traditional RPI), the main groups of which are shown in Table 3. Each item is in principle priced at least at 6 outlets (including, when applicable, street trading).

One official index is computed at a higher frequency, but it only covers food products. This is a 70 item, weekly index of food prices registered in 132 cities, currently published every Tuesday in Izvestia. The behavior of this index shows that within-month and average monthly inflation at times differ substantially in a period of high and volatile inflation. ^3/

Alternative price indices are sometimes reported in the press, particularly those computed by the weekly publication Commersant. ^4/ The coverage of these indices appears to be rather narrow, however, and the inflation rates reported in Commersant are not consistent across issues. These series are therefore omitted from the statistical appendix.

Other shortcomings associated with the indices used in this paper are related to the belated recognition of price increases which had been disguised as quality improvements, to the aggregation problems encountered in the context of massive price liberalization and to the fact that the deflationary trends in black markets tend to be overlooked.^5/ Lastly, information on seasonality is insufficient to construct properly deseasonalized series.

c. Developments

Of particular interest are the size of the January 1992 price jump, the subsequent pace of inflation, the extent of price reversion or overshooting, relative price changes and the degree of price dispersion.

The magnitude of the price jump in Russia in January 1992 by far exceeded what had been observed in Eastern Europe at the time of large scale price liberalization in 1990-91 (Charts 1 and 2). Depending on the index considered--and ignoring the predictably much smaller rise in the kolkhoz (collective farm) markets--prices increased, on average, by 3 to 4 times in Russia in one month, while the largest monthly jump in the sample of four Eastern European countries considered in Table 4 was a 123 percent increase (in Bulgaria). Russia’s replacement of the traditional turnover tax and the recently instituted 5 percent sales tax by a 28 percent value added tax in January 1992 would only explain a small fraction of the difference. More important, perhaps, was the effect on expectations of the Government’s announcement, in late 1991, that prices would rise by 3 to 5 times, and indeed the raising of certain key remaining administered prices by such amounts on January 2, 1992. Another potential explanation is the apparent existence of a larger monetary overhang in Russia than in other former centrally-planned economies. In relation to Poland for example, where the extent of accumulated past repressed inflation in 1989 may have been of a similar order of magnitude to that in Russia in late 1991, the significantly greater price burst in Russia could simply have reflected the fact that up to 50 percent of prices in Poland had already been liberalized prior to the initiation of its stabilization program in January 1990. ^{1/ 2/} Inflation following the price jump was also relatively high in Russia, remaining in the double digits throughout the first semester of 1992, whereas it subsided to single digit levels within two months in Eastern Europe.

The Russian price jump measured by the fixed-weight urban CPI was considerably larger than that measured by the current-weight RPI (296 versus 221 percent), while the “hybrid CPI” fell in between (at 245 percent). The interpretation of the differences between these alternative measures is not straightforward. For one thing, the sheer size of the price shock and the associated confusion (see infra) may have implied that the RPI measure of the price jump lacked precision. For another, it should be recalled that the weights used for the urban CPI broadly reflect the structure of consumption in early 1992, rather than in December 1991. Given the marked shift in consumption patterns that accompanied price liberalization, this measure is likely to incorporate the downward influence of the pure substitution effect. However, with food prices rising much faster than other prices (see infra), and with consumption having shifted towards food, it also embodies the strong upward influence from the income effect.

At the time of the price jump, numerous press reports mentioned price increases for individual items that were much larger than 3 to 4 times. ^1/ Table 5 and Chart 3 show that this was locally the case for prices in state stores of various food items. Table 3 shows that the price of butter, on average, increased tenfold. Two factors were at work. On the one hand, as was the case for butter, substantial relative price changes were required to reduce previous distortions. On the other hand, overshooting apparently occurred in a number of cases: Chart 3 shows significant price reversion after January for beef, for example, and to a lesser extent for butter, in state stores across cities. Similarly, Chart 4 reveals instances of overshooting for beef, milk and potatoes in some city markets. Even at the national level, the price of eggs overshot somewhat in January (Table 3). Weekly data show that nationwide overshooting was more widespread than monthly data may suggest, for example for beef and vegetable oil. ^2/ Whether the overall consumer price level overshot is a more difficult issue, discussed in Section IV.

One of the key relative prices is that of food. During 1991, it rose by about 12 percent (Table 3). In January 1992, it increased by another 16 percent. While it fell somewhat during subsequent months, it still stood 6 percent above its December 1991 level at the end of the first half of 1992. In part, these developments are related to the removal (and reintroduction, in some cases) of food price subsidies. They probably also reflect the shifts in relative supply and demand of food versus non-food items.

A certain degree of geographical price dispersion is to be expected in any country. In Russia, however, the geographical price dispersion registered around the time of the price jump, and for some items in the following months as well, seems extreme. In part, this probably reflects the fact that distances are great, implying larger transportation costs and greater potential for local monopoly power. Also, local subsidization varies considerably, as can be inferred from the observation that geographical price dispersion is much smaller across city markets (or “free markets”) than across state stores. But the fact that prices sometimes vary by a factor of ten or more as shown for state stores (Table 5) may also reflect the price uncertainty and confusion that is bound to accompany radical liberalization. ^1/ One would expect that effect to be largely dissipated after a few months, however, and indeed for a number of goods, geographical dispersion dropped remarkably between end-December 1991 and June 1992, presumably as a consequence of declining subsidies, increasing competition and diminishing price uncertainty. For example, the coefficients of variation for prices in state stores for beef, eggs, milk, butter and vegetable oil fell sharply between end-December 1991 and July 1, 1992 (respectively from 0.41 to 0.11, from 0.31 to 0.18, from 0.96 to 0.45, from 0.97 to 0.11 and from 1.26 to 0.31). ^2/

Another facet of price dispersion is the discrepancy between prices on city markets and prices in state stores (Table 6). The ratios of the former to the latter vary considerably across products and regions. In the case of milk, for instance, they remained significantly above unity through June 1992, since milk sales in state stores generally continued to be subsidized. For eggs or butter, in contrast, they fell rapidly, even dropping below unity in some cases. As to regional variations, the Moscow price ratios were typically among the highest throughout the first semester of 1992.

A final noteworthy aspect of price dispersion is the regional variations in inflation rates. For selected individual food items, these differences can be derived from the data presented in the rows of Table 5. At the aggregate level, the comparison between the two last columns of Table 2 shows that consumer prices rose more rapidly in 1991 in Moscow than in the rest of the country; accordingly, the January 1992 price jump was smaller in the capital.

a. Policies

An important reform of producer prices took place in January 1991, shifting many prices from the fixed to the “contractual” category. ^3/ After the reform, contractual prices accounted for 40 percent of the total in light industry, 50 percent in machine construction, and about 25 percent in the raw materials, energy, and metals sectors. As before the reform, the prices of “new products” were allowed to be set on a contract basis. ^4/

The January 2, 1992 price liberalization resulted in about 80 percent of the producer prices being free (again, in value terms, at 1991 relative prices). A restricted list of producer goods and services, however, remained controlled, including electricity and fuels, precious metals and stones, and freight tariffs.

The prices of energy products had long been very low by international standards. ^1/ In January 1992, they were increased by about five times: the price for crude oil was raised to rub 350 per ton, the price of gas to rub 260 per 1,000 m^3/, and the price of coal to rub 140 per ton (all exclusive of VAT). In the months following the January price jump, however, oil, gas and coal producers were allowed to sell an increasing proportion of their output at free prices, amounting to 40 percent in early May in the oil and gas sectors (and equal to the output exceeding state orders in the coal sector). A May 18 federal government decree increased the administered price of oil and gas by 4 to 6 times, but at the same time abolished the right to sell a fraction of production at a price exceeding the administrative ceiling (thus, the average effective price of oil and gas rose significantly less than the administered price). ^2/ The maximum price of oil was raised to rub 1,800-2,200 per ton (excluding VAT), with a steeply progressive tax applied on increments above rub 1,800. Similarly, the price of gas was raised to rub 1,100-1,600 per 1,000 m³ (excluding VAT), with an analogous provision. Furthermore, in the second half of May, freight prices were liberalized.

b. Data

The producer price index (PPI) published by Goskomstat relies on fixed 1989 weights. ^3/ In contrast to what its denomination (optoviy) may suggest, it measures prices at the factory gate rather than in wholesale trade. Hence, it excludes imported goods. The proliferation of barter deals, side payments and discounts for cash, as well as reportedly widespread price discrimination, ^1/ suggests that the PPI data should be interpreted with caution.

c. Developments

Industrial producer prices jumped, on average, by 5 times in January 1992, after having more than tripled within 1991, with large inter-branch variations (Table 7). Subsequently, they continued to grow rapidly, outpacing retail prices by a large margin in February and March (see Table 2 and Chart 1). Again, compared to the Eastern European experience, both the initial burst and the subsequent inflation were considerably larger in the Russian case (see Table 8 and Chart 5).

Since the industrial PPI had already risen much faster than consumer prices in 1991, the cumulative wedge between these two indicators had widened tremendously by the end of the first semester of 1992: taking December 1990 as a 100 base, the urban CPI had risen to 2,113 while the industrial PPI had grown to 7,106 i.e., 3.4 times more. ^2/ This cannot be explained by the lag usually observed between changes in producer and changes in consumer prices, since the divergence has increased almost monotonically. For the same reason, it cannot be accounted for by one-time factors such as the fact that in January 1992, shops were obliged to sell existing inventories at no more than two times (for food products) or three times (for other products) their December 1991 prices. More relevant could be explanations based on the definition of the PPI on the differential coverage of the indices under consideration, on the role of price controls and subsidies, and on the financial counterpart of higher producer prices.

A first potential reason is that the PPI may reflect listed prices rather than prices actually charged. In particular, the PPI may not reflect the significant discounts which were reportedly available for cash payment. Another factor that may help account for the divergence is the large amount of subsidies pledged and paid out following some of the producer price increases, in particular the ones for energy carriers. For example, the mid-May increase in oil and gas prices was reportedly followed by the announcement that additional subsidies for passenger transportation and for fuel used by the population would be granted. ^3/ However, sizable producer subsidies have also been announced for some sectors, in particular for agricultural producers. ^1/ Hence, the net impact of subsidization on the wedge between producer and retail prices is ambiguous.

Lastly, the exponential growth of interenterprise arrears during the first semester of 1992--from rub 34 billion on January 1 to rub 666 billion on April 1 and rub 2,919 billion on July 1--suggests that producer prices are contract prices but not necessarily settlement prices. It bears noting that in Romania, where interenterprise arrears also exploded, industrial producer prices rose much faster than consumer prices as well; while in Bulgaria and Czechoslovakia, where interenterprise arrears remained much smaller, no such discrepancy arose. ^2/ The “hardness” of the budget constraints faced by enterprises was questionable; in contrast, households were liquidity-constrained (literally, by the cash shortage, as well as by the absence of consumer credit). Thus the growth of prices of consumption goods was probably relatively more limited by demand than that of the intermediate and capital goods sold to other enterprises. The exclusion of the latter goods from the CPI may account for its much smaller increase in 1992.

Supporting this view is the fact that prices in the “food industry” component of the PPI rose by much less than those of the industrial goods in the PPI in the first six months of 1992. Indeed, the increase of 995 percent was not much greater than the 820 percent increase of the “food, beverages and tobacco” component of the CPI. ^3/

As regards energy prices, these remained far below world prices in U.S. dollars, even after their further increase in mid-May 1992. For example, the average producer price of oil in June 1992 was still only about rub 1,520 per ton. Evaluated at the average exchange rate during that month, ^4/ this would represent less than one-tenth of a prevailing world price on the order of US$20 per barrel. Even taking into account the possible undervaluation of the ruble in purchasing power parity terms, the relative price of energy remained very low in Russia.

a. Data

Two main types of evidence are available from Goskomstat for the study of income developments: average wages and the so-called “money incomes of the population,” a measure of household income. The average wage data have their origin in the total “wage fund” figures reported by enterprises to Goskomstat. This formerly annual series became available quarterly, and then monthly, in 1991. In addition to an overall national average, average wages in various sub-categories are compiled.

The household incomes data represent a broader concept, including for example pension payments, stipends, family allowances and interest income. These data are available on an annual basis as part of the Goskomstat table, “Money Incomes and Expenditures of the Population” ^1/ which also permits calculation of the rate of saving of households. Household income at the annual frequency is computed as the sum of direct estimates of the various types of incomes. At the monthly frequency, however, the published money income figures are estimated as the sum of estimates of household saving and expenditure. While the monthly incomes data are of considerable interest to the analysis of events in 1992, their reliability is much less certain, as discussed below. It should also be noted that in recent years the monthly data have at times been strongly influenced by special lump-sum “compensation” payments on households’ deposits. ^2/

The Goskomstat wage and household incomes data differ significantly in the way they were influenced by the so-called cash shortage. Numerous reports appeared in 1992 indicating that many workers were receiving less income than promised by their employer. Almost all Russian workers (excepting the military) are accustomed to being paid in cash, twice a month. However, the cash shortage meant that some enterprises were unable to obtain from the banking system the currency required to make all wage payments. This reportedly led, in some cases, to the relatively new phenomenon of enterprises making transfers to workers’ savings bank accounts as a substitute means of payment. A worker receiving such a transfer, however, may also not have been able to convert it into cash. ^{3/ 4/ 5/} Of course, the failure of some enterprises to make promised wage payments may have been attributable not to the cash shortage but rather to their own financial difficulties.

Significantly, Goskomstat wage data reflect wages earned, rather than wage payments actually received, during a given period. As discussed below, the difference between these concepts was significant during the first half of 1992. In contrast, the money incomes data are in principle a measure of payments actually received.

In analyzing income data, the main interest is the rate of growth relative to that of prices. For example, it will be useful to examine the ratio of average wages to a price level index. This statistical or measured real wage is to be distinguished, especially prior to 1992, from the real wage concept used in studying market economies. The latter construct can be considered an indicator of consumer welfare, if wages and incomes in the form of money are freely convertible into goods at market-clearing prices. Such was not the case in the shortage economy which prevailed in Russia before January 1992. ^1/ The statistical real wage measured prior to that date is of interest more as an indicator of macroeconomic disequilibrium than as an indicator of consumer welfare.

In constructing a real wage series with a base year of 1987, Goskomstat’s retail price index (RPI) is used through 1991; beginning in 1992, the urban CPI is used. Although the RPI is imperfect (as discussed in section II), the limited flexibility of the price level during most of the period prior to 1992 implies that variation due to error in measuring the price level is likely to be small in relation to the magnitude of wage inflation. Thus the main findings of this section are probably robust to the price level mismeasurement occurring before the 1992 price liberalization.

b. Developments

With the above considerations and data limitations in mind, one can analyze recent developments in incomes, starting with wages. Growth of average wages began to significantly outpace that of retail prices in 1988, resulting in steady growth of the statistical real wage (Chart 7). By 1990, average real wages stood 27 percent above their 1987 levels. The price reforms of early 1991 abruptly canceled this build-up, lowering the national average and industrial real wages approximately to their 1987 levels in the second quarter. However, while prices at the retail level were quite stable for some time after the April 1991 price reform, wage inflation continued: the national average wage rose 30 percent in the third quarter of 1991, reversing the recent real adjustment. In the final quarter, retail price inflation accelerated to 16 percent, but was again outpaced by a 48 percent surge in the average wage (57 percent in industry). By December 1991, statistical real wages were nearly twice their 1987 levels. ^1/

Nominal wage growth accelerated further in early 1992, with the national average wage rising 31 percent in January (50 percent in industry). ^2/ The statistical real wage, however, fell precipitously with the January price burst (by 62 percent in industry, and 67 percent overall). As Chart 7 makes clear, a large part of this decline represented a reversal of increases experienced only in late 1991, and only in a context of widespread shortages and an unsustainable run-down of retail inventories. Nevertheless, statistical real wages dropped below 1987 levels. Furthermore, in contrast to the experience after the (largely administered) price increases of early 1991, the January 1992 decline in the statistical real wage was not quickly reversed. The real wage did increase, however, after the January price burst, with overall real wages growing at an average monthly rate of 10 percent from January through June. By June 1992, the relationship between the national average wage and consumer prices was approximately the same as in 1987 (in industry, the measured real wage was almost 15 percent above its 1987 level).

This evidence appears to contradict the numerous reports to the effect that the standard of living of most Russian citizens had fallen to levels not seen in recent memory; certainly the strong growth of the real wage within the first half of 1992 has seldom been recognized. In this connection, it should be recalled that actual wage payments were less than indicated by the wage data analyzed here, which do not reflect wage arrears. Goskomstat has reported various measures of such arrears, beginning with April 1, 1992. ^3/ Although this data is not as complete as one would desire, an illustrative calculation suggests that in the second quarter of 1992, wage payments in industry may have been only about 90 percent of the wage bill implied by the average industrial wage figures.

While it is difficult to determine the appropriate level of the real wage in Russia, a number of observations are possible. There is some disagreement in the literature as to whether substantial macroeconomic disequilibrium (in the form of a monetary overhang and excess real wages) had been a long-standing phenomenon or whether it had developed only in the late 1980s. In the latter case, the 1987 base year considered here seems a reasonable choice, given the strong divergence of wage and price growth which began in 1988. Supposing that in 1987 the real wage per unit of aggregate output was approximately appropriate, subsequent real wage developments may be considered in conjunction with developments in both aggregate output and employment. Clearly, the rise of the measured real wage from 1987 through December 1991 was far in excess of any conceivable improvement in aggregate labor productivity. In fact, rough calculations of productivity suggest a substantial cumulative decline (reflecting mostly the abrupt fall in output which began in 1990-1991). The drop in the real wage which occurred in January 1992 would appear to have more than offset this decline. However, the recovery of the real wage, to beyond its 1987 level by June 1992, together with the continued decline in 1992 of all indicators of aggregate output, suggests that by mid-1992 the real wage per unit of output was again significantly above its 1987 level (which itself may have been somewhat excessive). ^1/

Chart 8 depicts real wage developments, since 1987, in four Eastern European countries as well as Russia. ^2/ It is not meant to imply that real wages in 1987 represented equilibrium values in any or all of these economies, but simply to allow a basic comparison of real wage developments during the last five years. For example, in the cases of Bulgaria and Romania, the increase of real wages prior to general liberalization of prices was much smaller than in Russia; in Czechoslovakia, almost no divergence of wages and prices had occurred. In each of these three cases, however, real wages fell steeply in early 1991.

The Russian experience is closer to that of Poland, particularly with respect to the much larger growth of the statistical real wage prior to general price liberalization. In the Polish case, however, the decline from the peak real wage value was less abrupt, as it began with the partial liberalization of prices in August 1989. At its lowest point just after the comprehensive January 1990 liberalization, the Polish real wage was about three-quarters of its 1987 level. The corresponding low point for real wages in Russia was fairly similar, at about 75 percent in industry and 65 percent overall. In both Russia and Poland, a significant recovery of real wages occurred within the six months following general price liberalization.

Another way to compare wage developments across countries is to compute wages in terms of U.S. dollars. By that standard, the level reached six months after comprehensive price liberalization was relatively modest in Russia, on the order of U.S. $40 per month in June 1992 for the overall average wage, as compared to U.S. $180 in Poland at the same stage.

Incomes policy was considered an important component of stabilization programs of the Eastern European countries depicted in Chart 8. ^1/ Because the managers of state enterprises might otherwise lack incentives or the ability to restrain wage increases, such a policy was thought particularly important. In Poland, for example, wage bill ceilings were set, ^2/ and payments in excess were subject to heavy penalties.

The Russian authorities opted for a much weaker policy, whereby the wage bill in excess of four times the minimum wage (times the number of employees) could not be deducted for purposes of calculating an enterprise’s taxable profits. The effectiveness of such a policy depends both on the profit tax rate and the setting of the official minimum wage. ^3/ As discussed below, the setting of the minimum wage was quite restrictive in the first semester of 1992; however, the prevailing profit tax rate of 32 percent would not seem prohibitive. Another factor which may have weakened the effectiveness of the policy was that it was not binding on those enterprises which would have earned negative profits even if they had paid an average wage equal to only four times the minimum wage.

Indeed, Russia’s wage policy did not induce enterprises to keep most wages below four times the minimum wage: by April 1992, the average wage in the so-called material sphere had reached 10 times the minimum wage. ^4/ The policy might have raised revenue for the government, but it evidently did not provide a nominal anchor. Monthly wage inflation ran at an average of 35 percent through the first quarter (42 percent in industry). A deceleration occurred in April 1992; however, wage inflation in May and June was 20 percent and 38 percent, respectively.

It is also of interest to consider, in addition to average wages, a broader measure of household incomes (Table 10). Growth rates of money incomes and average wages had been similar during 1988-1990. ^5/ Not surprisingly, given the large share of wages in money incomes, growth of such incomes also accelerated in 1991, with the largest part of the increase occurring in the final quarter. However, money incomes rose even faster than wages in 1991. ^1/ In 1992, strong growth of nominal household incomes continued, but the 1991 pattern was reversed: growth of money incomes fell significantly behind that of the average wage, as an abrupt divergence occurred in January 1992 that was not later offset. With the January price burst, measured real money incomes plummeted to only 51 percent of their 1987 level.

The shock to household income thus appears to have been considerably greater than indicated by the average wage data. Wage arrears are one factor behind this divergence, since Goskomstat money incomes data are meant to reflect only actual payments. Furthermore, anecdotal evidence suggests that payments arrears may have been proportionately more significant in the case of non-wage income. With the resolution of payments arrears, one would expect a narrowing of the divergence of the wage and household incomes series.

It also seems likely that the divergence in 1992 of the household incomes and wage series results from measurement problems imparting a substantial downward bias on the monthly money incomes series. The recent (i.e., monthly) incomes data are not a direct measure of income but rather are derived by Goskomstat as the sum of estimates of household saving and household expenditure. The latter have been based on reports of state retailing organizations, and thus exclude the likely increasing share of output distributed through unofficial channels, particularly with the decree liberalizing trading in January 1992. Furthermore, the reporting of sales by the retailing organizations is believed to have broken down significantly in 1992. ^2/ The resulting underestimation of household expenditures implies a corresponding underestimation of household incomes.

In summary, although it could not be expected that either the wage or money incomes series would be free from error, the likelihood that the latter has recently become significantly downward biased (at least at the monthly frequency) suggests that the wage series may be the more reliable (albeit partial) indicator of Russian incomes. However, it must be kept in mind that the wage data has tended to overstate the current income of wage earners because it does not reflect the significant wage arrears that developed in the first half of 1992.