1. Evolution of the exchange and trade system ^2/

Liberalization of the exchange and trade system started in the late 1980s. The state monopoly in foreign exchange was relaxed somewhat, as enterprises were allowed to retain foreign exchange that could be used, within certain limits, for imports. External trade started to be decentralized, with enterprises being permitted to trade directly abroad instead of via the foreign trade organizations. The external borrowing and foreign exchange monopolies of the Vneshekonombank (VEB) were terminated. Licensed commercial banks were permitted to deal in foreign exchange and limited foreign exchange auctions were initiated (see below). These piecemeal liberalization measures were implemented in the context of continued price controls and growing financial instability. They contributed to boost effective demand for imports but did not stimulate exports, and therefore bore some responsibility for the rapid increase in external debt. ^3/

Following the Supreme Soviet’s approval of presidential guidelines calling for the introduction of currency convertibility, a commercial exchange rate replaced the official exchange rate for most transactions in November 1990 (the official exchange rate was kept only for the valuation of external claims and selected statistical purposes). The commercial rate was fixed in terms of a basket of five currencies and set at a level three times as depreciated as the official exchange rate. At the same time, some export quotas were increased, import taxes were reduced, and export taxes introduced for the main raw materials. There was also a tourist exchange rate which originally, in November 1989, had been set at a level 10 times more depreciated than the official rate. Finally, a parallel foreign exchange market had long existed and was becoming increasingly tolerated as the previous control system broke down and the incentives widened for bypassing official channels. The spreads between the fixed and more market-determined rates widened during 1991, as the increase in domestic liquidity resulted in a depreciation of the latter.

In January 1992 the exchange and trade system was partially liberalized. A substantial depreciation of the rate for commercial operations helped reduce exchange rate distortions. However, a multiplicity of exchange rates remained in place. For the 40 percent surrender requirement that was introduced for raw materials, a “special commercial exchange rate” of Rub 55 per U.S. dollar was applied. For the 10 percent surrender requirement applying to all exports, a so-called quasi-market rate established by the Central Bank of Russia (CBR) was used, which was initially set at Rub 110 per U.S. dollar, but was periodically adjusted.

Alongside these controlled rates, several types of market rates continued to co-exist, including the interbank auction rate quoted on the Moscow Interbank Currency Exchange (MICEX), the rates offered by banks to individuals and the rates agreed in informal street trades. For government operations, several exchange rates were in effect. A rate of Rub 55 per U.S. dollar was applied for debt-service payments. A special rate of Rub 5.4 per U.S. dollar was used for so-called centralized import operations and services payments financed from the Hard Currency Reserve Fund (e.g., contributions to international organizations and business trips abroad for officials). A special accounting rate of Rub 10 per U.S. dollar was created for tax settlement with citizens who had income in foreign exchange. Finally, the official exchange rate was preserved for the valuation of external claims.

A set of major reforms of the foreign exchange and trade system were implemented in July-August 1992, including the unification of the exchange system and the introduction of convertibility of the ruble for current transactions. However, some of the accompanying measures were not taken: export quotas continued to apply to most goods, import subsidies remained large, foreign exchange reserves were not consolidated within the CBR, and there were delays in introducing the foreign exchange regulations.

The unification of the ruble was accomplished through the abolition of the special commercial exchange rate on July 1, of the quasi-market rate on July 3, of most of the special budgetary exchange rates on July 1, and of the remaining one on August 15. ^4/

The unification of the exchange system was expected to result in a large depreciation of the effective exchange rate applying to centralized imports, which accounted for about two-thirds of total imports during the first half of 1992, as import subsidies were to be confined to grain, medicine and imports under the World Bank’s Rehabilitation Loan. The new regulations that took effect from August 15, however, had little effect on the domestic prices of centralized imports, as the multiple exchange rates were effectively replaced by explicit subsidies. The latter varied by commodity and averaged approximately two thirds of the foreign currency value of subsidized centralized imports.

On the export side, effective taxation was greatly reduced, as exporters ceased to be subject to the surrender requirement at an appreciated exchange rate. The repatriation requirement was maintained, with 50 percent of all export receipts having to be surrendered to the banking system at the interbank market rate, the banks in turn being obliged to sell 30 out of the 50 percentage points to the CBR. The Government subsequently introduced many exemptions to the surrender requirement, both for individual enterprises and for regions. It also appeared that the banks experienced difficulties in enforcing the surrender requirement.

The rules governing currency controls after unification were spelled out in a set of texts (in particular the September 1992 Supreme Soviet “Law of the Russian Federation on Foreign Exchange Controls and Foreign Exchange Controls Monitoring”) which regulated movements of foreign exchange and provided for administrative enforcement. However, in practice, all current currency transactions could be carried out without any restriction, including forward import pre-payments of up to 180 days and unlimited purchases or sales of foreign exchange for legal current international transactions.

One major change in the exchange and trade system after unification was the implementation in January 1993 of a centralized exports scheme whereby the Government purchased a share of the quotas of exportables (in particular oil and gas) from Russian producers at domestic prices and in rubles, and received the foreign exchange revenues from their sale abroad, at world prices. Another noteworthy change was the authorization granted in April 1993 to banks holding general licenses to freely import and export foreign currency bank notes, treasury notes, coins and securities, provided customs regulations were observed.

Next, a number of regulatory measures were taken around mid-1993 that tended to boost the ruble, including: the introduction, as per a May 28 CBR instruction, of limits on banks’ open foreign exchange positions, which apparently forced a number of banks to reduce their holdings of foreign exchange; the authorization granted to non-residents, starting July 15, 1993, to open ruble accounts in Russian banks and to sell hard currency on the MICEX; ^5/ and the reduction of import subsidy coefficients for centralized imports. Moreover, the scope of the interbank foreign exchange market was broadened by the June 28 CBR decision to henceforth allow exporters to surrender foreign exchange to commercial banks rather than to the CBR (see Section II). ^6/

2. Development of the foreign exchange market

Foreign exchange auctions were initiated by VEB in November 1989. Transactions volumes were modest, partly because of the restrictions imposed on the use of foreign exchange. In April 1991, these auctions were replaced by an interbank foreign exchange market. Sessions took place each Tuesday at the Gosbank with representatives from 25 banks which had become members of the MICEX. The latter had been established in March 1991 as a joint stock company owned by commercial banks, finance companies and the CBR. In late 1991, weekly foreign currency auctions were also held at the Russian Exchange Bank.

From August 1991, Gosbank allowed the registration with commercial banks of foreign exchange transactions between enterprises. Recorded transactions amounted to about half of the volume of interbank auctions but actual interenterprise transactions in foreign exchange were probably at least as large. The exchange rate established in the interenterprise market often diverged from the interbank market rate since enterprises frequently agreed on various forms of side payments (e.g., the delivery of goods at the low official price rather than at the higher market price).

During the first half of 1992, the interbank market for foreign exchange remained embryonic (Table A1, Chart 1), and a large share of transactions consisted of CBR intervention sales. The thinness of the market, where volumes traded amounted to less than 5 percent of hard currency export receipts, was partly the consequence of highly negative real interest rates on ruble deposits and of the persistence of soft budget constraints, which encouraged enterprises to accumulate sizable foreign exchange deposits from retained export earnings. ^7/ Regulations on the admission of new members to an expanded MICEX and on the modalities of currency transactions were adopted in February 1992. ^8/ In April 1992, the frequency of sessions on the interbank market was doubled.

View Full Size

VOLUMES TRADED ON THE MICEX

January 1992 - December 1993 (Monthly totals, in millions of US dollars)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: MICEX; and CBR.1/ Net sales of US dollars.

• Download Figure
• Download figure as PowerPoint slide

Following exchange rate unification, the U.S. dollar/ruble rates determined at MICEX auctions were used as the official foreign exchange rates of the CBR to be used by all banks for accounting purposes. The frequency of U.S. dollar/ruble auctions, which were being held on Tuesdays and Thursdays beginning in April 1992, was increased by the addition of Mondays and Fridays starting May 31, 1993, and shortly thereafter of Wednesdays (Table A2). Volumes traded rose substantially, from about 10 percent of exports to countries other than from the former Soviet Union in the second half of 1992 to about 40 percent in the second half of 1993.

While interbank auctions were still confined to Moscow and limited to U.S. dollars through mid-1992, other markets were subsequently established in St. Petersburg, Yekaterinburg, Novosibirsk, Vladivostok and Rostov (Table A3). ^9/ Auctioning of other currencies also began, even though U.S. dollar transactions by far continued to dominate. Foreign currency trading outside of the MICEX grew, with the regional exchanges accounting for close to one fifth of U.S. dollar market turnover by the last quarter of 1993, excluding direct interbank trading (Table A4). The rates quoted on the regional exchanges occasionally diverged from those recorded at the MICEX and broken cross-rates emerged from time to time, as the integration of the foreign exchange market remained stymied by the underdevelopment of the financial infrastructure (poor communication facilities, unreliable ruble payment system, lack of widespread forex dealing expertise). However, CBR intervention reportedly helped contain the dispersion in the rates quoted across exchanges. Moreover, efforts were made to facilitate arbitrage across exchanges: in late 1993, an Association of Russian Currency Exchanges was formed, aiming at the creation of a unified information and trading network; furthermore, the MICEX started to encourage bank participants in other auctions to establish a Moscow branch and to participate in MICEX auctions.

The prominent ten or so MICEX participants were also engaged in interbank direct dealing transactions using telephones or Reuters dealing terminals. These banks bought from and sold to customers and traded on their own accounts as well. In the summer of 1992, the bulk of these transactions consisted of customer rather than interbank trading, with banks merely acting as brokers. By the end of 1993, such transactions had grown to an estimated one-third of the MICEX auction turnover, about half consisting of interbank trading. However, the uncertainty of ruble value dates and the lack of complete mutual confidence in other banks—due to incomplete disclosure of bank information, which discouraged the establishment of credit lines—hampered a more rapid development of these foreign exchange transactions. A few banks had set up credit lines among themselves and were trading on firm (contractual) order basis, but all other trading was done on non-firm (pre-payment) order basis.

A credit auction house specializing in small-scale transactions—the Moscow Interbank Financial House (IFH), with 80 member banks—commenced operations in November 1992, supplying bid and offer U.S. dollar/ruble and DM/ruble rates on a daily electronic information system facilitating direct dealing among members. Trading was initially slow but by the fourth quarter of 1993 had risen to an average of about US$6 million per week.

Turning to derivative instruments, trading of ruble/U.S. dollar futures contracts began at the Moscow Commodity Exchange (MCE) in October 1992. By late 1993 trading was taking place three times a week. In November 1992, the Moscow Board of Trade (MBT) launched its own ruble/U.S. dollar futures contracts, holding daily sessions. Futures trading subsequently extended to deutsche mark contracts. Both markets, and especially the MBT, were extremely thin, owing largely to the importance of counterpart risk and also to the lack of technical expertise of market participants. The volumes recorded in December 1993 for each delivery date were typically on the order of half a dozen $100 contracts per session on the MBT; and a few hundred $1,000 contracts and a few thousand $10 contracts per session on the MCE.

Finally, a few banks offered forward contracts to customers wanting to hedge their import payments. ^10/

Intervention by the CBR played an important role in the determination of the exchange rate (see Section III below). Until the recent establishment of its own trading room where direct interbank operations became feasible, the CBR placed its orders through the MICEX. Limited intervention also reportedly took place on the St. Petersburg exchange. The sole intervention currency was the U.S. dollar. Following exchange rate unification, the share of CBR intervention in total MICEX turnover remained large, averaging one third if measured by monthly net totals. ^11/

The CBR was not, however, the only official institution involved in exchange market operations. The Ministry of Finance (MOF) also held a substantial portion of the reserves of the monetary authorities in the form of foreign exchange and gold. It had its own sources of foreign exchange revenues, including the aforementioned centralized export scheme. Some transactions by the MOF on the foreign exchange market were reported in November 1993, but no reliable information has been made available on such operations. The MOF reportedly also tried to organize a consortium of commercial banks in December 1993 to sell centralized foreign exchange resources on the MICEX.

1. Unification and target zone schemes

As exchange rate unification was being considered in the first half of 1992, some policymakers were proposing that unification be accompanied by exchange rate anchoring. For example, then Deputy Prime Minister Gaidar (1992) argued that “it is important to “anchor” the price level by pursuing tight monetary and fiscal policies, and by stabilizing the value of the ruble against international currencies. (…) We want to stabilize the ruble at a realistic rate, (…) consistent with average monthly wages of about US$ 100 later this year. (…) We have asked the Group of Seven leading industrial countries to provide us with a stabilization fund similar to that provided for Poland”. A similar line was taken in a Policy Memorandum signed by Gaidar and then CBR Chairman Matiukhin (1992) and by a number of foreign observers, such as Fischer (1992). In this context, a specific target zone scheme involving a ± 7.5 percent band around a central parity of about Rub 60 to 80 per U.S. dollar was envisaged by some. ^13/ This would have entailed a very substantial further real appreciation and, by July 1992, a monthly U.S. dollar wage of the order of US$ 70-90. ^14/

The interbank exchange rate did appreciate substantially, from Rub 160 per U.S. dollar at the end of March to 113 at the end of May, helped by the increase in the CBR refinance rate from 20 to 50 percent in April and from 50 to 80 percent in May, ^15/ as well as by substantial CBR intervention on the MICEX (Tables A5 and A1). In the second half of June, however, the rate started to depreciate, notwithstanding continuing CBR intervention, to Rub 144 per U.S. dollar by end-month.

As it turned out, the exchange rate was unified but not pegged and, after a brief spell of appreciation in early July (to Rub 130 per U.S. dollar by mid-month), it started to depreciate steadily. ^16/

2. Exchange rate floating in the first year

From mid-1992 to mid-1993, the exchange rate policy followed by the authorities could be described as a managed float (as acknowledged, for example, in the CBR’s annual report for 1992). Most of the time, the CBR modulated its intervention so as to maintain a relatively steady rate of nominal depreciation, averaging some 16 percent per month. ^17/ Within the period, however, and more or less in parallel with the evolution of the real interest rate differential, the real exchange rate weakened substantially, through October 1992, recovered between November 1992 and January 1993, and subsequently remained in the neighborhood of its mid-1992 level (Chart 2).

View Full Size

INTEREST RATES, RESERVES, MONEY, AND THE REAL EXCHANGE RATE

July 1992 - December 1993 (Monthly Observations)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: CBR; MICEX; Goskomstat; US Bureau of Labor Statistics; and authors’ calculations.1/ Based on CBR refinance rate and LIBOR, and contemporaneous consumer price inflation.2/ Based on MICEX exchange rats and consumer prices in Russia and in the United States; July 1992 = 100.3/ At the beginning of the month, in millions of US dollars. Excludes gold and Fund proceeds.4/ Real ruble M2 is logged one month, July 1992 = 100.

• Download Figure
• Download figure as PowerPoint slide

The evolution of the nominal exchange rate was broadly in line with expectations as recorded in surveys of market participants or as reflected in the futures markets. A large scale poll conducted in July-August 1992 suggested that the vast majority of operators anticipated substantial nominal depreciation in the near future. ^18/ Evidence from the futures markets, available from the fall of 1992 onwards, also points to the fact that by and large the trend rate of depreciation was anticipated (Table A6 and Chart 3, first page). In that regard, one summary measure is the expected rate of depreciation at a horizon of one month associated with the quotes on the foreign exchange futures contracts (Table A7), which averaged 15 percent (on the MCE) and 25 percent (on the MBT) during the first half of 1993. ^19/ Furthermore, until April 1993 at least, the monetary authorities did not send any signals to counteract bearish expectations. ^20/

View Full Size

SPOT AND FUTURE EXCHANGE RATES

November 1992 - June 1993 (Weekly averages, in Rubles per US dollar)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: MICEX; Commersant; MBT; and MCE.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

SPOT AND FUTURE EXCHANGE RATES

July - December 1993 (Weekly averages, in Rubles per US dollar)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: MICEX; Commersant; MBT; and MCE.

• Download Figure
• Download figure as PowerPoint slide

3. Exchange rate smoothing in the second half of 1993

During the second half of 1993, the nominal exchange rate depreciated by less than 20 percent (end-December over end-June) while monthly domestic consumer price inflation averaged 20 percent (Table A8, Chart 4). At the same time, real ruble money contracted sharply, while gross CBR foreign exchange reserves accumulation surged in the third quarter and declined somewhat in the fourth (Chart 2).

View Full Size

CONSUMER PRICE INFLATION

July 1992 - December 1993 (Rates per month, in percent)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: Goskomstat; and Center of Economic Analysis.

• Download Figure
• Download figure as PowerPoint slide

More specifically, the nominal rate appreciated by some 12 percent between mid-June and end-July, and then stabilized at a plateau of about Rub 1,000 per U.S. dollar through mid-September (Chart 5). The political crisis associated with the dissolution of the Supreme Soviet in late September caused the rate to depreciate sharply to a then historical low of Rub 1,299. It subsequently recovered somewhat and then remained within a ± 5 percent band around a notional benchmark of Rub 1,200 per U.S. dollar through the end of the year. To a large extent, the relative stability of the nominal rate over this period was the result of a shift in policies that took place around mid-1993.

View Full Size

SPOT EXCHANGE RATES

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: MICEX; Goskomstat; Commersant; US Bureau of Labor Statistics; and authors’ calculations.1/ Inferred by the authors from the policy statements cited in Section III.2/ Based on MICEX exchange rate, weekly CPI In Russia, and monthly CPI in the United States.

• Download Figure
• Download figure as PowerPoint slide

In line with the joint CBR-Ministry of Finance agreement to implement the measures spelled out in the Systemic Transformation Facility program supported by the IMF, the CBR refinance rate was raised in steps from 100 percent end-May to 170 percent by July 15 (Chart 6). It was further raised in September, to 180 percent, and in October, to 210 percent, in accordance with a September 22, 1993 government ordinance entitled “On Urgent Measures to Ensure Financial Stability” stipulating that the CBR, in coordination with the Ministry of Finance, adopt measures “to protect the national currency of the Russian Federation and prevent sharp fluctuations of its rate”.

View Full Size

INTEREST RATES

July 1992 - December 1993 (Weekly observations)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: CBR; IMF; Goskomstat; and authors’ calculations.1/ Using contemporaneous weekly consumer price inflation as a deflator.

• Download Figure
• Download figure as PowerPoint slide

Around mid-1993, the Ministry of Finance started a public campaign in support of the stabilization of the nominal exchange rate, in the form of a series of press releases and of a Rub 50,000 public bet by the Minister of Finance himself that the exchange rate would remain stable through September 1. ^21/ On June 24, 1993, the Ministry of Finance issued a statement declaring that “the financial condition of the country will be improving in the next three months, and the inflation index [rate] will be down to less than 15 percent per month. (…) The exchange rate of the ruble against the dollar has entered a period of relative stability and is likely to stay at the level of about Rub 1,100 per dollar. Large amounts of foreign exchange are expected to appear on the market by the end of the year, which will make it possible for the exchange rate to come closer to the purchasing power parity.” ^22/

On August 10, 1993, another press release from the Ministry of Finance indicated that “The Russian ruble exchange rate against the dollar has been relatively stable for two months. The growing popularity of the ruble compelled the Central Bank to switch from its previous standard of Rub 1,100 to Rub 1,000 per dollar (plus/minus 5 percent). The gap between the market rate and the ruble’s buying capacity has shrunk notably. (…) The country’s currency reserves have actually increased by several billion dollars in the past two months. (…) The coordinated tactics between the Central Bank and the Finance Ministry in recent months reveal their desire to facilitate the stabilization of the ruble, without strengthening it too abruptly and excessively to damage exports. (…) All the above factors indicate that the ruble will remain quite stable not only until the beginning of September, but until the end of this year. This trend, which became evident in August, is certain to chip away at the rate of inflation.” That this was the joint policy of the Ministry of Finance and the CBR was suggested by central bank officials. ^23/

After the September crisis, the Minister of Finance reiterated on several occasions that the ruble was likely to remain stable until the end of the year, at around Rub 1,200 per U.S. dollar, ^24/ and the CBR continued to intervene in both directions, but on a smaller scale than during the summer.

With monthly inflation running well over 20 percent in the third quarter of 1993, market participants retained serious doubts about the sustainability, beyond the very near term, of a virtually stable nominal exchange rate, as indicated by the quotes registered on the futures markets (Table A6 and Chart 3, second page). Using the MCE US$1,000 contract as a benchmark, the expected rate of depreciation at a 30-day horizon on average exceeded 10 percent, and the expected rate of depreciation at a 60-day horizon on average exceeded 20 percent (Table A7). ^25/

The exchange rate policy pursued in the second half of 1993 by the monetary authorities could be characterized as one of large-scale smoothing, especially during the summer months. It could even arguably be described as a flexible target zone policy. As regards policy intentions, this is very clear from the above excerpts, which advertise an adjustable central rate and the width of the associated band. As regards policy implementation, this strategy led the CBR to lean heavily against the wind in June-July, when it purchased on net over US$ 1 billion on the MICEX, and in August-September, when it sold on net close to US$ 1.5 billion on the MICEX (Chart 1).

The combination of a very high domestic rate of inflation and of a very limited nominal exchange rate depreciation over a fairly long period implied a very sharp real appreciation (Chart 5). Symptoms of undervaluation were still abundant in mid-1993, typified by enterprise directors’ complaints that the gap between the purchasing power of the ruble and the U.S. dollar “allowed any American student to buy a Russian plant with his stipend”.

Half a year later, symptoms of overvaluation had become apparent, with a number of domestic prices for tradeable goods far exceeding world prices and exporting enterprises imploring a large depreciation of the ruble. ^26/ Also, by end-year, dollar wages in Russia had converged to the level registered on average in the Baltic states. ^27/ Comparing to Central and Eastern European countries, they ended the year at about half of the level observed in Poland and the Czech Republic at the same stage of the transition, but well above the level recorded in Romania, and at the same level as in Bulgaria. Notwithstanding this evidence, it remains very difficult to assess how far the real exchange rate stood at any point during this period from an “equilibrium” rate, as different approaches produce very different answers (Table A9, Chart 7).

View Full Size

ALTERNATIVE REAL EXCHANGE RATE MEASURES

July 1992 - December 1993 (Monthly observations)

Citation: IMF Working Papers 1994, 066; 10.5089/9781451848755.001.A001

Sources: Goskomstat; MICEX; World Bank; Center of Economic Analysts; PlanEcon; IMEMO; McDonald’s Moscow; and authors’ calculations.1/ Based on MICEX exchange rate and on the average, economy-wide wage.2/ In US dollars per ruble.

• Download Figure
• Download figure as PowerPoint slide

1. A simple model of exchange rate determination

In the tradition of Dornbusch (1976) and Frankel (1979), it is assumed that there exists a long run equilibrium exchange rate, but that in the short run the spot exchange rate may differ from its long run equilibrium value, reflecting sluggish adjustment in goods markets. In such an environment, the expected rate of depreciation can be written as:

where E_t denotes the expectation conditional on information available at time t, s is the logarithm of the spot bilateral exchange rate in rubles per U.S. dollar, and S its long run equilibrium value. Parameter α₁ measures the speed at which the spot exchange rate converges to its long run equilibrium rate, and is positive. Parameter β₁ reflects the extent to which changes in the expected long run equilibrium rate are reflected in changes in the spot rate. In the absence of “money illusion”, β₁=1.

Residents hold interest bearing assets denominated in rubles and in U.S. dollars. Let

• DA ≡ logarithm of the stock of interest bearing assets denominated in rubles,

• FA ≡ logarithm of the stock of interest bearing assets denominated in foreign exchange,

• i ≡ log(1+r), where r is the nominal interest rate on ruble denominated assets,

• i^* = log(1+r^*), where r^* is the nominal interest rate on U.S. dollar denominated assets.

Portfolio allocation is a function of relative yields:

and

with ω₁ and ω₂ the budget shares of the assets, i.e., ω₁ + ω₂ = 1, and Ω a scale variable defined as Ω_t ≡ ω₁ DA_t + ω₂ (s_t + FA_t).

Relative yields determine the portfolio shares of the ruble assets and of the U.S. dollar assets converted in rubles at the spot exchange rate. Since residents will hold an additional amount of financial assets denominated in rubles when they expect the currency to appreciate and/or when the domestic interest rate increases relative to the foreign interest rate, α₂>0. In the case of risk neutral economic agents, α₂ → +∞.

Dividing both sides of equations (2. a) and (2.b) by ω₁ and ω₂, respectively, subtracting equation (2.b) from equation (2.a), and rearranging terms, yields:

where α₃ ≡ (ω₁ ω₂)/α₂ ≥ 0.

Equation (3) describes equilibrium in the foreign exchange market. If economic agents are risk neutral (α₂ → +∞), or if there are no outside assets, the right hand side of equation (3) is equal to zero, and uncovered interest rate parity holds. In other words, the left hand side of (3) measures the risk premium.

Inserting equation (1) into equation (3) yields:

where

Equation (4) relates the spot exchange rate to the long run equilibrium exchange rate, the expected change in the long run equilibrium rate, the interest rate differential, and the relative stocks of interest bearing assets. The interpretation of the signs of the parameters is straightforward. Note that:

The latter constraint reflects the fact that the exchange rate clears the foreign exchange market through its impact on the expected return and the stock of assets denominated in foreign currency. Indeed, equation (1) shows how exchange rate expectations are driven by the divergence between the spot and the long run equilibrium rate, while the left hand side of equation (3) illustrates how the stock of foreign assets, measured in rubles, is adjusted to its desired level through changes in the spot exchange rate. Parameters α₄ and γ₄ are weights which reflect the relative importance of each of these two adjustment channels. In the polar case where economic agents are risk neutral (or in the absence of risk), α₃=0 and the parameters reduce to:

In other words, in the case of risk neutral economic agents the spot exchange rate restores foreign exchange market equilibrium only through its impact on the expected rate of return.

In order to estimate equation (4), the long run equilibrium exchange rate and the expected change in this long run equilibrium rate have to be modelled.

In principle, the concept of a long run equilibrium exchange rate involves a notion of current account equilibrium. Since the focus of this paper is on the impact of short run monetary factors on the spot exchange rate, it will be assumed that long run equilibrium conditions can be captured by using alternative measures of the purchasing power parity (PPP) rate. Three such measures are considered, based respectively on consumer goods prices (PPP1), on the price of a Big Mac sandwich (PPP2), and on the price of a basket of 19 basic food items (PPP3). ^28/ As shown in Chart 7, these measures differ significantly, especially in level terms.

The long run equilibrium exchange rate is measured by taking a weighted average of these alternative measures:

where the weights stand for the proportions of agents considering measures PPP1, PPP2 and PPP3, respectively, to be the relevant ones. Note that 0 ≤ α₅, β₅ ≤ 1.

Also needed to operationalize the model is a hypothesis on the expected change in the long run equilibrium rate. The expected long run nominal equilibrium exchange rate is equal to the expected long run real equilibrium exchange rate adjusted for the expected relative long run price levels. Here, it is assumed that no changes in the real equilibrium rate are anticipated. Hence, only the expected changes in the long run price levels are to be specified.

It is posited that agents base their expectation of the long run price level on recently observed price movements. Furthermore, in the case of Russia, one should take into account that expectations are formulated in a high inflation environment. High inflation is not sustainable in the long run: either it degenerates into hyperinflation or stabilization comes about. Market participants closely monitor inflation developments and are assumed to formulate their expectations on the basis of the following scheme:

where p and p^* respectively denote the logarithms of the domestic and foreign price levels, and MOV(.) Is an operator taking a lagged, ten-week moving average, anchored on the contemporaneous week. ^29/ A priori, the sign of parameter β₉ is undetermined. If β₉ > 0, market participants are expecting that inflation will degenerate into hyperinflation; if β₉ = 0, the recently observed inflation rate is expected to be sustainable in the long run; if β₉ < 0, stabilization is expected.

Inserting equations (6), (8) and (9) into equation (4), yields:

where α₄, α₅, β₄, β₅ ≥ 0, δ₄ ≤ 0, and the sign of γ₁₀ is indeterminate.

An equation is thus obtained relating the spot exchange rate to a long run equilibrium exchange rate concept, an interest rate differential, an inflation differential and the relative ruble and U.S. dollar asset stocks.

2. Estimation results

First the unrestricted version of equation (10) is estimated using weekly data. ^30/ The estimation period for the dependent variable ranges from mid-1992 to the end of 1993. ^31/ The first result is as follows (with t-values between brackets):

This result is also reported as version (1) in Table 1. The interest rate on ruble denominated assets is measured by the Moscow interbank interest rate (using the CBR refinance rate instead did not yield better results). ^32/ The estimated coefficients display the expected signs and high t-values, except for the coefficient associated with PPP1, which has a negative sign (but a low t-value).

Table 1.

Main estimation results

Estimated equation: $\begin{array}{c}{s}_t={\phi }_{10}+{\alpha }_4(\begin{array}{cc}{\alpha }_5& PPP{1}_t\end{array}+\begin{array}{cc}{\beta }_5& PPP{2}_t\end{array}+\begin{array}{cc}(1-{\alpha }_5-{\beta }_5)& PPP{3}_t\end{array})+{\delta }_4(i_t-{i^{*}}_t)\\ +{\beta }_4[MOV(p_t-p_{t-1})-({p^{*}}_t-{p^{*}}_{t-1})]+({\gamma }_{10}+\mu DUM)[MOV(p_t-p_{t-1}){]}^2+(1-{\alpha }_4)(D{A}_t-F{A}_t)\end{array}$

^1/The associated t-statistic is indicated in parentheses.

^2/Confidence level to reject the null hypothesis when tested against version (3) of this table.

^3/Confidence level to reject the null hypothesis when tested against version (1) of this table.

^4/Diagnostic statistics are for the transformed data.

^5/Confidence level to reject the null hypothesis when tested against version (5) of this table.

Table 1.

Main estimation results

Estimated equation: $\begin{array}{c}{s}_t={\phi }_{10}+{\alpha }_4(\begin{array}{cc}{\alpha }_5& PPP{1}_t\end{array}+\begin{array}{cc}{\beta }_5& PPP{2}_t\end{array}+\begin{array}{cc}(1-{\alpha }_5-{\beta }_5)& PPP{3}_t\end{array})+{\delta }_4(i_t-{i^{*}}_t)\\ +{\beta }_4[MOV(p_t-p_{t-1})-({p^{*}}_t-{p^{*}}_{t-1})]+({\gamma }_{10}+\mu DUM)[MOV(p_t-p_{t-1}){]}^2+(1-{\alpha }_4)(D{A}_t-F{A}_t)\end{array}$

Version	Type	φ10	α4	α5	β5	δ4	β4	γ10	μ
(1)	OLS	2.16 (4.48)	0.87 (6.93)	-0.12 (-0.61)	0.65 (6.62)	-54.00 (-6.33)	18.14 (2.53)	-206.41 (-2.31)	μ=0
(2)	OLS	2.31 (4.68)	0.84 (6.55)	-0.24 (-1.20)	0.68 (6.43)	-49.52 (-5.79)	1.84 (1.38)	γ10=0	μ=0
(3)	OLS	1.99 (4.64)	0.82 (7.35)	-0.19 (-1.07)	0.50 (5.56)	-41.88 (-5.23)	17.78 (2.80)	-272.07 (-3.38)	78.41 (4.58)
(4)	OLS	2.64 (14.97)	α4 = 1	-0.13 (-0.87)	0.46 (6.54)	-54.30 (-18.97)	18.38 (2.86)	-282.16 (-3.48)	75.68 (4.39)
(5)	CORC	1.48 (4.19)	0.73 (10.59)	0.31 (8.41)	-0.05 (-4.57)	-14.68 (-2.04)	8.20 (1.02)	-123.92 (-1.17)	73.25 (2.44)
(6)	CORC	1.32 (5.15)	0.73 (10.69)	0.29 (8.15)	-0.05 (-4.66)	-11.49 (-2.08)	β4 = -δ4	-164.35 (-2.15)	71.49 (2.39)
OLS = ordinary least squares, CORC = Cochrane-Orcutt procedure; t-values between brackets. The dummy DUM takes the value 1 for the period running through mid-1993, and 0 thereafter.
Version	R2		Durbin Watson		Log of likelihood		Confidence level		ρ 1/
(1)	0.98		0.75		79.80		1.000 2/
(2)	0.98		0.72		76.97		0.983 3/
(3)	0.99		0.74		90.01		not applicable
(4)	0.99		0.86		88.51		0.917 2/
(5) 4/	0.69		1.62		126.82		not applicable		0.96 (36.43)
(6) 4/	0.69		1.61		126.67		0.416 5/		0.97 (41.68)

^1/The associated t-statistic is indicated in parentheses.

^2/Confidence level to reject the null hypothesis when tested against version (3) of this table.

^3/Confidence level to reject the null hypothesis when tested against version (1) of this table.

^4/Diagnostic statistics are for the transformed data.

^5/Confidence level to reject the null hypothesis when tested against version (5) of this table.

View Table

Table 1.

Main estimation results

Estimated equation: $\begin{array}{c}{s}_t={\phi }_{10}+{\alpha }_4(\begin{array}{cc}{\alpha }_5& PPP{1}_t\end{array}+\begin{array}{cc}{\beta }_5& PPP{2}_t\end{array}+\begin{array}{cc}(1-{\alpha }_5-{\beta }_5)& PPP{3}_t\end{array})+{\delta }_4(i_t-{i^{*}}_t)\\ +{\beta }_4[MOV(p_t-p_{t-1})-({p^{*}}_t-{p^{*}}_{t-1})]+({\gamma }_{10}+\mu DUM)[MOV(p_t-p_{t-1}){]}^2+(1-{\alpha }_4)(D{A}_t-F{A}_t)\end{array}$

Version	Type	φ10	α4	α5	β5	δ4	β4	γ10	μ
(1)	OLS	2.16 (4.48)	0.87 (6.93)	-0.12 (-0.61)	0.65 (6.62)	-54.00 (-6.33)	18.14 (2.53)	-206.41 (-2.31)	μ=0
(2)	OLS	2.31 (4.68)	0.84 (6.55)	-0.24 (-1.20)	0.68 (6.43)	-49.52 (-5.79)	1.84 (1.38)	γ10=0	μ=0
(3)	OLS	1.99 (4.64)	0.82 (7.35)	-0.19 (-1.07)	0.50 (5.56)	-41.88 (-5.23)	17.78 (2.80)	-272.07 (-3.38)	78.41 (4.58)
(4)	OLS	2.64 (14.97)	α4 = 1	-0.13 (-0.87)	0.46 (6.54)	-54.30 (-18.97)	18.38 (2.86)	-282.16 (-3.48)	75.68 (4.39)
(5)	CORC	1.48 (4.19)	0.73 (10.59)	0.31 (8.41)	-0.05 (-4.57)	-14.68 (-2.04)	8.20 (1.02)	-123.92 (-1.17)	73.25 (2.44)
(6)	CORC	1.32 (5.15)	0.73 (10.69)	0.29 (8.15)	-0.05 (-4.66)	-11.49 (-2.08)	β4 = -δ4	-164.35 (-2.15)	71.49 (2.39)
OLS = ordinary least squares, CORC = Cochrane-Orcutt procedure; t-values between brackets. The dummy DUM takes the value 1 for the period running through mid-1993, and 0 thereafter.
Version	R2		Durbin Watson		Log of likelihood		Confidence level		ρ 1/
(1)	0.98		0.75		79.80		1.000 2/
(2)	0.98		0.72		76.97		0.983 3/
(3)	0.99		0.74		90.01		not applicable
(4)	0.99		0.86		88.51		0.917 2/
(5) 4/	0.69		1.62		126.82		not applicable		0.96 (36.43)
(6) 4/	0.69		1.61		126.67		0.416 5/		0.97 (41.68)

^1/The associated t-statistic is indicated in parentheses.

^2/Confidence level to reject the null hypothesis when tested against version (3) of this table.

^3/Confidence level to reject the null hypothesis when tested against version (1) of this table.

^4/Diagnostic statistics are for the transformed data.

^5/Confidence level to reject the null hypothesis when tested against version (5) of this table.

The coefficient associated with the interest rate differential is related to the speed at which the spot rate adjusts to its long run equilibrium value (α₁) and the degree of risk aversion (α₃). Equations (5.a) and (5.c) and the earlier reported point estimates imply for α₁ a value equal to 0.016. ^33/ In other words, after one week the spot exchange rate adjusts for 1.6 percent of a change in the long run equilibrium rate. The risk aversion coefficient comes out positive, at 0.003.

The negative sign of the coefficient associated with the quadratic inflation term suggests that the market expects stabilization. Nevertheless, it is worth examining the regression results if this term were to be left out. Version (2) shows that the value of the parameter associated with the inflation differential, β₄, drops sharply. However, at the same time, the diagnostic statistics indicate that the hypothesis γ₁₀=0 can be rejected with a fair degree of confidence. ^34/ Moreover, the description of policies and expectations in Section III implies that a distinct break occurred around mid-1993. A time dummy is used in version (3) to contrast the first year following exchange rate unification and the second half of 1993, the prior being that in the latter period, agents were more confident that stabilization would come about. This presumption is indeed strongly corroborated, as version (1) can be rejected when tested against version (3). Hence, the quadratic term and the associated time dummy are maintained in subsequent versions.

Still regarding agents’ expectations, it is worth assessing the significance of the risk premium. Version (4) reports the estimation results for the case where it is assumed that the relative stock of ruble and U.S. dollar assets has no impact on the spot rate, i.e., α₄=l. The null hypothesis that there is no risk premium can be rejected when tested against version (3), but not with a very high degree of confidence.

The low values taken by the Durbin-Watson statistic indicate high positive autocorrelation. Therefore, version (3) is reestimated using a Cochrane-Orcutt procedure. Version (5) shows that the qualitative conclusions reached above are not invalidated. However, some differences deserve mention. The implied value of α₁—the parameter measuring the speed of adjustment of the spot exchange rate to its long-run value—rises from 2 percent (per week) in version (3) to 5 percent (per week) in version (5). Also, the relative weights of the different PPP measures change. ^35/

The elasticities with respect to the interest rate and inflation differentials come somewhat closer together in version (5). Version (6) serves to test the null hypothesis H0 : δ₄ = -β₄ against the alternative hypothesis : H₁ : δ₄ ≠ -β₄. It appears that one cannot reject the null, meaning that “money illusion” does not prevail in the foreign exchange market. ^36/

In sum, it appears that the interest rate differential and the expected inflation differential had a significant impact on the evolution of the ruble-U.S. dollar exchange rate. Moreover, the evidence seemed to suggest that market participants were well aware of the risks associated with high inflation. The hypothesis that the relative supply of interest yielding assets had a significant impact on the spot exchange rate was partly corroborated.

A number of further tests were carried out to assess the influence of particular institutional or policy changes on the spot exchange rate. For the sake of conciseness, and because on the whole no clear-cut impacts could be identified, the results are not reported. Presumably, the weak or seemingly non-existent influence of such factors results from the fact that in many cases changes in the rules are accompanied by many exemptions and are only partly and belatedly enforced.

Potential simultaneity was also tested for by reestimating the equations using instrumental variables. Indeed, one cannot a priori rule out a simultaneity bias induced by causality running from the exchange rate to domestic inflation and interest rates. The results suggested that there is no major simultaneity problem.

Finally, the hypothesis that the CBR intervened to achieve a certain exchange rate was tested using weekly data on CBR intervention on the MICEX. Notwithstanding the policy intentions and actions described in Section III, the econometric analysis did not support the conjecture that the CBR had any intervention strategy other than smoothing short-run fluctuations. However, this may be due to the fact that interventions were not circumscribed to CBR sales or purchases on the MICEX, as well as to our incomplete information on the very short-run exchange rate rules guiding the authorities’ intervention, which reportedly changed fairly often or were not always fully clear.