Black-Markets for Currency, Hoarding Activity and Policy Reforms

(published in the Journal of International Economics vol. 42 3/4 May 1997 pp.349-370)

Linda S. Goldberg Il'dar Karimov

Federal Reserve Bank of New York IIASA, Austria
and NBER

Abstract

In emerging market economies, under-development of financial markets leads to goods

hoarding and foreign currency accumulation as forms of investment. In this paper, an asset

market model, supplemented by explicit treatment of smuggling and second-economy

activity, is used for studying the paths of black-market exchange rates, second-economy

prices, hoarding stocks, and privately held dollar balances following policy reforms. We

discuss conditions for overshooting and related dynamics of exchange rates and prices

following: official exchange-rate adjustments, price reforms, and altered risks of monetary

confiscation or currency reforms. The saddlepath trajectory of financial adjustment is

simulated using data from recent Russian experience.

Keywords: Black Markets, Exchange Rate, Hoarding, Overshooting, Emerging Markets,

Transition

JEL Classification Codes: F3, F31, F41,O17, P21

The opinions expressed in this paper are those of the authors and do not necessarily reflect those of the Federal
Reserve Bank of New York. Linda Goldberg is grateful for the technical support provided by the C.V. Starr
Center for Applied Economics at New York University and the financial support of the National Science
Foundation and the Social Science Research Council. Maury Obstfeld provided particularly useful comments on
the paper. Linda Goldberg also appreciates useful discussions with Pierre-Richard Agenor, Barry Ickes, and
Randi Ryterman. Dr. Il'dar Karimov is grateful for the technical support provided by the Economic Transition
and Integration Project at the International Institute for Applied Systems Analysis, Austria and the Japan
Foundation, and for the technical and financial assistance of the International Centre for Economic Research in
Turin, Italy.
1. Introduction
In emerging markets and some developing economies, restrictions still hamper free-market

transactions in some goods and financial assets. Financial markets remain under-developed and

few investment assets and outlets for savings are available to individuals. The existence of

pervasive shortages, rigidities, and the vestiges of central planning and controlled prices, lead

second-economy or black-market production structures to operate in tandem with secondary

markets for fully-convertible foreign currencies. In this context, characterized by a paucity of

financial instruments and uncertainty regarding future consumption-goods availability, one form

of investment and intertemporal transfer of wealth has been through purchases of storable

consumption goods.1

In the present paper, these features of the transition economy are shown to bear upon the

dynamic effects of various policy initiatives, such as official price reforms, official exchange rate

devaluations, and monetary reforms. With investment opportunities available to the consumer/

investor consisting of hoarded goods, accumulated hard currency, and domestic-currency savings,

we trace the dynamics of adjustment of the stocks and prices of domestic currency assets and

foreign exchange. An explicit treatment of overshooting is provided. These dynamics of exchange

rates and prices depend on events in both financial and goods markets. An important implication

is that demand for second-economy goods arises from both investment and consumption motives,
rather than purely from consumption motives. Thus, a broader range of reform initiatives have

significantly richer effects on exchange rates and prices than those predicted by more conventional

approaches. Moreover, we highlight a role for hoarding activity beyond that typically cited as

important in discussions of policy credibility and "big bang" approaches to reform during

transition [van Wijnbergen (1992)].

The key theoretical contribution of the present paper arises from the simultaneous

treatment of exchange rate and price dynamics, coupled with an explicit treatment of smuggling

1Weitzman (1991) also studies the importance of goods as a store of value.

1
activities. While the methodology and dynamic treatment of the problem per se are not novel, the

specific application is: its solution provides useful insights for transition and developing

economies. By explicitly studying the likelihood of overshooting of exchange rates and prices, we

provide insights into the observed pattern of relatively high volatility of nominal and real

exchange rates in emerging markets. Our results support criticisms of the use of black market

exchange rates as a guide to equilibrium fixed exchange rates in transition economies. Both real

and nominal black market exchange rates can overshoot in response to goods market events, in

addition to overshooting in response to actual or pending foreign exchange market reforms.

The theory that we present complements two other important approaches to modeling

exchange rate and price dynamics. First, we contribute to the theory of black-market exchange

rates. This prior literature was motivated mainly by the experiences of Latin American economies,

wherein black-market currency demands are treated as arising from "portfolio" motives: changes

in the risk, return and stocks of various financial assets drive the currency price.2 Recent

applications include Pinto's (1991) analysis of the fiscal ramifications of unification of black and

official market exchange rates, and Lane's (1992) study of Polish household behavior in money

markets during the 1980s.3

Our approach also is closely related to the literature on the dynamics of exchange rate

adjustment which stemmed from the seminal Dornbush exchange-rate overshooting framework.

Such models typically have two distinct markets: the market for financial assets and the market

2Calvo and Rodriguez (1977) provide an early model of currency substitution. Branson and Henderson (1985)
provide a comprehensive presentation of portfolio theory applied to the asset market approach to exchange rate
determination. In the context of black market exchange rate determination, the existing literature, as in de Macedo
(1982, 1987) and Dornbush et al. (1983), models dollar holdings by individuals arising as a means of diversifying a
"portfolio" of assets held to maximize expected returns on "invested" wealth and minimize the variance of these
returns.
3Lane (1992) considers portfolio demands for two assets: domestic currency and foreign exchange.

2
for domestic goods.4 As such, these models are useful for analyzing the dynamic effects of

changes in the financial and goods market structures when financial market structures are fully

developed and when goods prices are sticky.5 However, when there are underdeveloped financial

and goods markets the theory requires modification. Useful steps in this direction are taken by

Calvo and Frenkel (1991) in their analysis of the implications of financial market reforms, and by

Agenor and Flood (1992) and Goldberg (1995) in their analyses of exchange rate unification.

However, unlike the present paper, all of these papers maintain the assumption of distinct

dynamics in foreign exchange and goods markets.

Finally, our analysis provides insights into the meaning of "monetary overhang".

According to the traditional view for socialist and in emerging market economies, money

holdings are, in part, involuntary.6 Instead, we maintain that, in an environment with under

underdeveloped financial markets, money balances are held as part of an optimizing decision

by households. Therein, individuals take into account the variances and covariances among the

returns on alternative savings vehicles, in addition to considering the pure expected values of

these returns. In this treatment, the risk-adjusted returns on money may not be dominated by

those adjusted returns on alternative investment assets.7 The theoretically "optimal" demand

for money balances are distinct from those generated by current transactions motives and are

not related exclusively to set-asides for stochastic future consumption opportunities.

4See Obstfeld and Rogoff (1995) for an important reworking of the sticky price framework to incorporate
monopolistic competition and intertemporal issues.
4Frenkel and Rodriguez (1982) present a two-asset model which relaxes the assumption that goods prices are sticky.
However, in that paper prices still are determined in goods markets, rather than alongside investment vehicles in
assets markets. In contrast to our approach, goods are held purely for consumption purposes, and smuggling and
hoarding activities are not considered. This treatment of real goods as an investment asset, with the smuggling
linkages, further distinguishes our work from simple three-asset models of the portfolio approach to exchange rate
determination.
6For example, see Birman (1980) and Cottarelli and Blejer (1992).
7This type of treatment also is provided by Lane (1992).

3
 The organization of the paper is as follows. First, Section II develops the model of the

black-market for foreign exchange and hoarding of real goods. Section III analyzes the

adjustment paths of black-market exchange rates, second-economy goods prices, dollar

balances in private portfolios, and hoarding stocks, in response to a set of announced and

unanticipated policy initiatives. Section IV concludes.

II. The Model: Black Markets for Foreign Currencies and Goods Hoarding
Two aspects of the demand and supply for (black) foreign exchange and for second-

economy consumer goods are modeled. The first aspect is derived from "financial" or

investment motives for holding foreign currency and hoarding goods, and is subsumed under

the heading of "portfolio" or speculative currency-demands. The second aspect is related to

current consumption needs and is subsumed under the heading of transaction demands with

smuggling.

A. Portfolio Demands in the Black Market

The consumer/investor chooses an optimal division of his wealth among domestic

savings, foreign savings, and hoarding of goods. The investment assets have some trend rates

of return, and each of these returns have a stochastic element (represented by a Brownian-
motion diffusion term). Given the choices available to consumer, we derive their optimal

portfolio allocations,8 and later use these mean /variance computations to motivate a more
general asset market model. Specific formulas resulting from the portfolio problem are meant

to be suggestive, since asset prices may not precisely follow the Brownian motion processes as

specified in our derivations.

Domestic savings, denoted by B, have an expected nominal return of ib equal to the

interest paid on domestic-currency savings deposits. The risk associated with this form of
savings, σ b , arises due to the possibility that savings accounts can be confiscated in a monetary

8 The technical details of the derivation of optimal demands are provided in the Appendix.

4
reform or, alternatively, that savings are the main officially-recognized assets in the event of a

new issue of notes.9 Example of such events abound in emerging market economies, and

include: the treatment of savings by East German households savings in the process of German

unification; the January 1991 Russian invalidation of cash holdings outside of the savings

accounts; and the July 1992 Russian announcement that ruble notes issued prior to 1993 were

to be invalidated.

The second asset available to the consumer/investor is foreign-currency (i.e. dollars)

accumulated through the black market. For our exposition, the foreign currency is called dollars

and the domestic currency is called rubles. The domestic-currency value of these dollars is the
product of the quantity of dollars held, F, multiplied by the black-market exchange rate S,

where the exchange rate is defined as units of domestic currency per unit of foreign exchange.

Domestic residents are prohibited from opening interest-bearing foreign-currency accounts, so

foreign-exchange holdings do not yield a nominal return.10 The valuation and rate-of-return on
these dollar stocks depends on the expected evolution of the short-run black-market exchange

rate. The expected rate of depreciation of the ruble against the dollar is given by φ and the
standard deviation of this forecast is denoted by V s .

The third investment opportunity is via the accumulation of storable goods, G, valued at

second-economy prices, P. The expected rate of return on the hoarding of goods is the expected

rate of increase in durable-goods prices, π, (i.e. the inflation rate on domestic goods) which is
adjusted for the cost c of obtaining or storing these goods over the same interval of time for

9This risk enters through the standard deviation of the diffusion process on returns on domestic currency assets. It
also is possible to model this risk as a Poisson process. Analogous interpretations will apply to other risk terms
introduced.
10 This assumption is not important for our results and easily could be relaxed. In Poland, domestic residents are
permitted to maintain both domestic currency and foreign exchange accounts. Also, one could easily introduce
expected costs of obtaining and later liquidating black-market foreign exchange, as consumer/investors pay a range
of fees to money dealers. Also, it is possible to introduce a parameter reflecting the risks associated with holding
foreign exchange, for example, those due to the likelihood of being caught by authorities and penalized.

5
which the inflation series is measured. This cost can be interpreted as the opportunity cost of

waiting in queues or searching for the goods.11 The uncertainty in the forecast of future goods
availability and future inflation is subsumed within the index σg.

These investments are made out of available nominal wealth stock W, defined by:

W = B + PG + SF (1)

and the portfolio shares allocated to the respective assets are:

PG SF B
λ1 = ,λ2 = , and λ 3 = 1 − λ 1 − λ 2 = (2)
W W W
with λ 1 the share of wealth invested in hoarded goods, λ 2 the share of wealth held in foreign
currency assets, and λ 3 the share of wealth in domestic currency assets.12 The objective of the

consumer/investor is to choose portfolio shares optimally in order to maximize the expected

real return on his portfolio while minimizing its variability.

dw 1 dw
maximize λ ,λ U = E  − R ⋅ Var   (3)
1 2  w 2  w

where R is the coefficient of relative risk aversion, w = W/Q is real wealth, and Q is the price

index of the consumption basket.

Consumption occurs over two types of goods, domestic and imported, which enter the

price index with weights α and (1-α), respectively. We assume that importers are price-takers
in the market for foreign products and normalize so that the foreign currency price equals one.

Thus, the domestic currency prices of imported goods follow the black-market exchange rate.

The optimal-portfolio demands for each asset, provided in the Appendix, are divided

among minimum-variance and speculative portfolios. The minimum-variance portfolio

demands precisely equal their shares in the future consumption baskets on individuals: αW is

11See Stahl and Alexeev (1985) and Weitzman (1991).

12In our model the demand for domestic currency assets arises out of investment motives. This system also could be
supplemented by a transactions demand for money motive, as in Lane's (1992) study of Poland, but this would add to
the complexity of the model and would not qualitatively change our results.

6
invested in real goods, while (1-α)W is invested in foreign exchange. A short position is taken

in domestic currency assets to offset the long minimum-variance position in the other two

investment alternatives. However, this does not imply that negative stocks and demands for

domestic currency (i.e. net borrowing) will prevail. For positive domestic asset holdings, the

speculative portfolio demands for domestic-currency assets would be required to be positive

and at least as great as the negative minimum-variance demands. This leads to the following

proposition:

Proposition 1: For positive holdings of domestic currency assets, domestic interest rates
must be at least as large as the rates of return on each alternative asset, net of a linear
combination of the expected variances and covariances of those returns.

Proof: See the Appendix for conditions required for λ 3 > 0 .

For conciseness and brevity, in this main body of the paper and in Table 1 we present

only the signs of the asset-demand elasticities with respect to the parameters of financial

markets, relegating to the appendix the full set of equations and results on optimal portfolio
shares. Asset demands are normal, so that OS1 ! 0 and OI1 ! 0 : the subscripts here refer to the

specific asset of the portfolio; the superscripts refer to the derivative of this portfolio share with

respect to durable goods prices or expected currency depreciation. Increased expected hoarded
good inflation (π) raises the current demand for durable goods ( λ 1 ). This responsiveness is

decreasing in the relative-risk-aversion of the investor, since the source of demand for durables

is the investor's attempts at minimizing the variance of his investment portfolio. The

responsiveness of portfolio demands to changes in π is influenced by the variances and

covariances of the returns on alternative assets, but not by the expected returns on those assets.

7
 Table 1 Asset Demand Functions: Signs of Demand Elasticities
σb2 σg2 ib c φ π
effect on demand for \

goods-hoarding λ1 +1 - 2 - - - +

dollars λ2 +1 +3 - + + -

home-assets λ3 -1 +3 + + - -

1: Unambiguously satisfied if λ 3 ! 0
2: Unambiguously satisfied if (λ 1 − α ) +
α
>0
R
2
3: Unambiguously satisfied if (λ 2 − (1 − α )) <
α 1  φ − i b − σ s  where
+  
R R  σ 2s + σ b2 

( ) (
A = σ 2b + σ 2g (σ 2b + σ 2s ) − σ b2 + ρ gs )
2
and ρ gs is the correlation between shocks
to black market exchange rates and to second economy goods prices.

Optimal portfolio demands also shift with changes in the expected return on domestic

investments, in the costs of holding or acquiring hoarded goods, and in the uncertainty

surrounding the expected returns on the investments. The signs of these effects are presented in
Table 1. Increased goods supply uncertainty ( σ 2g ) reduces the hoarding demand for goods and

increases the demand for foreign and domestic currency.13 This risk also has a second-order
effect: it alters the responsiveness of asset demands to other policy and process changes. For
example, as σ 2g rises, the expansionary effect on goods hoarding of increasing π is reduced.

Likewise, an increased risk of monetary reform ( σ 2b ) generally increases the portfolio demands

for hoarding and foreign currency.

These optimal portfolio demands for assets are summarized by the reduced form

equations (4) through (6). All demand influences other than φ and π are collected into the

vector Z. The total amount of each asset demanded is the product of the optimal portfolio share

and the investable wealth. Stock equilibriums in the asset markets are given by:

13 The theoretical possibility of reversing the sign of those elasticities is likely to occur only if λ3 < 0.

8
 − +
λ 1  φ , π , Z  ( B + PG + SF ) = PG (4)
+ −
λ 2  φ , π , Z  ( B + PG + SF ) = SF (5)
− −
λ 3  φ , π , Z  ( B + PG + SF ) = B (6)

( )
where Z = σ 2b , σ 2g , i b , c and λ 3 ⋅ ( B + PG + SF ) = (1 − λ 1 − λ 2 ) ⋅ ( B + PG + SF )

Equations (4) and (5) provide two equations in two endogenous unknowns, S and P.

Both the black-market exchange rate and the second-economy price of goods are assumed to be

perfectly flexible. Note that this treatment of goods-price flexibility is more general than the

conventional approaches and also is appropriate if goods are traded as assets in the short run.

Also, observe that the values of the black market exchange rate and the second economy price

of goods depend on agents forecasts of expected rates of depreciation and expected inflation, φ

and π. We will assume that investors have rational expectations: they correctly anticipate the

values of S and P that will result after the market responds to some initial anticipated or

implemented disturbances.

Equilibria in asset markets are assumed to be maintained by the instantaneous

adjustment of all asset prices. In the short and medium run, flow disequilibrium can exist in the

second-economy goods markets and in the black market "trade balances". These markets

respectively provide adjustments in quantities of hoardable goods and dollars accumulated or

decumulated through illicit cross-border trades. In the long-run, asset markets, second-economy

goods markets, and the illicit trade balance are in equilibrium. Specifically, in the long-run
.
there is balance both in the current account (i.e. balanced smuggling), F 0 ; in the second-
.
economy goods market (i.e. no additions or reductions of hoarding stocks), G 0; and φ = π =

0.14

14 We use φ = π = 0 as the steady-state condition of our model. However, in practice these inflation and depreciation
rates will return to some "normal" rates determined by forces outside of the scope of the model, which need not
equal zero. For example, these forces would include the dynamic path of money creation. Any discussion herein of

9
B. Stock Adjustments on Black Markets
To complete the specification of markets for foreign exchange and storable goods, it is

necessary to introduce non-portfolio-related flows. In this respect, the emerging market

economies are not fundamentally different from other countries with active parallel-markets. In

the foreign-exchange market, illegal foreign-currency flows arise from under-invoicing or

smuggling of exports, over-invoicing of imports and from exchanges made by tourists.15

Surpluses or deficits in the second-economy trade balance, generated by changes in the net

supply of private dollars, are represented by equation (7).16

F = f (S P , S S 0 ) where f1 > 0, f 2 > 0 (7)

A surplus or accumulation of dollars in the second economy occurs when the

transactions or consumption demands for foreign goods and foreign exchange is less than the

foreign demand for goods smuggled out of the domestic economy. This second economy trade

balance is a function of the relative price of smuggled imports and the black market exchange

rate premium. As S/P, the relative price of imports, rises, the demand for domestic goods rises

and demand for foreign goods falls. This leads to an improvement in the second-economy trade
balance and accumulation of dollars, so that f1 = fS P > 0 . Increases in the black-market

exchange-rate premium, S/So, where So is the official exchange rate offered to tourists or

short-run and long-run activity clearly is limited to the specific forces triggered by the events analyzed within this
paper.
15For example, see Pitt (1984) and Kamin (1992).
16Our specification of foreign asset creation assumes that foreign asset stocks change due to cross-border activities.
An alternative approach would be to model the banking sector's ability to create foreign liabilities by indexing
household deposits to foreign currency returns. We opt for the former approach as a more accurate representation of
activities outside of the official sector. By contrast, the latter approach may be more appropriate for a country such
as Poland that has implemented more radical banking sector reforms.

10
would-be emigrants, improve the second-economy trade balance by increasing the transactions
supply of dollars, so that f2 = fS S > 0 .17
0

Equation (8) shows the determinants of changes in the availability of hoardable goods in

the second economy. Goods availability is a function of the relative price of imported and

domestic goods, as well as the premium from smuggling goods out of the official market and

into the second economy.

G = g(S P , P P 0 ) where g1 = g S < 0 and g2 = g P > 0

.
(8)
o
P P

Consumption demand for second-economy goods is increasing (net supply is

decreasing) in the relative price of imported to second-economy prices of domestic goods, S/P,
so that fewer goods are available for hoarding and g1 = gS P < 0 . Also, the net supply of goods

is increasing in P/ P0, the ratio of the free-market to the government-controlled price of goods.

In the second-economy market for goods, demand decreases and supply increases in P. This

supply response is attributed to the motives for increased theft of goods from the official sector.

Thus, an increase in P unambiguously raises hoarded goods availability. By contrast, the

impact on goods availability of changes in government-controlled prices is less clear-cut. First,

an increase in official prices can lead to a reduction in the supply diverted to secondary

markets, since the premium to theft is reduced. In the official sector, more goods are available
but at higher prices. Hence, demand in the second-economy declines as P0 increases. The
supply effects dominate, so that g2 = gP P > 0 . o

III. Dynamic Effects of Reform Initiatives

Using our asset market model supplemented by hoarding and smuggling activity, in this

section we analyze the dynamic effects of: i) official/ tourist exchange-rate devaluation (an

17 See Dornbusch et al. (1983). Note also that the function representing the net supply of dollars easily can be
modified to account for the considerable obstacles that limit East-West trade. Historically, these obstacles have
included prohibitive border-taxes, tariffs, and more subtle costs of illicit activity.

11
increase in S0); ii) relaxation of the state-controlled prices of goods, either through one-shot

price adjustment or liberalization (increase in P0); iii) a sale of government bonds to the

domestic population; iv) changes in the risk of currency confiscation, and v) changes in the

degree of uncertainty surrounding future goods supply.

The general principles behind the adjustment dynamics are similar to those observed in

all asset market models. Unlike the basic two-asset models usually studied, our system of

equations has two prices (the price of goods and the black-market exchange rate) adjusting

instantaneously to yield equilibrium in the markets for three assets. Thus, in our model both the

real and the nominal black-market exchange rates have well-specified (and not identical)

dynamics in the very short run. The dynamics of G and F (i.e accumulation or decumulation of

foreign exchange holdings and hoarding activity) also are well-specified along the path to the

steady-state equilibrium.

Analysis and simulation of the dynamic system shows that this system has a saddlepath

solution, i.e. it is characterized by two positive and two negative eigenvalues. Below, we

discuss the properties of this saddlepath and the implications of constraining adjustment along

this path. In conjunction with an analysis of the steady-state values of each variable following

policy initiatives, the saddlepath results are used to address the issue of price and exchange rate

overshooting. Before turning to the results of the analysis, the next section provides a

description of the parameters used in the simulations of our model.

Parameterization using data from Russia: Data from Russia in the early 1990s are directly

relevant for our experiment and simulations because, in addition to having experienced

extensive smuggling and hoarding, various discrete policy announcements led to overshooting

and sharp short-term speculative movements in both exchange rates and prices. Computation of

the linearized version of the system of dynamic equations (i.e. equations 4,5,7 and 8) requires
initial parameter values for: household portfolio shares λ 1 , λ 2 ; elasticities of these portfolio

shares with respect to changes in expected inflation and the expected rate of black market

12
exchange rate depreciation, λπ1 , λφ1 , λπ2 , λφ2 ; secondary market premium on foreign exchange

and durable goods, S S 0 , P P 0 ; and partial derivatives of the flow equations for foreign
exchange and durable goods accumulations, i.e. f , f , g , g .18
1 2 1 2

The true allocations of liquid household wealth in reforming economies generally are

unknown and elude survey efforts.19 In general, we expect that at least 50 percent of household
wealth is held in domestic currency assets. Thus, λ 3 = 1 − λ 1 − λ 2 ≥. 50 , and the sum of

portfolio shares toward foreign currency and hoarded assets will be at most 0.50. Our grid of
values allows λ 1 ∈ (.1,.4) and λ 2 ∈ (.1,.4) , subject to λ 1 + λ 2 <. 50 .

Since we also lack reasonable data on the price elasticities of the portfolio allocations,
we turn to the portfolio model presented in Section IIA for intuition. The model shows that
λπ1 > λφ1 , λφ2 > λπ2 , and λφ1 = λπ2 < 0 . The ranking of λπ1 and λφ2 depends on the relative size of

exchange rate variance about its trend, compared with inflation variance about its trend. Data

from the Russian economy in 1992 and 1993 show that black market exchange rate volatility

generally exceeds by threefold detrended inflation volatility. Moreover, since positive stocks of

all three assets are held by the population, Proposition 1 implies that the riskiness of domestic

18For further discussion of the Russian foreign exchange system and the implications of exchange rate reforms in this
economy, see Goldberg (1993, 1995). Goldberg and Tenorio (1995a, 1995b) discuss the link between the foreign
exchange market microstructure in Russia and its outcomes. Goldberg and Karimov (1995) consider the implications
of price and official sector foreign exchange market reforms for producer decisions regarding foreign exchange
auction participation and for their allocations of production across domestic versus external destinations.
19We have undertaken extensive efforts to determine the actual allocation of liquid household wealth in Russia. Even

those detailed household surveys that have been conducted under the auspices of international organizations and the
Soviet Interview Project have not been able to have households respond adequately to questions about liquid wealth.
The main available data related to our portfolio allocation issue are household deposits in savings accounts and
estimates of currency in circulation. But even this data most likely misses significant forms of domestic currency
savings, as well as foreign exchange holdings, and holdings of storable goods. We also consider anecdotal evidence
on this issue unreliable.

13
currency denominated assets is far inferior to the riskiness of the other two assets.20 Finally, the

data show that the covariances between these series are small relative to their variances. The

rankings of volatilities which emerge from this appeal to the portfolio approach suggest that
λπ1 ≥ 3λφ2 . Our simulations use λπ1 ∈ ( 0.1,0.3), λφ2 ∈ (0.05,0.2), and λφ1 = λπ2 ∈( −0.1,−0.02).

Finally, we assign values to the initial: black market exchange rate premia; ratio of free

to controlled prices of goods; and elasticities of response of hoarded goods stocks and foreign

exchange accumulation in the second economy with respect to their respective arguments.
These values are: S S 0 ∈(1,2), P P 0 ∈(1,2), g1 ∈( −0.1,−0.5), g 2 ∈(0.1,0.5), f1 ∈ (01
. ,0.5),

and f 2 ∈(01
. ,0.5) .

The vast majority of parameter combinations yielded an eigenvalue sign pattern of:
θ1 > 0, θ 2 < 0, θ 3 < 0, θ 4 > 0 , where the θ i represent the respective eigenvalues of the model. A

distant second pattern of eigenvalues (Case #2) is θ1 > 0, θ 2 < 0, θ 3 > 0, θ 4 < 0 . This case
appeared infrequently, and was associated with λ1 at the low end of its range (around 0.10) and

λ2 was at the high end of its range (around 0.30): in these initial conditions hoarding activity is

nearly absent.

Short-Run Overshooting and Steady-State Responses to Policy Initiatives. For our

exposition of the overshooting results we will use eigenvalues and eigenvectors generated by

20The black market exchange rate data are weekly series reported in Commersant and compiled by the authors.
Commersant is a Russian business and news publication. The price data also are weekly, and are drawn from Russian
Economic Trends. We use two alternative measures of consumer prices: an aggregate consumer price index (cpi) and
a nonfood consumer price index. It is difficult to vouch for the quality of the data. However, we were convinced that
the data are satisfactory by a range of checks for consistency with monthly series on consumer and retail prices. Both
of these measures exclude the prices of services. For computing our benchmark results on portfolio share
elasticities, we assume that the variance of returns on holding domestic currency assets is at most one half of the
variance of the return on holding real goods. We also assume that the coefficient of relative risk aversion can take on
benchmark values of one and two.

14
two sets of parameter combinations and corresponding to each of the benchmark cases.21 The

parameter combinations, corresponding eigenvalues and eigenvectors, and resulting equations

(after working through the algebra) for exchange rate overshooting are presented in Table 2.

Table 2 Eigenvalues, eigenvectors, and overshooting

Benchmark Case 1:

parameters: λ 1 = 0.30 λ 2 = 010

. λφ1 = λπ2 = −0.05 λπ1 = 0.30 λφ2 = 010
.
S S = 125
0
. P P = 125
0
. g1 = −0.25 g2 = 0.50 f1 = 0.25 f 2 = 0.50

negative eigenvalues (real parts) with eigenvectors [...]:

-4.8934 [ 2.8401, -3.8289,-0.2868, 1]

-0.1472 [-0.1167, -0.1910, 1.2792, 1]

Equations for instantaneous adjustment of exchange rates and prices:

dS0 − dS = 1.8881 ⋅ dF − 2.2986 ⋅ dG ; dP0 − dP = − 2.3231 ⋅ dF + 31626
. ⋅ dG

Benchmark Case 2:

parameters: λ = 010
. λ = 010
1
. λφ = λπ = −0.05 λπ = 0.30 λφ = 010
2 1 2 1
. 2

S S 0 = 125
. P P 0 = 125
. g1 = −010
. g 2 = 0.50 f1 = 010
. f 2 = 0.50

negative eigenvalues (real parts) with eigenvectors [...]:

-2.6767 [2.0803, -3.0513,-0.4314, 1]

-0.1079 [ -0.1054, -0.1596, 0.9799, 1]

Equations for instantaneous adjustment of exchange rates and prices:

dS0 − dS = 1.5487 ⋅ dF − 1.4122 ⋅ dG ; dP0 − dP = − 2.0490 ⋅ dF + 2.1674 ⋅ dG

The equations presented in Table 2 show the exchange rates and prices which prevail

immediately after a policy initiative places the economy on the saddlepath toward the new

21 We follow the generally accepted approach to rational expectations models which emphasizes the saddlepath
solution instead of non-saddlepath adjustments.

15
steady state. This initial jump occurs at time t=0. Using the dynamic equations of exchange

rates and prices, overshooting occurs if the instantaneous jumps in these variables exceed their

steady-state adjustments.22

Steady State and Overshooting Results: The overshooting possibilities and subsequent

dynamics of exchange rates and price depend on the steady-state results of the model. These

steady-state effects are summarized in Table 3 for: (i) an increase in the official price of

controlled goods; (ii) a devaluation of the official exchange rate; (iii) a government bond sale;

(iv) an increase in the uncertainty on government bond yields; and (v) an increase in uncertainty

about the future availability of real durable goods.23

A devaluation of the official exchange rate leads to a steady-state increase in the price

of hoarded goods and a steady state depreciation of the black market exchange rate. The

intuition is straight-forward. Since the official devaluation reduces the attractiveness to tourists

of supplying foreign currency in the second economy, in the long run there are lower F stocks

in black markets. The steady-state effect on asset markets is a nominal and real depreciation of

the black-market exchange rate. Moreover, since imported goods are relatively more expensive,

there is a shift toward greater consumption of storable goods so that the supply of goods

available for hoarding or accumulation contracts. Thus, in the steady state we observe both

lower F and lower G, and an increase in the price of hoarded goods and in the price of foreign

currency.

22 Maury Obstfeld provided exceptionally useful suggestions for addressing the overshooting problem.
23 To compute the actual overshooting results, the long-run comparative statics results for each shock,
provided in the Appendix, are used in the overshooting equations for each benchmark case. The resulting
conditions then are evaluated at the initial parameterizations of the respective benchmark cases. The
interested reader may refer to these equations to generate more detailed information about the extent of
exchange rate and price overshooting across the respective cases, and with respect to the particular policy
experiments.

16
Table 3 Sign of Steady State Implications of Policy Initiatives

Steady-State Effects
on
Policy Initiative \\ S P F G
______________________________________________________________________

Price Reform (∆Po) increase increase decrease decrease

Official Devalu. (∆So) increase increase decrease decrease

Bond sale (∆B) no effect no effect increase increase

Confiscat. threat (∆ σ 2b ) no effect no effect increase increase
Supply uncertainty(∆ σ 2g ) no effect no effect increase1 decrease1
______________________________________________________________________
1: See Table 1 for necessary and sufficient conditions on this partial derivative.

In the very short run, black market exchange rates will depreciate. Nominal exchange

rate overshooting is possible but depends on the initial parameterization of the economy. In

particular, overshooting requires real exchange rates to be “high” initially. Real exchange rates

in transition economies are likely to satisfy this overshooting criteria, since the domestic

currency price of foreign goods generally is greater than the price of domestic goods. Thus, in

the context of transition economies, a devaluation of the official exchange rate will cause a

large initial jump depreciation of the black market exchange rate, followed by a period of

corrective (but partial) nominal appreciation. Immediately after the foreign exchange reform,

the black market exchange rate premium will be at much higher levels than at the new steady-

state.

Overshooting of nominal prices is unlikely following the official exchange rate

devaluation. Prices jump instantaneously and continue to rise with gradually declining rate until

the steady state is attained. Together with the nominal exchange rate depreciation results, the

model shows that in the presence of hoarding and black markets for foreign exchange, there

17
will be both real and nominal exchange rate overshooting following this type of official foreign

market reform.

An official sector price reform, represented as an increase in the price of a controlled

price good, has the same qualitative effect on asset stocks and prices in the steady state as did

the official exchange rate devaluation. The implications of the two policy initiatives differ,

however, in the relative size of the resulting increases in S and P, and the resulting declines in

G and F. In comparison with the effects of an official exchange rate devaluation, an official

sector price reform has smaller steady-state proportional effects on S and F than on P and G,

respectively. Also, the price reform ultimately leads to a real appreciation of the black market

exchange rates, rather than a real depreciation: the proportionate increase in prices exceeds the

proportionate depreciation of the nominal exchange rates. Compared with an official

devaluation, the price reform leads to larger reductions in hoarding stocks and smaller

reductions in foreign exchange holdings by the population of consumer/investors.

In the very short run, second economy goods prices do not overshoot in response to the

increase in prices in the official sector. The nominal exchange rate bears the brunt of the

instantaneous adjustment to the price reform, with likely immediate overshooting followed by a

period of gradual and partial recovery of value. Again, the exchange rate overshoots both in

nominal and real terms, and exhibits greater volatility than prices.

Next, consider the effects of an increase in the riskiness of the returns on holding
domestic currency assets, perhaps due to changes in the perceived threat of savings or monetary

confiscation. The direct effect is to reduce the attractiveness of domestic currency assets and

increase portfolio demands for goods and foreign exchange. The immediate jump by exchange

rates and prices onto the saddlepath implies a jump appreciation of the nominal exchange rate

and an immediate increase in the price of hoarded goods. In the scenarios captured by our

benchmark cases, the immediate increase in both nominal and real black market exchange

rates implies overshooting behavior in the transition economy. As hoarding stocks and foreign

exchange availability adjust, the exchange rates will gradually depreciate and return to their

18
real and nominal levels. The initial inflation surge will be followed by a period of deflation. In

the steady state, individuals will hold larger stocks of foreign exchange and of hoarded goods

in their investment portfolios.

The other policy initiatives or circumstances which we analyze using the portfolio and

asset accumulation model are an increase in the supply of domestic bonds (or domestic

currency assets in general), and an increase in the uncertainty surrounding the availability of

real goods in the future. The steady-state implications of these measures are summarized in

Table 3. Detailed formulas for these steady state results are provided in the Appendix.

IV. Concluding Remarks

This paper has documented the behavior of black-market exchange rates, second-

economy prices, hoarding activity and foreign currency holdings in an economy with under-

developed financial markets and smuggling activity. In contrast to previous studies, we use a

framework that: (i) introduces hoarding activity and smuggling into an asset market approach,

and (ii) assumes that goods prices are determined in asset markets along with the prices of

financial assets that are used stores of value. This set of innovations enables us to more

realistically integrate the dynamics of the real and financial sectors of an emerging market

economy.

The key point is that both exchange rates and prices reflect speculative and transitory

demands for currency and for goods which arise from optimizing behavior: goods prices are

driven by the same type of speculative forces that are observed in financial markets. The

speculative influence on goods prices and exchange rates injects increased volatility into these

variables in transition economies. The existence of black markets and smuggling means that

such volatility cannot be regulated away using, for example, capital controls or taxation on

official transactions.

The results of our paper are especially important for economies in transition or in the

process of economic stabilization. We show that observed sharp increases in spreads between

19
black market and official exchange rates need not be indicative of unsuccessful reforms.

Instead, observed black market premia may be purely transitory overshooting phenomena,

reflecting inter-temporal speculation and the channels for clearing asset markets. Analogously,

prices (and the spread between free market prices and controlled prices) may rise at the outset

of specific reforms, but this type of inflationary burst also can be purely transitory.

Adding fuel to the skepticism voiced in Edwards (1989), Kamin (1993), and Montiel

and Ostry (1994), our results show that parallel market premia in both exchange rates and

prices can be misleading indicators of misalignment in developing economies. The biases

(relative to long-run equilibria) implicit in black to official premia depend on the structure of

announced and implemented policy initiatives operating in the economy in transition.

Moreover, the adjustment paths for black-market exchange rates and prices, and for smuggling

and hoarding activity, are dependent on observable features of the reforming economies,

including real exchange rates and initial portfolio allocations. These conclusions are especially

pressing for transition economies, wherein the frequency of policy changes adds to the

likelihood that exchange rates, at any moment in time, are still adjusting to previously

implemented or announced policy initiatives.

20
 Bibliography

Agenor, Pierre-Richard and Robert Flood, 1992, "Unification of Foreign Exchange Markets",
IMF Staff Papers vol.39 (December) pp.923-47.
Birman, Igor, 1980, "The Financial Crisis in the USSR," Soviet Studies vol. 32 no1 (January)
pp.84-105.
Branson, William and Dale Henderson, 1985, “The Specification and Influence of Asset
Markets,” in eds. Ronald Jones and Peter Kenen, The Handbook of International
Economics vol.II (Elsevier Publishers, The Netherlands) pp.749-804.
Calvo, Guillermo and Jacob Frenkel, 1991, “From Centrally Planned to Market Economy: The
Road from CPE to PCPE,” International Monetary Fund Staff Papers vol.38 no.2 (June)
pp.268-99.
Calvo, Guillermo and Carlos Rodriguez, 1977, "A Model of Exchange Rate Determination
under Currency Substitution and Rational Expectations," Journal of Political Economy vol.
85 no. 3 pp.617-625.
Cottarelli, Carlo and Mario Blejer, 1992, "Forced Savings and Repressed Inflation in the Soviet
Union: Some Empirical Results," International Monetary Fund Staff Papers vol.39 no.2
(June) pp.256-86.
de Macedo, Jorge Braga, 1982, "Exchange-Rate Behavior with Currency Inconvertibility,"
Journal of International Economics vol 12 pp.65-81.
de Macedo, Jorge Braga, 1987, "Currency Inconvertibility, Trade Taxes and Smuggling,"
Journal of Development Economics vol.27 pp.109-125.
Dornbusch, R., D.V. Dantas, C. Pechman, R. de Rezende Rocha, and D. Simoes, 1983, "The
Black Market for Dollars in Brazil," Quarterly Journal of Economics February pp.25-40.
Edwards, Sebastian, 1989, Real Exchange Rates, Devaluation and Adjustment: Exchange Rate
Policy in the Developing Countries (Cambridge, Mass.: MIT Press, 1989).
Frenkel, Jacob and Carlos Rodriguez, 1982, "Exchange Rate Dynamics and the Overshooting
Hypothesis," International Monetary Fund Staff Papers vol.29 no. 1 (March) pp.1-30.
Goldberg, Linda, 1993. "Foreign Exchange Markets in Russia: Understanding the Reforms,"
Staff Papers International Monetary Fund vol.40 no.4 (December) pp.852-864.

21
Goldberg, Linda, 1995, "Exchange Rate Unification with Black Market Leakages" Journal of
Development Economics vol.48 no.1 (October).
Goldberg, Linda and Il’dar Karimov, 1995, “Policy Reforms, Currency Auctions, and
Production Choices in Transition Economies”, Journal of Comparative Economics vol. 21
no. 3 (December) pp.267-288.
Goldberg, Linda and Rafael Tenorio, 1995a, "Strategic Behavior in a Two-sided Foreign
Exchange Auction," NBER working paper #5187, July.
Goldberg, Linda and Rafael Tenorio, 1995b, "Explaining Order Imbalances in Russia's

Tatonnement Foreign Exchange Auction," in eds. Janos Gacs and Merton J. Peck,
International Trade Issues of the Russian Federation (IIASA, Austria, 1995) pp. 138-159.
Kamin, Steven, 1993, "Devaluation, Exchange Controls, and Black Markets for Foreign
Exchange in Developing Countries," Journal of Development Economics (February) vol.40,
pp.151-69.
Koen, Vincent and Steven Phillips, 1993, Price Liberalization in Russia, IMF Occasional
Paper 104 (June).
Lane, Timothy, 1992, "Household Demand for Money in Poland: Theory and Evidence," IMF
Staff Papers vol. 39 no.4 (December) pp.825-54..
Montiel, Peter and Jonathan Ostry, 1991, "Macroeconomic Implications of Real Exchange Rate
Targeting in Developing Countries," IMF Staff Papers vol. 38 no. 4 (December) pp.872-
900.
Montiel, Peter and Jonathon Ostry, 1994, "The Parallel Market Premium," IMF Staff Papers
vol. 41 no.1(March) pp.55-75.
Obstfeld, Maurice and Ken Rogoff, 1995, "Exchange Rate Dynamics Redux," Journal of
Political Economy vol.103 no.3 (June) pp.624-60.
Pinto, Brian, 1991, "Black markets for foreign exchange, real exchange rates and inflation",
Journal of International Economics vol.30 pp.121-135.
Pitt, Mark, 1984, "Smuggling and the Black Market for Foreign Exchange," Journal of
International Economics vol. 16 pp.243-257.

22
Stahl, Dale O. and Michael Alexeev, 1985, "The Influence of Black Markets on a Queue
Rationed Centrally Planned Economy," Journal of Economic Theory (April) vol.35 pp.234-
250.
van Wijnbergen, Sweder, 1992, "Intertemporal Speculation, Shortages and the Political
Economy of Price Reform," Economic Journal vol.102 (November) pp.1394-1406.
Weitzman, Martin, 1991, "Price Distortion and Shortage Deformation or What Happened to the
Soap?" American Economic Review vol.81 no.3 (June) pp.401-414.

23
Appendix:
The problem of the consumer/ investor is captured by the objective function:

dw 1 dw
maximize λ ,λ U = E  − R ⋅ Var   (a1)
1 2  w 2  w

where w= W/Q is real wealth, Q is the price index of the consumption basket, and R is his
coefficient of relative risk aversion. Nominal wealth, W, is the sum of stock holdings of
domestic savings, B, foreign currency holdings, SF, and domestic goods hoarded, PG. The
risks and returns of the various processes are given by
dB
= ib dt + σ b dZb (a2)
B
dS
= φdt + σ s dZs (a3)
S
dPG
= ( π − c)dt + σ g dZg (a4)
PG
so that the portfolio shares allocated to the respective assets are given by:
PG SF B (a5)
λ1 = , λ2 = , λ 3 = 1 − λ1 − λ 2 =
W W W

The process driving the price index is:

dQ
= π q dt + σ q dZq (a6)
Q
Moreover, we assume that the domestic aggregate price index is a weighted average of the
unofficial price of domestic goods P and the price of imported-goods price PT .24 The black-
market exchange rate is used to convert the foreign price (hereafter set equal to unity) into
domestic-currency values. The domestic aggregate price index is:

Q = P α S1−α (a7)

where α is the weight of the price of domestic goods and (1-α) is the weight of the imported-
goods price in the consumer price index. Applying Ito's Lemma, assuming the risks associated
with domestic currency savings accounts are uncorrelated with other stochastic processes, we
derive the optimal portfolio shares:

24This price index could be modified to include expenditure on goods supplied at fixed official prices. However, we
view the unofficial price of goods as the relevant marginal cost of these products.

24
 1  (π − ib − c − ασ g − (1 − α )ρ gs )(σ b + σ s ) 
2 2 2

λ1 = α +  (a8)

AR  − (φ − ib − αρ sg − (1 − α )σ s )(σ b + ρ sg )
2 2

1  −(π − ib − c − ασ g − (1 − α )ρ gs )(σ b + ρ gs ) 
2 2
(a9)
λ 2 = (1 − α ) + 
AR  + (φ − ib − αρ sg − (1 − α )σ s )(σ b + σ g )
2 2 2 

1  (π − c − ib − ρ gs )(σ s − ρ gs ) 
2

 (a10)
λ3 = −
AR  + (φ − ib − σ s2 )(σ 2g − ρ gs )

( ) (
where A = (σ 2b + σ 2s ) σ 2b + σ 2g − σ 2b + ρ gs ).
2

Proposition 1: For positive holdings of domestic currency assets, domestic interest rates must
be at least as large as the returns on each alternative asset, net of a linear combination of the
expected variances and covariances of those returns.

Proof :
From (a10) positive holdings of domestic currency assets require that ib is large enough so that
(σ 2
− ρ gs ) (σ 2
− ρ gs )
(
λ 3 > 0 . ib > π − c − ρ gs )σ s
+ (φ − σ 2
)σ g
.
( 2
s + σ 2g − 2 ρ gs ) s
( 2
s + σ 2g − 2 ρ gs )
______________________________________________________________________
Table A1: Short-Run Comparative Statics Results for Asset-Demand Equations
Sensitivity of Hoarding Demands
∂λ 1 1 ∂λ 1 1
∂π
=
AR
(σ 2b + σ 2s ) > 0
∂φ
=−
AR
σ 2b + ρ gs < 0 ( )
∂λ 1 1 α ∂λ 1 1 2
= − (σ 2b + σ 2s ) + (λ 1 − α ) = σ s − ρ gs ⋅ λ 3 > 0 ( )
∂σ g2
A R  ∂σ 2b A
∂λ 1 1
∂λ 1
=−
1
(σ b2 + σ 2s ) < 0 ∂ib
=−
AR
σ 2b − ρ gs < 0 ( )
∂c AR
Sensitivity of Portfolio Demands for Black-Market Dollars
∂λ 2 1 ∂λ 2 1
∂π
=−
AR
(
σ 2b + ρ gs < 0 ) ∂φ
=
AR
σ b2 + σ 2g > 0 ( )
∂λ 2 1 2  α 1  φ − ib − σ 2s   ∂λ 2 1
∂σ 2g
= −
A
(σ s + σ 2
b )

(λ 2 − (1 − α ) ) − −  
R R  σ 2s + σ 2b   ∂i b
=−
AR
(
σ 2b − ρ gs < 0 )

∂λ 2 1 2
=
∂σ 2b A
( )
σ g − ρ gs ⋅ λ 3 > 0 ∂λ 2
=
1
(σ 2b + ρ gs ) > 0
∂c AR
( ) ( )
where A = σ b + σ g (σ b + σ s ) − σ b + ρ gs > 0 and demands are normal.
2 2 2 2 2 2

________________________________________________________________________

25
Determining the Dynamic Adjustment of Prices and Stocks to Policy Changes

I. Long-Run Equilibrium: Comparative-Statics Matrix


(
 f1 ( P)−1 + f 2 (S o )−1 ) − f 1 S ( P)
−2
0 0   ∂S 
  ∂P 
g1 ( P) g 2 ( P o ) − g1 S ( P)
−1
 
−1 −2
 0 0
 ∂F 
 −λ 1 F (1 − λ 1 )G −λ 1 S (1 − λ 1 )P ∂G
  
 (1 − λ 2 )F −λ 2 G (1 − λ 2 )S −λ 2 P 

( )
−2
f S So 0 0 0   ∂S o 
 2  o 
( )
−2
=
 0 g2 P Po 0 0   ∂P 
 
 0 0 Wλ Z1 λ 1   ∂Z 
 

 0 0 Wλ Z2 λ 2   ∂B 

Long-run Comparative Statics Results

For the examples of official exchange rate devaluation, official sector price adjustment, and
increased risk of holding domestic currency assets, the long-run comparative statics results are:

 ∂S   c11 c12 0 0   ∂S o 
    
 ∂P  = c 21 c 22 0 0   ∂P o 

 ∂F  c 31 c 32 c33 c 34  ∂σ b2 
    
∂G  c 41 c 42 c 43 c 44   ∂B 
where

Sf 2  g 2 g1 S  Sg2  f1  Pf 2  g1 S 
c11 =  −  > 0 , c =   > 0 , c = 2 − 2 
>0
κS o  P P  κP o  P  κS o  P 
2 o 2 12 2 21

Pg f f −f F g gS −f F g 
c 22 = o22  1 + 2o  > 0 , c31 = 2o 2  2o − 1 2  < 0 , c32 = 1  o 22  < 0
κP  P S  κS  P P  κ  P P
λ
c33 =
W
SA
(( ) ( ) )
σ 2s − ρ gs λ 2 + σ 2g − ρ gs (1 − λ 1 ) > 0 , c34 = 2 , c 41 = 1 22
λ 3S
g f SG
κS o P 2
< 0,

− Gg 2  f1 λ1
c 42 = 2 
κP o  P S 
+
f2 
o  < 0 , c 43 =
W
PA
σ 2
s − ρ ((
gs (1 − λ 2 ) + )σ 2
g − ρ gs (
λ 1 > 0 , c) )
44 =
λ3P

f1 g2 PS o + f 2 g2 P 2 − f 2 g1 P o S
where κ = > 0.
So Po P2

26
Table 2 Eigenvalues and Eigenvectors of the Dynamic System

Benchmark case #1
parameters:
λ1 = 0. 30 λ 2 = 0.10 λφ1 = λπ2 = −0. 05 λπ1 = 0. 30 λφ2 = 0.10
S S 0 = 1. 25 P P 0 = 1. 25 g1 = −0. 25 g2 = 0. 50 t1 = 0. 25 t2 = 0. 50

eigenvalues (real parts):

8.3344 -4.8934 -0.1472 0.1244

eigenvectors (real parts):

-0.3035 0.4050 -0.0499 -0.0901
-0.7666 -0.5460 -0.0817 0.0122
-0.0709 -0.0409 0.5471 -0.6880
-0.0783 0.1426 0.4277 0.3696

Benchmark case #2
parameters:
λ1 = 0.10 λ 2 = 0.10 λφ1 = λπ2 = −0. 05 λπ1 = 0. 30 λφ2 = 0.10
S S 0 = 1. 25 P P 0 = 1. 25 g1 = −0.10 g2 = 0. 50 t1 = 0.10 t2 = 0. 50

eigenvalues (real parts):

5.2769 -2.6767 0.1190 -0.1079

eigenvectors (real parts):

-0.3465 0.4017 0.1272 -0.0703
-0.7122 -0.5892 -0.0860 -0.1064
-0.0711 -0.0833 0.7200 0.6533
-0.0947 0.1931 -0.7202 0.6667

27