Policy, Researc,j, and External Affairs
WORKING PAPERS
Office of the Vice President
Development Economics
'The World Bank
Mlay 1990
WPS 418
Does Devaluation Hurt
Private Investment?
The Indonesian Case
Ajay Chhibber
and
Nemat Shafik
Inl tte slmrt run. de'i aluat ti.sN;Lill ht't prlivalte i\l\N.^tm t becauSe
hlighetr reall impo;)rt ^0t tol- caplital and{ inltermeldeiate goods limiit
priVate scnwtor prfti t   vhi i i.  In the  loni  rull, ttc reco\erv inI
tradablet gooods SetOs irease' prot'itabilit\ andi pri\ ate inv'est-
nent recovery. BuLt       11on 1e-' iS tlle 1- rtilln ' r 



Policy, Research, and External Affairs
_                 4 ewj
Office of the Vice President
'I'iliS paper    a produi ct olI te OIi   lic0 ' e o' thIic Vicx  Presidet   1)'\ elopo IoIICIIt FConomics  is pafIl ola' i ielI'I
elffort in PRE to understand the response ' (the privalte sector to jX)vlicies so as 1t  better design) ldjUlSnIWt
programs.  Copies arc av ailahle fre  Irnit the World Bank. Ik  1818 If Street NW, Washington DC' 2))-! '
Please  :onltac:t .\lMaureen (Cohlnt, room  SiM')-) I2, ctesiono  34W (4) (4   pags   ith figures and Itales).
DeVaIluaItiorn aICCeS lHVeC>tment f%eCJU�e of it,                  C\port demand,liland,  cinc'   im prov ements in
ef lc. on the real sIuppi>  price 1 o Lfaital cood:                erca>e proti Iity.  lnl pnvatc ivte   vstmntil t
the' re-Cal price of imponCd  upits .  ll '.', 'icI-h                t'. ers ullick!
'.ith cIapitd! gOOdS aire U\ed 1to pI'rod(1.1ce outpu1.1
tihe real product wka,e and therehy prottai  ii6                        13L t ho'0. long are the shc rt and loner rund  O
adi invcstrmenlt: real iniome, '.. ilinha Lek:'ts the
dCmand Ilor domesticallk  prodLIuceid   od    and1.                     ('h hlihber anId ShS.lTik. LlSin ln LCo0onletulic
nomirnl and  ieal inere>t riath \               ml         In      model of the  InonesianeConon1   fOlIo d t11a1 It'-
atfeet ir' LNnTICIIInt.                                            d       lnei  ald U>ted inl about t%o or threc   ecaiN
M..hich ih    a relative!l  quick tljrntaroulndr  co   :>1
Intormation on t]e "'t - X),!                                   itli othir  LountrieS> Ullnderoi m   adjutniem
ut<!  cIt pi'u;tlina    par teil;! i:z 4i   r'e-.~i                     the <iono  of tiet:               the
., t!,,p\ti', it  anloW0.It of Iitl'l    p-.< 0..:V&T'           4    t1o <        ecOlol     policie" pl\e  ad    .'
ieiu' ;t[   tle"    llort fenu   elle>L, '`-1\    v    '\            Ie           lS in    d 'aturvooud 3l'. oicl\ l 1ta:.
h'2trelt' 11'u lh h  1111'    i                                   cmpi            in     l>   it] tatXiill, tite 1tin 4!  .
11]   mti   tlII TIIl lil 11 a!! viv!   l    1"d                C111MC !l\tllilIlt CI  J.li  ;1 kitill sii <11 11  [t)II, <    I I
m,l~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~etr                                                            Liltu   xciitc   rt, h   tri    l,i'In
i! I         II Jthe                iiim                                  t 'Iise!t 1o   rNituli tdo  ntenIl   itI
~ipi~II .iid j i[ii~  uk ~  '.  nji    r   '.x  e  >ct1r uit  cthcn  ~ihYtol Vif'ol  adjl <nl;eh
'i"l'' 1'rt i'ts  i <]['<   1   'i t ''tn iix'i Ke,   Iii   ifi   cftiint > lulix']    \'.,IN  i?\li r nu Ittjii t,1 vthe4 neoc.
,;~~~~~~~~~~~I,



TA BQLE  F.NTYEN 
TEXT                                                              PAGE
1  Introduction                                                         I
2  Devaluation and Investment: The Model       ..                       6
2.1. The Channels.                                                   6
2.2. The Investment Model.                                           8
2.3. Stationarity Testing .12
2.4. The Engle-Granger Technique ..
3. Empirical Estimation for Indonesia .18
4. The Macro-Economic Mod1 .28
5. The Effect on Private Investment of Slower Devaluation              35
6. Conclusions .39
References.                                                         41
TABLES & FIGURES                                                       PAGE
Table I Indonesia: Economic Indicators.                                    4
Tabie 2  Testing for Unit Roots: Cointegrating Regression
Durbin-Watson Statistics .15
Table 3  Private Investment Equation in Levels .19
Table 4  Private Investment Equation in First Differences
with error Correction Term .21
Table 5 The Macroeconomic Model .29
Figure I Private Investment As a Share of GDP .23
Figure 2 Real Interest Rate in Indonesia .25
Figure 3 Real Import Prices in Indonesia .26
Figure 4 Real Exchange Rate Index .37
Figure 5 Private Investment/GNP .37
The authors are economists in the World Bank. We are grateful to Sadiq Ahmad, Bela
Balassa and Amar Bhattacharya for comments and help with this paper.



1. Introduction
The lack of investment response after the initiation of adjustment programs has
raised concern about their long run sustainability.' With exchange rate devaluation
2
as the sing'e most important instrument in many adjustment programs , questions
have arisen about the likely contractionary effects of a devaluation on aggregate
demand and on capital formation. These concerns have become widespread especially
since the publication of the Krugman-Taylor3 paper formalising the various channels
by which a devaluation may have contractionary effects on output and employment
in the economy. Since then a large and growing liberature has developed on this
* 4
issue4, that has been summarised in a recent paper by Lizondo and Mon.iel (1989).
A devaluation affects investment through several channels. Firstly, a
devaluation alters the real supply price of capital goods. Secondly, it affects the real
price of imported inputs which are used in conjunction with capital goods to produce
output. Thirdly, a devaluation has an impact on the real product wage and thereby
affects profitability and investment . Fourthly, a devaluation produces changes in real
income which affects the demand for domestically produced goods. Finally, a
'The World Bank recently carried out a co-nprehensive review of adjustment prograrns
and is currently conducting a second more updated review. For details see Thomas et. al.
(1989).
2See Edwards (1989) for a thorough analysis of the theoretica! and empirical
determinants and effects of exchange rates in developing countries.
3See Krugman and Taylor (1978)
See for example Branson (1986), Buffie (1986, 1989), Caval,o (1977), Gylafson and
Schmid (1984), Risager (1984), Shea (1976) and van Wijnbergen (1986).



- 2 -
devaluation affects . minal and real interest rates which in turn have an impact on
investment. The net effect of a devaluation on investment will therefore be a
composite of several factors and is theoretically indeterminate.
The short versus long run effects of a devaluation on private investment can go
in opposing directions. Even if the short run effects are negative because of increases
in the real cost of imported capital and inputs, the long run effects may still be
positive.  The increase in exports and growth due to devialuation could result in
higher private investment. Therefore, the information on the short versus long run
effects of a devaluation is critical in the design of adjustment programs and in
particular in assessing the size of external assistance required for their successful
implementation. The short run negative effects appear largely due to the
contractionary demand effect of a devaluation and to the higher costs of imported
inputs into production.' However, as the economy responds through higher exports
to increased competitiveness, the long run effects on private investment can be
significantly favorable.
It is necessary to view the effect of devaluation on private investment in a
macroeconomic context.  If a devaluation is to achieve its primary objective of
improving the trade imbalance then for a given level of output, domestic demand must
fall:
5Some ascribe a negative effect of stabilization on private investment can also arise
from the increased uncertainty about po!icy changes once the government embarks on a
program. The opposite is also possible if the i:.ception of a reform program signals that the
government is willing to take action, then uncertainty can be reduced and credibility
enhanced.



- 3 -
Y = Cp + Ip + Cg + Ig + ( X - M ).
For fixed output levels (Y) an increase in private investment (Ip) is possible
only if public expenditures (Cg + Ig) fall by more than the improvement in the trade
imbalance (X-M).6  If aggregate output falls, the short run effect on private
investment is likely to be negative. On the other hand, if the effect on output sf a
devaluation is positive due to a strong and rapid response from the tradeable goods
sectors, these negative effects can be avoided. Even if these positive effects are not
large enough in the short-run because of a weak supply response, the long-run effects
need not be. How quickly would a turn-aro, .d take place and under what conditions
would this come about?  These would natArally vary from one country to another.
In this paper, we examine the Indonesi3n case where the economy has adjusted rapidly
to oil shocks and private investment has begun to respond since 1986. The turn-
around has come about fairly quickly and provides useful lessons to countries
undergoing reform programs. The evolution of a number of macroeconomic indicators
in Indonesia is presented in Table 1.
We estimate the channels through which a devaluation affects private investment
empirically for Indonesia and attempt to quantify the composite effect of devaluation
on private investment. We do this with a model of investment and growth which is
formulated to capture the various channels mentioned above. The econometric analysis
EAssuming no effect of a devaluation on private consumption. This is verified later for
Indonesia.



-4-
TABLE
INDONESIA - ECONOMIC INDICATORS
Mid- 1987 Population (mils )171.4
1987 Per Capita GNP In US$:           450
Shares ot Gross Domestic Product                             Growth Rates(% per annum)
(from current price data)                                  (trom constant prIce data)
1965    1973    1980    1986    1987  1988p            19651-73 1973.80 1980-87    1,987  1988p
Gross Domestic Productm P.          100 0   100 0   100 0   100.0  lOJO    100 0                 8 2     7.2     3.6      3 4     4.7
Net 1b-dIrect Taxes                  2 3      46      3 3       .                .
Agrlcunture                         56 0    38.2    24.0    25.8    26.0        ..4.8                   3 8     3.0      k 8     3.8
Industry                            12 6    26.5    41.7    31 4    33 5        ..15.2                  7.3     2.1      1.6     4.8
(of whilch Manulacluulng)            8.4    10 6    13 0    14 2    14 1                       9!)    14.7      7.8      7.5     8.6
Services                            31.41   35.2    34.3    42 8    40 6         ..5.0    10.1                  5.6      5.4     5.1
Resource Balance                     -0 1      1 7    12 8     0 3      2 8     2 4..                      .
Exporfs of GNFS                      5 3    20 3    33 0    22 1    26.3    24.6              12.6      5.8    -0.3    16.8      5 5
Impod1 of GNFS                       5.4    18.6    20.2    21.8    23.6    22.2              17.6    12.6      1.1     4.4      3.1
Total ExpendItures                  100 1    98.3    87.2    99.7    97.2        ..7.5                   8.9     4.1     0.6      4.2
Total Consumpfion                   92 3    77.5    62 9    75 1    71 9         .5 6                   7.8     4.9     6.9      3.4
Pilvate Consumption                87 2    67.8    52.3    63 2    61.4        ..5 1                   6 9     5 1     8.4      4.3
General Government                  5.1      9.7    10.5    11.8    10.4       ..9.8    13.9                   4.1    -0.2    -1.5
Gross Domestic Investment            7 8    20.8    24.3    24 6    25 4        ..17.5    11.9                  3 1    -11.4     6 0
Fixed tnvestment                    7.8        . 21 6    21 7    21.8          ...                       .     0.6     4.4      7.0
Changes In Stocks                     .       ..     2.7     2.9     3.6       .                .       .
Gross Domestic Saving                 7 7    22 5    37 1    24 9    28 1         ..21.6    15.0                 0.0    -4.7      5.0
Net Factor Income                    1.2    -1.2    -4 1      -42     *5 3
Net CurTent Transfersm                        0 0     0 0     -0 1    0 1      0.2..                      .
Gross National Gaving                   ..   20 8    33 0    20.6    22 9        ..43.8    15.9                  0.4    -6.5      5.2
(1980 =100 )                                    Infiation Rates (% pa.)
Price Indices                     19A0   1984    1985    1986    1987  1988p              1965-73 1973-80 1980-87    1987  1988p
Consumer Prices (IFS 64)            100 0   1St?7   158 9   168 2   183 8   198 6              56 3    17.0      8 9     9 3      81I
MholesalaPRicos (IFS 63)            tooc0   156 2   1 63 9   16 75   198 4        ...                   18 8     9 9    18 4
implicit GDP Deflator               1 000  -15l 8   161 1   156 7   178 8   206 6              62 9    19 8      8.4    14.1    15.5
Implicfl Expendiltures Defl        100 0   148 6   159 0   158 5   181.1        ..62.7    16.8                  8.9    14.3
Shiare ot GDP I%)                                          Growih Rates
Central Government Budget          19'IC)   1S'1?4   1985    1988,   1987  1988p                   1980-84    1986    1987  1988p
Current Re-ceipts                    21 1    21 2    19 6    16 1    18 0    16 9                       16 4    -11 8    30 5     7,5
Current ExpesndfMures                14 1    12 1    13 4    13 2    13 0    12 3                       1 1.9    5.5    1 57      8 4
Current Budgel Baie,ice                7.0     9 1     6 2     2.9      5 0     4 6 
Capital Roeceipts
Capital Expenditures                  9"       8 4     9 2     7 2      77      7.7                     12.2   -16.1    26.0    14.1
Adjustments
Overall DeficIt                       -2 8     0 7     -2 9    -4 2    -2 8    -3 2
Otficial Capital Grants                        0 2     02      0 3      0 3     0 4 
Exlemal Borrowing (net)               2 7      2 6     1 9     3.6      2.4     3 1                     16.6   105.6   *24.3    50.0
Domestic Non-Bank Financing
Domestic Bank Financing               0 2     -3 5     0 8     0.3      0 1    -0.3
Otlicial x Rate (Rupiah/uss)       626 99 1025 90 1110 60 1282 60 1643 80 1696 00
Od)(ex Real Elf. X.R Base 1980     100 00   92 38   89 75   69 12   50 77   48 99



uses the recent literature on cointegr.tion and error correction models to highlight
the relationship between devaluation and private investment.
The results, while interesting in terms of their implications for Indonesia, are
also meant to be illustrative of the application of cointegration techniques.  The
approach allows for an analysis of the nature of the long run equilibrium relationship
between investment and its determinants as well as the short run dynamics, without
generating the types of spurious correlations often found in time series econometrics.
In addition, the error correction model provides a realistic framework for describing
how rational, but fallible, agents make optimizing decisions.
The paper is divided into five sections. Section I discusses the channels between
devaluation and private investment. In Section 2 we set out the error correction
growth-investment model. In Section 3 we describe the tren,  in Indonesia's key
macroeconomic variables and policy changes that have driven investment behavior.
The estimates of the model are presented in Section 4 with a set of simulations that
Icok at the short versus long run implications of devaluation on private investment.
The final section presents our conclusions.



- 6 -
2. Devaluation and Investment: The Model
2.1. TIhe Channela
There are several possible channels through which a devaluation affects
private investment. In general, a devaluation raises profitability in the traded goods
sector and lowers it in the non-traded sector. If capital is sector-specific then total
investment demand for non-traded goods must rise since the higher capital demand
of the traded goods sector cannot be met by negative gross investment in the non-
traded goods sector. In the more general case when capital is not entirely immobile
between sectors the impact of devaluation on investment is indeterminate in theory
because disinvestment in non-tradeable sectors may be larger than the increase in
investment in the tradeable goods sector.  The net outcome will depend on the
strengths of the various channels, which we discuss individually:
Cost of Imported lnputs: In recent years there has been a substantial literature
on the effects of devaluation on investment through the cost of imported inputs . A
devaluation raises the cost of imported inputs relative to non-traded goods prices but
not in terms of traded goods prices. Its effect therefore is unambiguously negative on
investment as profits in the non-traded goods sector will fall. This will not be offset
by a rise in profitability in the traded goods sector through this channel.
Supply Price of Capital:  The effects of devaluation on the supply price of
capital are indeterminate. In sectors that depend heavily on traded capital goods, a



7-
real devaluation lowers the overall supply price of capital. Where nontraded goods
constitute a larger shate of investment costs, a devaluation raises the supply price of
capital. Consequently, the net effect of devaluation on the supply price of capital
depends on the relative shares of traded and nontraded goods in total investment
costs.
Real Wages: The effect of devaluation on the real product wage is dependent
on how nominal wages are allowed to respond to higher prices.7 If nominal wages
remain fixed, then obviously a devaluation will lead to a fall in real wages, an
increase in profitability and a rise in investment demand. If wage indexation is
allowed then real wages could rise or fall depending on the degree of indexation. In
the more general case where wages adjust with a lag to overall inflation in the
economy, real wages would fall in the short run.  This should lead to a rise in
profitability and a positive effect on investment.  The net effect, however, is
ambiguous and requires empirical verification.
Real Interest Rate: In an economy wit' an open capital account the domestic
real interest rate is approximately equal to the real foreign interest rate plus the rate
of expected real depreciation.
ri = rf + ne
7These effects have been discussed by van Wijnbergen (1986) and Branson (1986).



where ri is the real interest rate in the dowestic market, rf is the real foreign interest
rate and ne is the expected real depreciation.
It is important therefore to distinguish between the effect of an anticipated
future devaluation and the effect of a previously unanticipated devaluation. Given
financial arbitrage between domestic and foreign assests, an unanticipated devaluation
therefore should have no effect on real  .erest rates. In the case oi an anticipated
devaluation, real interest rates would rise in response to the expected devaluation.
Once the devaluation took place, real interest rates would subsequently fall to the
level of foreign interest rates if domestic and foreign assets are perfect substitutes.8
2.2. The Investment Model
We now turn to a model which can be empirically estimated and which captures
these theoretical linkages. The process by which firms move from actual to desired
levels of capital stock is hypothesized to follow an error correction mechanism. The
rationale for such a hypothesis is, firstly, that error correction models have proven
to be empirically useful at explaining a number of long run macroeconomic
relationships.9  Secondly, unlike the more comt-on partial adjustment model, the
error correction approach implies that agents incur no costs for changes that are
See Ahmed and Chhibber (1989) for a theoretical exposition of this issue.
9For empirical examples, see Davidson et al, 1978; Currie, 1981; Salmon, 1982; Henry
and Minford, 1988; and Shafik (1989).



planned.   In addition, the recent literature on cointegration provides a theoretical
justification for the empirical success of error correction models. By including
variables in both levels and differences, error correction models capture both the long
run equilibrium and the short run dynamics between macroeconomic time series.
Granger (1983) has shown that error correction models produce cointegrating sets of
variables and that all cointegrating series can be represented by an error correction
mechanism.
AP, error cerrection model can be derived from the minimization of a quadratic
cost or loss  function."    Consider the  case  w' ere  the  firm's  intertemporal
optimization problem is to minimize the expected costs associated with adjusting to
the desired capital stock over an infinite horizon defined as:
co
(1)  Min EZ<c(K, - K,*)2 + (K1-Kt 1)2j
Differentiating and rearranging the abovc, an equation of the error correction
form is obtained:
(2)  A Kt = aK   + a,(Kt, - Kt *) +a2A6K,-,.
10The partial adjustment model results in consistent undershooting if the economy is
growing since agents incur a cost for any change, even a aesirable one.  See Nickell, 1985
for an analysis.
11 For a more dctailed discussion of the derivation of error correction models, see
Nickell, (1985); Hendry and von Ungern Sternberg, (1981).



10-
This formulation .iiplies that capita'l stock responds to both changes last period
as in the partial adjustment model as well as to changes in the target, K. The levels
term, (K, l -Kt,1*), vaptures the divergence from the long run equilibrium as agents
respond to new information.
The re:ationship between investment and che stock of capital is defined
conventiona'ly as:
(3) It = K, - (1-d)K,
where 8 = depreciation rate.
The long run equilibrium relationship between desired investment and the
capital stock is:
(4) It =        (1-8)Kt .
Applying the lag operator, LK=Kt,
(5) it* = []-(1-8)L],Kt*
and in differenced form,
(6) A1* = [1-(l-8)L]AK*.



- II -
Substituting equations (5) and (6) into equation (2)
(7) AIt * = ao[l-(-8)L]AKt* + a,{lt, - [l-(l-8)L]AK,.,j} + a2AI,-,
The determinants of the equilibrium capital stock are hypothesized to be
expected profits which are a function of demand and costs:
(8) K = f(PF') = f(Y, C)
where PFe = expected profits,
Y = aggregate demand
C = costs.
The importance of both demand and costs is a reflection of the nature of
technology which is hypothesized to be putty-clay. Costs are a function of domestic
and world prices which are proxied by the real domestic interest rate (r), real import
prices (m), and the real exchange rate (n). We enter the real exchange rate because
it has an effect on the supply price of capital and the real wage rate as discussed in
Section 2.1. Its net effect on investment, as we discussed ear]ier, is indeterminate.
(9) C= f(r, m, n).
Substituting equation (9) and (8) into (7) and rearranging, one obtains a dynamic
model of investment which is of the error correction form:



- 12 -
(10) Alt = boAYt + b,r, + b2&Mt + b3&n, + b4[It.1-b5Y,, -
b6r,.,-b7mt.1 - b8nt.,].
An equivalent model derived in terms of the capital stock would take the
form:
(11) AKt = boAYt + b,rt + b2AmM + b3An,
+ b4[K,_ - b5Yt, - b6rt-I - b7mt.1 - b8nt.,].
These relationships will be tested econometrically below.
2.3. Stationarity Testing
The problems associated with econometric analysis of non-stationary time
series were first identified by Granger and Newbold (1974) in their discussion of
"spurious regressions." They recommended differencing as a strategy for "de-trending"
data when spurious correlations were suspected.   However, simple differencing of
time series is both ad hoc and results in the loss of information about the equilibrium
relationship between the levels.
The recent literature on cointegration, combined with advances in testing
properties of time series, addresses both the spurious regressions critique while



- 13-
retaining  long  run  information  about the relationship  in equilibrium. "2    The
intuition is that if the econometric interaction of a set of stationary variables results
in a "white noise" error, a non-spurious statistical relationship exists and the model
is "sufficient" since there is nothing left to explain econometrically.
The first step in the process is to explore the time series properties of the
variables involved to determine what order of difierencing is required to achieve
stationarity. Many macroeconomic time series, such as output or investment, tend to
have unit roots (1(1)) implying that differencing once results in a stationary series.
Variables such as real interest rates or stock prices are often stationary without any
differencing, i.e., are 1(0).  Series that need to be differenced twice to achieve
stationarity are called 1(2). Once the order of the series is determined, it is possible
to assess in what form they should enter in the equation so as to obtain unbiased
coefficients
Given that stationarity testing is relatively new, there is still some controversy
about the power of alternative tests.  Hence, two different techniques for evaluating
the time series' properties were used.       The first, the Cointegrating Regression
Durbin-Watson (CRDW) test is the standard Durbin-Watson statistic that results from
12For a survey of the literature, see a special August 1986 issue of the Oxford Bulletin
of Economics and Statistics, volume 48, number 3, with articles by Hendry, Granger, Hall,
Jenkinson, Banerjee et al, and Molinas.   For a further discussion, see Dolado and
Jenkinson, 1987 and Engle and Granger, 1987.



- 14-
regressing the difference of the variable on a constant when the null is 1(1) and the
second difference on a constant when the null is 1(2).'3
The results for the CRDW test on data for Indonesia are reported in table 2.
The variables are defined as:
I     =      real private investment
GDP  =       real gross domestic product
KP   =       private sector capital stock
r     =      real interest rate
m     =      real import prices
n     =      real exchange rate
The CRDW  test results indicate clearly that all of the relevant variables have
unit roots, i.e. that first differencing results in stationary series. The test statistics
for the null of 1(2) are considerably over the critical value of approximately unity for
the CRL)W test.
The alternatives to the CRDW statistics are DiCkey-Fuller (DF) and Augmented
Dickey-Fuller (ADF) tests. 14  The Dickey-Fuller test where the null hypothesis is
a unit root (1(1)) takes the form:
(12) AXt = AXt, +  EajAXt, + et
See Sargan and Bhargava, 1983 for the derivation of the Cointegrating Regression
Durbin-Watson test statistic.
14 For a derivation, see Dickey and Fuller, 1979 and Dickey and Fuller, 1981.



- 15-
Table 2: TESTING FOR UNIT ROOTS: COINTEGRATING REGRESSION DURBAN-
WATSON STATICS (CRDW), DICKEY-FULLER (DF) AND
AUGMENTED DICKEY-FULLER (ADF) TESTS
[NULL HYPOTHESIS IS A UNIT ROOT, I.E., i(l)]
Variable         JE          DF  W/C           DF WiC&T          CRDW (I(IU
1                 0.79       -2.60             -1.04              0.15
GDP               1.50       -1.40              -1.66             0.05
KP               -0.31        1.95             -1.75              0.05
r                 0.35       -1.67             -2.37              0.78
m                 1.52        0.50             -0.87              0.48
n                 1.92        1.22              -0.69             0.40
ADF        ADF W/C           ADF  W/C&T        CRDW(1(2)
1                 0.09       -2.67             -2.14              3.33
GDP               0.55       -1.19             -1.77              2.42
KP                1.63       -0.46              -2.54             1.86
r                 0.63       -1.23              -2.29             2.64
m                 1.36        0.67             -0.70              2.53
n                 1.39        0.90             -0.78              2.08
Where the null hypothesis is 1(2), the test is:
(13) AAXt = bAXt l + Za AAXtj + et
j=l
The test statistic is the standard "t" test on the lagged dependent variable "b"
Because the test is sensitive to whether a drift (C) or time trend (T) are included, it
was repeated in different forms for each variable. The Augmented Dickey-Fulller test
includes second and third lags of the left hand side variable to capture any additional



- 16-
dynamics. The critical values for the ADF test are the same as those for the DF
test. '
The DF and ADF test results are ambiguous. The critical values for the DF
and ADF tests are somewhat debatable and therefore the results must be considered
tentative, especially given the relatively small sample size.  The critical values are
between -2.61 and -3.20, which are above most of the test statistics reported  in
Table 2.  The test results are much better when a constant and time trend are
included.
Stationarity testing is still at an early stage and in general there is greater
confidence in the power of the CRDW  test ovei the DF and ADF tests.  In addition,
visual inspection of the relevant variables indicates that they appear stationary after
differencing once.'6  Consequently, it seems plausible to conclude that the variables
have unit roots, i.e. that they are 1(1).
2.4.  The Engle-Granger Technique
A technique proposed by Engle and Granger for modelling cointegrated time
series will be used.'7 Engle and Granger suggest a two stage modelling procedure
Engle and Granger, 1987, p. 269.
16The fact that the variables are trended in the levels can also be seen in Figures 1, 2,
and 3 in the text.
Engle and Granger, 1987.



- 17 -
in which the equilibrium levels relationship is explored prior to estimating a dynamic
equation.  The alternative procedure of estimating an unrestricted dynamic equation
and then reparamaterizing may be   more attractive, but is not well suited to
modelling under limited degrees of freedom."I
The first stage of the Engle-Granger procedure consists of econometric analysis
of the hypothesized long run relationship in the levels to determine whether the
variables cointegrate.   Evidence of cointegration includes an R2 that is close to
unity, significant coefficients, 19 a significantly non-zero Cointegrating Regression
Durbin-Watson statistic, and significant Dickey-Fuller and/or Augmented Dickey-
Fuller test results on the residuals from  the levels regression.   The coefficient
estimates at this levels stage can be interpreted as the long run multipliers.
The second stage of the Engle-Granger procedure consists of using the lagged
residuals from the levels regression to capture the process by which agents correct for
expectational errors in previous periods.    These lagged residuals, along with
differenced terms, reflect the short run dynamics as rational, but fallible, agents
respond to changing economic signals.
is For a discussion of the advantages and disadvantages of the Engle-Granger procedure
versus general dynamic modelling, see Jenkinson, 1987 and Banerjee et al, 1987.
19Note that because of autocorrelation of the residuals, the "t" statistics from the
levels regression are biased upwards and therefore it is not possible to evaluate the true
significance of the coefficients.  However, it is possible to accept the insignificance of
coefficients at the levels stage since if a variable is insignificant when the "t" statistics are
upwardly biased, it will certainly be insignificant for the true value of the coefficient.



- 18-
3. Empirical estimation for Indonesia
We estimate the investment rnodel described in Section 2 with Indonesian data.
The model described in Section 2.2 hypothesizes that the interaction of these
variables determines the equilbrium level of investment or capital stock in the
economy. The model was tested using both private investment and the private sector
capital stock as the dependent variables.2" The investment equation is linear whereas
the capital stock equation is in logarithms, with the exception of the interest rate and
the import price index which are linear. All variables are in real terms. The first
stage levels results for the 1974-87 period are presented in Table 3. All the variables
are significant, with the exception of the real interest rate which has a slightly low
"t" statistic.  This may reflect the time series properties of the real interest rate
since it is the only variable that may be stationary without differencing. Otherwise
the interest rate, like all the other variables, is appropriately signed.
The cointegration statistics that have been constructed frorn these levels
regressions appear promising.  The Cointegrating Regression Durbin Watson test,
which resulted from regressin- the first difference of the residuals on a constant and
20 The private investment equation was run as a linear relationship whereas the capital
stock equation was run in logs (cxcept for the interest rate and the import price series which
were not in logs).
2 1
Note that the 't" statistics from the cointegrating levels regression are upwardly
biased owing to autocorrelation of the residuals. The statistics are reported here, however,
in order to assess whether certain variables should be dropped since if they are insignificant
when the 'tm statistics are are biased upward, they are likely to be irrelevant to the long run
equilibrium relationship.



- 19-
Table 3: PRIVATE INVESTMENT EQUATION IN LEVELS
- ------------------_------------------------------------------.-
Private                   Private
Investment                Capital Stock
(1)                         (2)
Regressor
Constant                   -1.59                       -0.05
(2.30)                    (0.64)
GDPt1X                      0.60                        0.55
(8.87)                    (5.31)
r,                         -2.30                       -0.18
(4.66)                    (5.02)
Mt lt          l          -0.02                        -0.001
(4.33)                    (3.36)
nt.,                        0.01                        0.0004
(2.12)                       (1.16)
KPt l                      -0.29                         0.62
5.85)                       (8.12)
R2                          0.98                         0.99
CRDW                        3.11                         3.08
F                          78.26                        6638.09
DF                         -3.79                         -4.96
ADF(2)                     -0.85                        -2.54
�__ _ _ _ _ _ _ _ _-             -__ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
GDP  =            real gross domestic product
r        =        real interest rate
m        =        real import price
n        =        real exchange rate
KP       =        private sector capital stock



- 20-
evaluating the Durbin-Watson statistic, is significantly non-zero. Both the Dickey-
Fuller and Augmented Dickey-Fuller tesEs are sufficiently high to indicate that the
residuals are "white noise." These results imp.y that some equilibrium relationship
exists between private investment, output, interest rates, and import prices.
Using the lagged residuals from the above levels regression, an unrestricted
equation using differences was estimated for both the level and stock of private
capital and are presented in Table 4. Ideally, it would be possible to include several
lags of each differenced variable and to reparamaterize according to significance until
the most parsimonious model was obtained. However, in order to preserve degrees
of freedom and given that the data are annual. only one lagged difference for output,
interest rates, and import prices was included in Table 4. The reparameterizations
in equations (4) and (6) in Table 4 have highly significant coefficients for all of the
independent variables. Output affects investment in lagged form implying that it is
serving as an indicator of expected demand. The effect of the interest rate is highly
significant reflecting the negative effects of the cost of capital on private investment.
The coefficient of the real import price index is significantly negative and is larger
than the positive coefficient on the real exchange rate. This implies that the first
round effect of a devaluation on private investment will be negative. This issue will
be explored further in the next section with a complete macroeconomic model. The
above results provide strong support for the use of an error correction model to
describe private investment behaviour in Indonesia. The lagged residuals term (Res(-
1)), which reflects the equilibrium error term, is always significant and correctly
negati\ely signed. This implies that there is a long run equilibrium relationship



- 21 -
Table 4: PRIVATE INVESTMENT EQUATION IN FIRST DIFFERENCES
WITII ERROR CORRECTION TERM
Difference                   Difference
of Private                   of Private
Investment                 Capital Stock
(3)         (4)           (5)            (6)
Regressor
DGDPt                0.518                        0.2461
(0.55)                      (1.86)
DGDP,-,              0.5517         0.6014        0.3365         0.5742
(4.77)        (11.68)       (1.93)         (4.44)
Dr,                  -1.9338        -1.9228       -0.1349        -0.1691
(7.29)        (9.26)        (3.36)         (4.98)
Dmt                  -0.0007
(0.55)
Dmt l                -0.0257        -0.0267      -0.0012         -0.0016
(6.87)       (13.75)        (3.49)         (5.54)
Dnt                                                0.0001
(0.46)
Dnt;                 -0.0128        -0.0136        0.0002        -0.0007
(3.52)        (8.04)        (0.76)         (2.69)
DKPt I               -0.2908        -0.2929        0.5978        0.6017
(7.40)        (8.36)        (7.26)         (6.42)
RESt,I               -1.6725        -1.8350       -0.7362        -1.3841
(4.07)        (8.69)        (1.55)         (3.35)
R2                     0.98           0.98          0.99           0.99
CRDW                   2.95           3.11          2.41           2.99
F                     28.78          47.51        346.59         341.34
DF                    -4.56          -3.79         -3.02          -3.04
ADF(2)                -1.42          -0.85         -2.17          -1.48
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - -



*22-
between private capital formation, output, real interest rates, and import prices and
that agents learn from past expectational errors about the target capital stock over
time.
The cointegration statistics are very high for the unrestricted versions of the
equations (3 and 5). In the reparamaterizations (equations 4 and 6), the regression
with the capital stock has better cointegration statistics than that with investment.
The cointegration statistics for equation 6 indicate that the residuals from the model
are white noise and therefore the equation explains of the time series dynamics.
Recursive estimation with one step Chow tests also showed that the parameters were
stable.
The model provides an interesting explanation of the evolution of private
investment irn Indonesia over the period.  Figure I depicts thc rise in the private
investment ratio. Non-oil private investment peaked at almost 11 percent of GDP in
1979 and then fell to a low of between 6-7 percent of GDP in the late 1980s. The
aggregate private investment series with oil included follows roughly the same pattern,
although the decline after 1979 is more gradual. The graph indicates the degree to
which investment has borne the brunt of adjustment to both favorable and
unfavorable external shocks.22
2 In contrast, consumption growth has been steady, implying that Indonesia successfully
smoothed consumption.



- 23 -
Figure I
PRIVATE INVESTMENT AS A SHARE OF GDP
(IN REAL TERMS AND IN LOGS)
0.16 -
0.15 
0,14-
0.13 -/\
0.12 -
0.1
0.12a                         
73    74    75    76    77    78    79    d0    81    B2    83    84    55    86    87
YEAR
O    PRIU/GOP



- 24 -
The moveme,it of the real interest rate in Indonesia is depicted in figure 2. The
definition of the interest rate variable is somewhat complicated by the existence of
a dual financial market up to 1983 with the state banks providing subsidized credit
to certain borrowers coexisting with private banks that lend at more market rates.23
The existence of subsidized interest rates in the state banks was not motivated by a
need for cheap public borrowing since the balanced budget law constrained the
government's ability to borrow domestically.24 Instead, the subsidized credit waL a
mechanism whereby the state transferred some of its oil rents to the private sector
in certain targeted areas, such as export-oriented projects or those undertaken by
native investors. This policy of heavily subsidised credit could not be maintained
after the oil prices declined. The 1983 financial liberalization reflected the passing
of the government's terms of trade losses due to falling oil prices to the private
sector. Real interest rates were positive after 1983, in response to the financial
reform and their subsequent movement reflected expectations about inflation, the
exchange ratte and risk.
Changes in real import prices depicted in figure 3 reflect the substantial
devaluations associated with the adjustments of the 1980s. After a decline in the real
import price index during the 1970s, the trend thereafter is steadily upward, with the
23
Fur a summary of the evolution of the Indonesian financial system, see Erquiga, 1987,
pp. 42-79. The very negative interest rates in 1979 and 1980 are aue to the high growth in
the GDP deflator of over 25 percent per annum in those years.
24 Chamley and Hussain, 1988, p. 19.



- 25-
Figure 2
REAL INTEREST RATE IN INDONESIA
(I N LOGS)
1. 14
1.12-
1.1
1.08
Li.
*      1 02
-      0.98
0.96
0.94
0.92
0.9
73    74    75    76    77    78    79    80    81    82    83    84    85    86    87
YEAR
O     REAL R



-26-
Figure 3
REAL IMPORT PRICES IN INDONESIA
180
160 -
150 -
0
z       140-
0       130-
120-
110 -
I100-
90-
80 
73    74    75    76    77    78    79    80    el    82    83    84    85    86    87
YEAR
0    REAL IMPORT PRICES



- 27 -
exception of the 1979-83 period when import prices declined slightly. This effect was
reinforced by the fact that costs were exceptionally low during the windfall period.
Interest rates were negative in real terms and the exchange rate was overvalued, thus
lowering the costs of borrowing and of importing.  These policies, which were
sustainable as long as the government had access to rents from oil, encouraged a
greater investment rate.
The decline in the investment rate during the first half of the 1980s was a
reversal of the circumstances that spawned the earlier boom. The fall in the world
price of oil had direct effects on output, and therefore aggregate demand, as well as
indirect effects on the sustainability of government subsidies of interest rates and the
exchange rate. The devaluations of 1979 and 1983 and the financial liberalization of
1983 were symptomatic of growing pressure on the government budget.  The rise in
the cost of borrowing and of imports further exacerbated the decline in private
investment as is confirmed by our econometric results in Table 4.
By the middle of the 1980s, there was a recovery in private investment which
has been sustained despite higher import costs due to the large devaluations starting
in 1986 and high real lending rates in the economy. Why has private investment
recovered in Indonesia despite rising import and credit costs?  Why is Indonesia
unique among countries that have undertaken a sweeping adjustment program in that
private investment has recovered and led to a restoration of growth in the economy?
To answer this question, we embed the investment model estimated in this section into
a larger macroeconomic model of the Indonesian economy reported in Ahmed and



- 28-
Chhibber (1989), which is briefly described in the next section. This is necessary to
capture the general equilibrium effects of devaluation on variables such as interest
rates, the real exchange rate and output that ultimately affect private investment.
4. The Macro-Economic Model
The simulation model used in this section is presented in Table 5. The model
consists of five sections - growth, private sector demand, fiscal, external and inflation
and interest rates. In the first section, a non-oil growth equation (3.1) is derived
from a Cobb-Douglas production function:
(14)    Q = f(B, K, L)
where Q is non-oil output, B is technical change , K is the non-oil capital stock and
L is labour supply.
We use two alternate formulations for B, technical change. In the first,
technical change follows a linear trend.
(15)     1 = f (time)
!n the second, we postu!ate that export orientation and less import protection
will impart greater efficiency to domestic industry and increase the rate of technical



- 29-
change. We use the real exchange rate as an indicator of changes in outward
orientation 25
(16)    B = f(n)
TABLE 5: THE MACROECONOMIC MODEL
A.  GROWTH
3.1.  gy    =      b, gk + b2gPOP + b3n + b4 DY
3 2.  gk           In K - In K(-l)
3.3.  gy           In Y - In Y (-1)
3.4.  K            Kp + Kg
3.5.  Kp           Kp(-l)(1-Sp) + Ip
3 6.  Kg           Kg(-1)(1-Sg) + Ig
B.    PRIVATE SECTOR DEMAND
3.7.   Ip   =      ao + a, Y(-l) + a2n + a3rL + a4m + a5Kp(-l)
3.8.   Cp/p =      co + c1 rf + c2 Yd + C3 Yd(-l )
3.9.   Yd   =      GNP - Net Revenue
3.10.  Sp   =      Yd - Cp
25The trade regime has not undergone major changes in Indonesia during the period
under consideration.



- 30-
C.   FI SCAL
3.11.   DEF       =        + Cs + INf - CUDT - EXPT - OREV
3.12.   CUDT      =    e .tm. M
3.13.   EXPT      =    e. tx. X
3.14.  IN         =    e. i. DEBT
D.   EXTERNAL
3.15.      CA           =    Sp - Ip - [DEF]
3.16.      n            =    e . P/Pr
3.17.      nf          =    do + di TOT + d2 Y + d3 Y,, + d4 CA/GNP
3.18.       DEBT        =    DEBT., - CA/e
E.   INFLATION AND INTEREST RATES
3.19.  P    =    f0 + f, . M2 + f2(Pf + e) + f3 DEF/GNP + f4(Pr.I) + f5 KIY
3.20.  r    =    r + (n ref - n)
3.21.  rL        r + SPRD
where:
CA          = Current Account Surplus in Rupiah
Cg          = Government Non-interest Current Expenditure
Cp          = Private Consumtpion
CUDT        = Re-enues from Customs Duties
DEBT        = Foreign debt in USS
DEF         = Fiscal Deficit
DY          = Dummy equal to I after 1981
Sg          = Depreciation rate for public capital
Sp          = Depreciation rate for private capital
e           = Nominal Exchange Rate Rupiah/US$



- 31 -
EXPT       = Revenue from export taxes.
gy          = Growth of Non-oil output
gk            Growth of Capital Stock
gPOP       = Growth of population
GNP        = Gross National Product
Ig         = Government Investment
Ip         = Private Investment
INf        = Interest payments on foreign Debt in Rupiah
if         = Average interest rate on foreign debt
K          = Total Capital Stock
Kp         = Private Capital Stock
Kg         = Government Capital Stock
M2         = Broad money supply
M          = Imports in US $
mc          = Import Cost Index
n          = Real Exchange Rate Rupiah/US$
nref       = Equilibrium Real Exchange Rate
OREV       = Government Revenue Other
P          = Consumer Price Index
Pf         = Foreign Price Index
Pr         = Rice Price Index
rL         = Domestic real deposit rate
rf         = Foreign real Deposit rate
rL         = Domestic real lending rate
SPRD       = Spread between domestic lending and deposit rates
Sp         = Private Savings
tm         = Import tax rate
tx          = Export tax rate
TOT         = terms of trade index
X           = Exports in US$
Y          = Non-oil GDP
Yw         = World Income
As there are no reliable labor statistics for Indonesia we use population growth
(gPOP) as a proxy for labor force growth. In an economy with a flexible labor market
such as Indonesia there are unlikely to be major shifts in the unemployment rate. In
agriculture-based economies like Indonesia the problem is not unemployment but
underemployment with the rural sector and the informal sector absorbing and
releasing surplus labour in response to economic cycles.



- 32-
The model is estimated in growth rates (gy) with a dummy variable (DY) after
1981 to represent the permanent downward shift in the resource base of the economy.
The two alternative formulations of the model give the following results:
gy = 0.3263 gK + 0.8561 gPoP + 0.0037 t - 0.0588 DY
(2.77)       (2.76)      (3.20)      (4.84)
TSLS, R2 = 0.77, DW = 1.56
Instruments: time,lagged capital stock, population growth, lagged real exchange rate,
world income, dummy.
gy = 0.2783 gK + 0.6592 gPoP - 0.0007 t - 0.0468 DY + 0.00053 n
(2.92)       (2.57)      (0.38)      (4.44)    (2.89)
TSLS, R2 = 0.86, DW = 1.76
Instruments : time, lagged capital stock, population growth, lagged real exchange
rate, world income, dummy.
The second equation in which technical change and efficiency changes are
represented by the real exchange rate performs much better. The share of capital in
output is 0.28, and the share of labor is 0.66, the sum is 0.94. The remaining share
0.06 is explained by technical change. The inclusion of the real exchange rate to
represent technical change is not very conventional. Others such as Balassa (1985)
have used the share of exports in growth equations to represent outward orientation.
This is also the approach used in recent work on growth by Roemer (1989). We use
the real exchange to get a direct handle on the impact of exchange rate policy on
growth. Equation 3.2 through 3.6 are self-explanatory.



- 33-
The next block (Block B) describes private sector demand. Equation 3.7 on the
determinants of private sector investment is taken from Section 2. Private sector
consumption in Equation 3.8 is based on permanent and transitory income and real
interst rates. The relevant interest rate for consumption is the Sinlgapore interest rate
as Indonesia has an open capital account and there is interest rate abitrage.26 In the
fiscal block (Block C) Equations 3.11-3.14 we single out customs duty, export taxes
and interest payments on foreign debt as these items change in response to changes
in the exchange rate.
Block D describes the determinants of the external accounts and the real
exchange rate. Equation 3.15 is the transformed national income identity . The
current account surplus equals the difference between net private savings ( private
savings minus private investment) and the fiscal deficit. Equation 3.16 defines the real
exchange rate. Equation 17 defines the real exchange rate in equilibrium. Its
arguments are the terms of trade (defined as the price of exports relative to the price
of imports), domestic output, world output and the current account deficit as a share
of GNP. The equation is derived from a more detailed external trade model. In this
model, non-oil exports are a function of the real exchange rate, world income and
domestic income.  Non-oil imports are a function of the real exchange rate and
domestic income.  This model is described in Ahmed and Chhibber (1989).The
estimated equation gave the following results:
26For detailed tests on this see Ahmed and Chhibber (1989).



- 34-
n  = 423.1108 -0.9086 TOT + 31.6469 Y -6.4465 Yw +  7.3784 CAJGNP
(2.10)   (3.50)         (2.52)     (1.86)         (3.07)
TSLS, R2 = 0.80, DW = 1.27,
Instruments: terms of trade, world income, lagged real exchange rate, fiscal deficit
as a share of GNP, lagged GDP.
The important point to note about this equation for the purpose of the
simulations is that the reference or equilibrium exchange rate is determined in terms
of basic fundamentals such as the terms of trade, and domestic and world income
growth. Note that a I point increase in the terms of trade index lowers the
equilibrium real exchange rate by about I point (-0.90), as predicted by theory. It
is also defined in relation to inflows from abroad. A larger current account deficit
represented by a lowering of the CA/GNP ratio lowers "n" thereby resulting  a real
exchange rate appreciation.
Block E determines the rate of inflation and is described more fully in Chhibber
(1989). Inflation (Equation 3.19) is determined by wage push (represented by lagged
rice price inflation), underutilised capacity (represented by the capital-output
ratio)2' and the fiscal deficit. Import costs in local currency also affect the rate of
inflation. Monetary growth determines expectations about inflation and also enters
the equation.28 Equation 3.20 represents interest rate arbitrage. The domestic real
21This is derived as follows: Underutilized capacity (U) equals capacity output divided
by actual output (y).  Capacity output is equal to the average output capital ratio (a)
multiplied by the capital stock (K). Therefore: U = a.K/y.
28See Dornbusch and Fischer (1988) for a theoretical exposition of this model.



- 35-
deposit rate is equal to the foreign real deposit rate plus expectations on depreciation
of the real exchange rate, which are defined as the difference between the equilibrium
and the actual exchdllge rate. Equation 3.21 defines the spread between domestic
lending and deposit rates.
5. The Effect of Privatc Investment of Slower Devaluation
The model is used to examine the effects of a slower devaluation on private
investment and growth. The rate of devaluation is slowed from 1983-1987 such that
the real exchange rate depreciation that Indonesia experienced did not take place (See
Figure 4). The outcome of the slower devaluation is that the real exchange rate index
remains at about the levels prevailing from 1980 to 1982. According to our model this
has implications for the fiscal deficit, the current account deficit, and private
investment and growth. Here we focus on presenting the implications for private
capital formation so as to assess why investment recovered in Indonesia despite a very
sizeable depreciation.
We refer to Sections I and 2 where we had outlined the various channels through
which devaluation might affect private investment. .-irst, domestic interest rates rise
because the real exchange rate becomes overvalued in relation to the reference or
equilibrium real exchange rate. This is an inevitable outcome of the open capital
account. Note of course that the equilibrium rate also changes (Equation 3.17, Table
3). This is becaase the current account deficit and domestic growth change. The net
outcome is a more appreciated equilibrium exchange rate as is depicted in Figure 4.



- 36 -
There  is nevertheless considerable  overvaluation  between  the actual and  the
equilibrium exchange rate resulting in expectations of a depreciation and higher
interest rates.
Second, the slower depreciation results in a smaller increase in lagged real
import costs. This has a positive effect on private investment in year 2, i.e., 1984
according to the estimated equation in section 2. Third, with an overvalued exchange
rate, export growth suffers.  As a result, the efficiency benefits from  the more
depreciated exchange rate do not materialise as export orientation declines and
growth suffers. This is captured in our model by the direct positive effect of the real
exchange rate in the growth equation. Fourth, the real exchange rate overvaluation
also hurts private investment directly as described in Section 2.2. This effect comes
from the fact that real wages are higher with a more appreciated real exchange rate
which reduces profitability. It could also be due to the negative effect of reduced
demand for investment goods in the traded goods sector which is larger than the
higher demand in the non-traded goods sector.
Figure 5 shows tlie combined effect of all these changes through the simulation
model on private investment. With a slower devaluation private investment is higher
in the second year 1984 primarily due to lower real import costs.  However, the
positive export and output performance eventually outweigh the benefits of lower real
import costs. Private investment declines thereafter and is substantially lower by
1987 than it would have been with a faster devaluation. The recovery in private



- 37-
Figure 4: REAL EXCHANGE RATE INDEX
140-
135- / r .. - . Fdl.~~~~~-i,ir..
1'O �0      -              --        _;7I- -lower
RD4.~~~~~~~~~~~~~~e a> Ikatv-
1.93C      1 9 5B3    1954       1ER.5         5
Figur5: PRIVATE INVESTMENT/GNP
.1.1.5i
.1191-'l
7 1     4-          '       ';'      
r) 1 4
:~X  '                                              I
-? A                                                   j'
n~~~~~~~~~~~~~~~~~~~~~~~~~~~ a1 t.,,   
In -, nI       R            



- 38 -
investment which has occurred since 1986 does not materialize. The downward trend
is cumulative as lower private investment leads to lower growth which in turn lowers
future private investment. It is interesting to note that the medium-term effect of a
devaluation is different from the short-run effect. The switch-over comes by the third
year of the simulation.
How does the higher private investment get financed? According to the model
simulations, partly through higher private savings but primarily through larger
foreign borrowing. The higher private savings is an outcome of higher growth in the
economy. We have assumed that the improvements in revenue are passed on to the
private sector in the form  of transfers and subsidies so that there is no direct
improvement in public savings from the devaluation. Since the government is a net
seller of foreign exchange in the economy (due to oil) the devaluation should lead to
higher revenue. But this is compensated by larger transfers to the private sector.
As a result, all of the remaining increase in p ivate investment (i.e., the difference
between the increase in private investment and private savings) is financed by
borrowing from abroad.
The availability of foreign financing w  therefore a key aspect of the recovery
in private investment. This allowed Indonesia to maintain social spending and private
consumption while financing increases in private investment by borrowing abroad.
The availability of commercial financing (despite a high debt/GNP ratio) was an
outcome of stable and correct macroeconomic poiicies followed by the government and



- 39-
the growth in non-oil exports which reduced Indonesia's debt-service ratio (as distinct
from the debt/GNP ratio).
6. Conclusions
Does a devaluation hurt private investment'? The answer is yes in the short-run
according to the simulation results presented in this paper. It does so because of the
higher import costs for capital goods which hurt private sector profitability and thus
dampen investment.  This conforms with the arithmetic of the national income
identity. There must be a cut in domestic demand if the devaluation is to succeed in
its objectives of improving the balance of payments position.  If exports rise and
imi.urts fall and if supply response is weak in the short-run, then private investment
must be squeezed unless all of the burden is put on private consumption or
government expenditure.
In  the longer run  these effects get reversed.   The real exchange rate
depreciation leads to restructuring of domestic industry to meet the rising export
demand and to efficiency improvements which improve profitability. As growth is
revived, private investment recovers quickly.  How short is the short-run and how
long is the long-ruti? T'his paper shows that Indonesia has adjusted in a relatively
short amount of time, i.e., in about two to three years. This has been a relatively
quick  turn  around  in comparison  with  other countries  undergoing  adjustment
29
programs.   Clearly other factors that are not included in our model also matter.
2 For examp!.s of adjustinent experiences, see Thomas et al (1990).



- 40-
One of these which is difficult to quantify is uncertainty and the credibility of
government policies. By orchestrating a comprehensive set of reforms in areas
including taxation, the financial sector, and the exchange rate, the government sent
a clear signal to the private sector which imparted credibility to its adjustment
efforts. This was a crucial element in the recovery of private investment, in the
emergence of an array of non-traditional exports, and in the restoration of growth
in Indonesia.



- 41 -
REFERENCES
Ahmed S. and Chhibber, A (1989): "How did Indonesia Maintain Creditworthiness and
Non-Inflationary Growth", PPR Workini, Paper (forthcoming), World Bank,
Washington, D.C.
Balassa, B. (1985): Change and Challenge in the World Economy, London: Macmillan.
Banerjee, A., J. Dolado, D. Hendry and G. Smith (1986): "Exploring Equilibrium
Relationships in Econometrics Through Static Models; Some Monte Carlo
Evidence, Oxford Bulletin of Economics & Statistics, Vol. 48, No. 3, pp. 253-
278.
Branson, W.H. (1986): "Stabilization, Stagflation, and Investment Incentives: The
Case of Kenya 1979-80," in Economic Adjustment and Exchange Rates in
Developing Countries, edited by S. Edwards and L. Ahmad, Chicago:
University of Chicago Press.
Buffie, E.F. (1986), "Devaluation and Imported Inputs: The Large Economy Case,"
International Economic Review, volume 2.
Cavallo, D. (1977), "Stagflationary Effects of Monetarist Stabilizatior, Policies,"
Ph.D. thesis, Harvard University., Cambridge, Massachusetts.
Chamley, C. and Q. Hussain, (1988): "The Removal of Interest Ceilings and Other
Regulations on Financial Assets in Thailand, Indonesia, and the Philippines:
A Quantitative Evaluation," World Bank mimeograph.
Currie, D. (1981): "Some Long Run Features of Dynamic Time Series Models," The
Economic Journal, 91, pp. 704-715.
Davidson, J., D. Hendry, F. Srba and S. Yeo (1978): "Econometric Modelling of the
Aggregate Time Series Relationship between Consumer's Expenditures and
Income in the U.K.," The Economic Journal, 88, pp. 661-692.
Dickey D. and W. Fuller (1979): "Distribution of the Estimators for Autoregressive
Time Series with a Unit Root," Journal of America' Statistical Association,
volume 74, pp. 427-431.
Dickey, D. and W. Fuller (1981): "The Likelihood Ratio Statistics for Autoregressive
T;ne Series with a Unit Root," Econometrica, volume 49, pp. 1057-1072.
Dolado and T. Jenkinson (1987): "Cointegration: A Survey of Recent Developments,"
Applied Economics Discussion Paper No. 39, Institute of Economics &
Statistics, Oxford University.
Dornbusch, R. and S. Fischer (1988) Macroeconomics, New York: McGraw-Hill.
Edwards, S. (1989): "Real Exchange Rates. Devaluation and Adjustment, Cambridge,
Massachusetts, MIT Press.



- 42 -
Engle, R. and C. Granger (1987): "Co-integration and Error Correction:
Representation, Estimation, and Testing," Econometrica, volume 55, No. 2, pp.
25 1-276.
Erquiga, P. (1987): "Improving Domestic Resource Mobilization through Financial
Development -- Indonesia," Asian Development Bank Economic Staff Working
Paper No. 40.
Gelb, A. and B. Glassburner (1988): "Indonesia: Windfalls in a Poor Rural Economy,"
in A. Gelb and Associates, Oil Windfalls: Blessing or Curse? World Bank and
Oxford University Press.
Granger, C. (1983): "Co-integrating Variables and Error Correction Models,"
University of California, San Diego Discussion Paper, number 83-13.
Granger, C. and P. Newbold (1974): "Spurious Regressions in Econometrics," Journal
of Econometrics, 2, pp. 111-120.
Granger, C. (1986): "Developments in the Study of Cointegrated Economic Variables,"
Oxford Bulletin of Economics and Statistiu, vol. 48, number 3, pp. 213-228.
Gylafson T. and NM. Schmid (1983): "Do Devaluations Cause Stagflation?" Canadian
Journal of Economists, vol. 16, pp. 641-54.
Hall, S. (1986): "An Application of the Granger and Engle Two-Step Estimation
Procedure to United Kingdom  Aggregate Wage Data," Oxford Bulletin of
Economics & Statistics, vol. 48, No. 3, pp. 229-240.
itWnry, S. and M. Nfinford (1988): "United Kingdom Manufacturing Investment in the
1980s: Some Estimates using Error Correction Models," Discussion Paper
Number 137, National Institute of Economic Research.
fiendry, D. and T. von Unjern-Sternberg (1981): "Liquidity and Inflation Effects on
Consumer's Expenditure," in Dealton, A. editor,  ssavys in the Theory and
MeasurerneoQfonsumer's Behavior, Cambridge: Cambridge University Press.
Jenkinson, T. (1986): "Testing Neo-Classical Theories of Labor Demand: An
Application of Cointegration Techniques," Oxford Bulletin of Economics and
Statistics, vol. 48, No. 3, pp. 241-252.
Krugman P. and L. Taylor (1978): "Contractionary E.ffects of Devaluation," Journal
of lnternational Economics (Amsterdam), Vol. 14, pp. 37-60.
Lizondo J.S. and Peter Montiel (1989): "Contractionary Devaluation in Developing
Countries: An Analytical Overview," IMF Staff Papers.



- 43.
Nickell, S. (1985): "Error Correction, Partial Adjustment and all that: An Expository
Note," Oxford Bulletin of Economics and Statistics, vol. 47, No. 2, pp. 119-
129.
Risager, 0. (1984): "Devaluation, Profitability and Investment: A Model with
Anticipated Future Wage Adjustment, Institute for International Economic
Studies, Seminar Paper 287, Stockholm, Sweden.
Salmon, M. (1982): "Error Correction Mechanisms," Economic Journal, 92, pp. 615-
629.
Sargan, J. and A. Bhargava (1983): "Testing Residuals from Least Squares Regression
being Generated by the Gaussian Random Walk," Econometri a, 51, pp. 153-
174.
Serven L. and A. Solimano (1990): "Private Investment and Macroeconomic
Adjustment: An Overview," PPR Working Paper 339, The World Bank.
Shafik, N. (1989): "Private Investment and Public Policy in Egypt, 1960-86" D. Phil.
Thesis, Oxford University.
Shea, K.L. (1976): "Imported Inputs, Devaluation and the Balance of Payments: A
Keynesian Macro Approach," Southern Economic journal, vol. 43, pp. 1106-11.
Solimano, A. (1990): "How Private Investment Reacts to Changing Macroeconomic
Conditions: The Case of Chile in the 1980s," PPR Working Paper 212, The
World Bank.
Thomas, V. et al (1990): Structural Adjustment and the World Bank, forthcoming,
Oxford University Press.
van Wijnbergen, S. (1986): Exchange Rate Management and Stabilization Policies in
Developing Countries," Journal of Development Economics, vol. 23, October
pp. 227-47.



PRF Working Paper Series
Co ntact
IILik                             Author                    Date           for paper
WPS402  The GATT as International Dtscipline J. Michael Finger      March 1990       N. Arlis
Over Trade Restictions: A Public                                            38010
Choice Approach
WPS403  Innovative Agricultural Extension  S. Tj,p Walker
for Women: A Case Study of
Cameroon
WPS404  Chile's Labor Markets in an Era of    Luis A. Riveros       April 1990       R. Luz
AdJustmert                                                                 34303
WPS405  Investments in Solid Waste         Carl Bartone             April 1990       S. Cumine
Management: Opportunities for    Janis Bernstein                           33735
Environmental Improvement        Frederick Wright
WPS406  Township, Village, and Prnvate     William Byrd             April 1990       K Chen
Industry in China's Economic     Alan Gelb                                 38966
Reform
WPS407  Public Enterprise Reform:          AhmedGalal               April 1990       G. Orraca-Teie'h
A Challenge for the World Bank                                              37646
WPS408  Methodological Issues in Evaluating  Stijn Claessens
Debt-Reducing Deals              Ishac Diwan
WPS409  Financial Policy and Corporate     Mansoor Da lami          April 1990       M Ragga- '
Investment in Imperfect Captal                                             37657
Markets  The Case of Korea
WPS41 0  The Cost of Capital and Investment  A!an Auerbach          April 1990       A Bha,la
in Developing Countries                                                    37699
WPS411 Institutional Dimensiors of Poverty   Lawrence F. Salmen     May 1990         E Madrona
Reduction                                                                  37496
WPS412  Exchange Rate Policy In Develop ng  W  Max Corden           May 1990         M. Corder
Countries                                                                  39175
WPS413  Supporting Safe Motherhood         L. M Howard
A Review of Financial Trends
Full Report
WPS414  Supporting Safe Motnerhood. A      L. M. Howard
Review of Financial Trends,
Summary
WPS415  How Good (or Bad) are Count,y      Normar Hicks             April 1990       M. Berg
Projections'                     Mlchel Vougeois                           31058



PRE Working PaDer Series
Contact
Litle                            Author                     �t             for paper
WPS416  World Bank Efforts at Poverty      Paul Glewwe
Measurement in the Third World: The
Living Standards Measurement Study
WPS417  Modeling the Macroeconomic         William Easterly         May 1990         R. Luz
Requirements of Policy Reform    E. C. Hwa                                 34303
Piyabha Kongsamut
Jan Zizek
WPS418  Does Devaluation Hurt Private      Ajay Chhibber            May 1990         M. Colinet
Investment? The Indonesian Case   Nemat Shatik                              33490
WPS419  The Design and Sequencing ot       Brian Levy
Trade and Investment Policy Reform:
An Institutional Analysis
Wr0S420  Sustainability in Irrigation      Gerald T. OMara
Investments
WPS421  Taxation of Financial Intermediation: Christophe Chamley
Measurement Principles and       Patrick Honohan
Application to Five African Countries
WIPS422  Civil Service Reform and the World  Barbara Nunberg
Bank                             John Nellis
WIPS423  Relative Prce Changes and the     M. Shahbaz Khan
Grcwth of the Pubiic Sector