Policy Rosearch j
WORKING PAPERS
Public Economics
Policy Research Department
The World Bank
September 1993
WPS 1178
Productivity of Public
Spending,
Sectoral Allocation Choices,
and Economic Growth
John Baffes
and
Anwar Shah
The model results suggest that reshaping public spending priori-
ties in favor of human resource development and away from
military spending would positively stimulate world economic
renewal.
PRcy pR-o WodsPapas    t6   of   propesp and cacouagetheexchange dides amongBPnk staffand
sUcths intaeaed in dvcl vng  lexpap, dimbutedby fiRahAdvyStaff,cy "-offlwauth-euftd
ly theirviews,andshould ouand ciudaccordingly.Thefindhn,intelations, ando  aetheauthon'Own. 1heyshould
not be tatibuted to dte Wodd Bank. its Board of Di;,tois its managunro, X any of its manbo counimes.



Policy Reseach
Public Eoonomlc
WPS 1178
This paper - a product of the Public Economics Division, Policy Research Department- was prsented
at the 1993 Annual Meetings of the American Economic Association in Anaheim, California, iii January
1993. Copies of the paper are available free from the World Bank, 18i3 H Street NW, Washington, DC
20433. Please contact Carlina Jones, rom NA0-063, extension 37699 (September 1993, 15 pages).
Baffes and Shah examine the composition of         private capital and labor. Output elasticity of
public spending and its implications for eco-      infrastructure capiWa1 was found to be relatively
nomic growth.                                      small, with the exception of Latin American
countries where it exhibited relatively high
They use a translog production function by      values. Militar! capital had negative output
treating gross domestic product as the output and  elasticity in slightly more than half of the cases
labor, private capital, and several types of public  considered.
sector capital slocks as the inputs, using time-
series data for 25 countries for 1965-84.              The rsults suggest that reshaping public
spending priorities in favor of human resource
Thn production functions of all but four        development and away from military spending
countries exhibited increasing returns to scale.   would positively ,timulate world economic
The highest output elasticity was for human        renewal.
resource development capital, followed by
The Policy Research Working PaperSeries disseminates terfindings of work wnder way in theBaniL Anobjecaveofthe sties|
is to get these findings out quickly, evenl if presentations are less than fully polished. The findings, interetations, and
conclusions in these papers do not necessarily represent official Bank policy.
,,Produced by the Policy Research Dissemination Center
-) 0\t           ,>                 ,39



PRODUCTIVITY OF PUBLIC SPENDIANG, SECrTORAL ALLOCATION CHOICES
AND ECONOMIC GROWTH
John Baffes and Anwar Shale
The World Bank
This paper was presented at the 1993 Annual Meetings of the American Economic Association,
Anaheim, Califorrua, January 5-8, 1993. The authors are grateful to the session participants
(David Canning, Michael Dietch, Charles Hulten, Dale Jorgenson, Ishaq Nadiri and Robert
Schwab) for comments, and to Costas Christou for substantial input. They would also like to
thank Lawrence Summers for inspiration for this paper and Nancy Birdsall, Sadiq Ahmed,
GobindNankani, JavadKhalilzadeh-Shirazi, WilliamMcCleary, AntonioEstache, Vikram Nehru,
Michael Stevens, Shanta Devarajan, Vinaya Swaroop and Min Zhu for comments.
Please address correspondence to: Anwar Shah, Principal Economist, Public Economics
Division, Policy Research Department, The World Bank, Room N1O-053, 1818 H Street, N.W.,
Washington, D.C. 20433. Tel:(202)473-7687, FAX:(202)676-0025



PRODUCTIVITY OF PUBLIC SPENDING, SECTORAL ALLOCATION CHOICES AND
ECONOMIC GROWT
1. INTRODUCTION
Government deficits especially in developing and emerging market economies have been rising
at an alarming pace in recent years. Some countries have responded to this fiscal crisis by
reducing public spending on infrastructure and human resource development. At the present
time, policymakers di not appear to have access to any empirical evidence on the sectoral
allocation of public investmem and its implications for economic growth. Such evidence, as
pointed out by Summers (1991, 1992) is vital in making tough choices on public spending
priorities. In the absence of this empirical guidance, Summers suggested that in the developing
world at least a pnima facie case can be made to protect public investment in critically deficient
aspects of infrastructure and female eduication.
According to .ecent advances in endogenous growth theory, factors contributing to the
crors-country differences in both the level of per capita income and the growth rates are:
investment in human capital (Lucas, 1988), knowledge spillovers (Romer, 1989), and investment
in physical capital and infrastructure (DeLong and Summers, 1990; Murphy et al., 1989).
Looking at the microeconomic aspect of those issues, a number of recent studies have
reflected on the productivity of public spending on infrastructr (see e.g. Aschauer, 1989; U.S.
Congress, 1991; Shah, 1988, 1992; Berndt and Hansson, 1992; Richards, 1992; Lynde and
Richmond, 1993). However, sectoral allocation of public investment (i.e. on infrastructure,
human resource development capital, defense capital, etc) and its implications for economic
growth remains largely an unexplored area of research.
This paper takes an important first step in the above direcdon. Specifically, it employs
a flexible production structure methodology where various public and private inputs interact and
contribute to national output. Public capital is disaggregated into infrastructure, human resource
development and military capital stocks. Based on an analysis of time-series (1965-84) and
cross-section (25 countries) data, the paper concludes tF 4 public investment in human resource
development provides a stimulus for economic growth, whereas the contribution of military
spending to economic growth appears to be negative for a substantial number of countries.
1



Thus, the paper fmds some empirical support for the development strategy that argies for
curtailment of public spending on defense and higher spending on education training and health.
The rest of the paper is organized as follows. Section 2 provides a description of the
empirical framework within which the estimation procedures will be carried out, while section
3 describes the data used. Section 4 presents and discusses the econometric results. Finally,
a concluding section addresses the policy implications of results.
2. MODEL SPECIFICATION AND ESTIMATION
The Solow-type models (Solow, 1957) postulate that a country's output can be represented by
a well-behaved aggregate production relationship of the following form:
(1)                              Q -.fK; t),
where Q denotes a measure of aggregate output, K is a vector referring to measures of different
types of inputs used in the production process (typically laoor and physical capital), while t
represents time and is supposed to capture technical change. With a few exceptions, relationship
(1) is approximated by a Cobb-Douglas type of technology.'
In response to the limited class of variables included in K, another body of literature has
emerged. This body of literature uses cross-country regressions in order to detect linkages
between growth rates and a variety of economic indicators.2 The obvious advantage of the latter
class of models is the flexibility of including a variety of explanatory variables. However, the
structure of this (linear) formulation does not allow one to detect any substitutability or
complementarity conditions among factors of production that might be present.
In order to account for the inflexibility of Cobb-Douglas fumntion and at the same time
take into account the fact that factors of production may be substitutes or complements with one
another, relationship (1) is approximated by the following translog specification (Christensen et
al, 1973):
n              n a
(2)          InQ - ao + Ea1lnK, + (1/2) E a InKKj + yt + ,,
1-1           t1  -1U
2



where u, is an error term assumed to possess all classical statistical properides, while c& c*i, cr
and 'y denote parameters to be estimated, with a = ct,,; one can impose constant returns to scale
in (2) by restricting the parameters as follows: crai - I = ,acr = 0, v i, j. However, since one
of the major objectives of this study is to test rather than assume such a structure, we considered
the unrestricted version of the model.
Differentiation of (2) with respect to input i gives the marginal product of input i, which
expressed in elasticity form defines the output elasticity with respect to input i:
(3)                         e      3 s  aInQ  ~ at + <a   n
Note that the e, defmed above can be viewed as a unit free measure of the relative contvibution
of capital input i to the overall output and its interpretation is straightforward: How much the
output is going to change if the use of input i changes by one percentage unit? Furthermore,
based on the estimates of the e,, one can calculate the elasticity of scale as; e = E,c,.3 Values
of scale elasticity exceeding unity are consistent with increasing returns to scale; conversely,
valhes of f less than unity are consistent with decreasing returns to scale.
As a measure of output we have taken per capita Gross Domestic Product (GDP). We
have considered five types of capital stocks. These include labor, private capital, infrastructure
capital, human resource development capital and military capital, and again divided by the
population in order to come up with the per csnita figures for the inputs. It is important to note
that the human resource development capital, in a sense, captures different quality levels of labor
that might exist among countries or even within the same country at different time periods.
That, in other words, corresponds to the skill and experience augmented notion of labor or the
conrept of effective labor (Lucas, 1988).
Before we proceeded with the estimation, some preliminary tests tvere performed in order
to further invest':ate the nature of the employed technology. As mentioned earlier, one of the
justifications and innovations of this study is the introduction of a flexible production structure.
Therefore, a reasonable point of departure was to test the hypothesis of whether the above
production specification is consistent with the Cobb-Douiglas technology. In ternms of the
parameters defined in (2) such a test requires that all the aois (i.e., the cross-terms coelficients)
3



be zero. F-statistics uniformly rejected a Cobb-Douglas specification for the production function
in favo; of a translog form for all five regions and tt -refore pointing to the conclusion that a
flexible specification is more appropriate than a Cobb-Douglas one.
The second set of tests involved the question of whether the employed production
function is homogenous of degree one, or stated otherwise whether the technology is
characterized by constant returns to scale. Such a hypothesis was tested by imposing the
restriction: ,cv, - 1 = Ealj = 0. V i, j. This hypothesis was also rejected, at the 5% level of
significance, for all five regions considered, suggesting that the employed technology is not
characterized by constant returns to scale.
Ordinary least squares was utilized in order to estimate the above translog specification.
Because it is likely that there exist many differences among the countries considered within each
regional group, two types of dummies corresponding to each country were introduced in each
equation. An intercept dummy to capture differences in levels (level effects) and a time trend
dummy to capture differences in the slope (slope effects) were used.
3. DATA DESCRIPTION
Our sample covers the 1965-84 period and includes a total of 25 countries from different
regions of the world, characterized by a wide range of income levels. Although it would have
been desirable to apply the translog specification on a country by country basis, because of
degrees of freedom limitations we pooled the data and considered regressions by region and
income level rather than by country.
The countries of our sample are divided among five regions as follows: (1) AFRNCA:
Egypt, Kenya, Tanzania, and Zimbabw2; (2) ASIA: Indonesia, India, Pakistan, Korea, Sri
Lanka, Malaysia, Philippines, and Thailand; (3) EMENA: Greece, Iran, Israel, Portugal, and
Turkey; (4) LAT: Bolivia, Colombia, Mexico, and Venezuela; (5) OECD: Canada, Japan,
U.S.A., and Germany. The description of the variables used in the analysis is as follows:
GDP: Gross Domestic Product was obtained from the IMF, International Financial
Statistics database aW4 was deflated by the GDP deflator and divided by nopulation in order to
express it in per CE :ta terms.
4



Labor: Data on manhours worked were obtained from the International Labour Office,
Yearbook of Labour Statistics (various issues) and from unpublished World Bank data sources.
Private Capital Stocks: Series wer- .uonstrucled using the perpetual inventory method
from published and unpublished data on private investment in the U.N. System of National
Accounts. A depreciation rate of 10% was assumed.
Infrastructure Capital Stocks: Series were constructed from the IMF, Government
Finance S:..istics (various issues) and U.N. System of National Accounts (various years). The
perpetual inventory method was used with a depreciation rate of 5%.
Human Resource Development Capital Stock: Ser,es were constructed using the
perpetual inventory method and assuming a 10% depreciation rate, from relevant public
investment series on health, education and training obtained from IMF, Government Finance
Statistics (various issues) and the U.N. System of National Ac xounts (various years)4.
Military Capital Stock: Series were constructed from the IMF, Government Finance
Statistics (various issues) and the U.N. System of National Accounts (various years). The
perpetual inventory method was used with a depreciation rate of 10%.
4. RESULTS
Table 1 reports parameter estimates consistent with specification (2). To conserve space we do
not report parameter estimates associated with country specific intercept and time trend
dummies. It is important to note that the signs of the parameter estimates do not have a direct
interpretation for our purposes. However, most of the coefficients are significant at the 5%
level of significance; further, looking at the R2s and the F-values (bottom of Table 1), one can
conclude that the model performed in a quite satisfactory manner. ror example, with the
exception of the LAT region, which had an adjusted R2 of 0.97, all other adjusted R2s were 0.99.
At the same time, all F-statistics were significant at the 1 % level of significance.5
The first five columns of table 2A report elasticity estimates for the five types of capital
stocks (i.e. labor, private capital, infrastructure capital, human resource development capital and
military capital), evaluated at the sample means. Furthermore, the last column of the same table
presents the scale elasticities for each of the sample countries. Table 2B reports the same class
of results based on a different classification scheme. Specifically, we have classified tLle
5



countries of our sample according to their 1990 per capitv GDP (ranked from the lowest to the
highest).6
The overall conclusions drawn from Tables 2A and 2B can be summarized as follows.
In most countries scale elasticities exceeded unity, ttus pointing to the conclusion that the
economies considered exhiLdt increasing returns to %-ae.. Four notable except.ons were Canada,
Israel, Pakistan, and Zimbabwe. Private capital and human resource development capital
seemed to be the major determinants of GDP; labor stocks were the third, in terms of
importance; infrastructure capital exhibited low output elasticity; and finally, military capital had
negative output elasticities in many instances.
As already noted, Table 2B reports the same elasti:ities as Table 2A, but with the
countries classified in a different scheme. In particular, the countries have been ordered from
the ones with the lowesL to the ones with the highest per capita GDP. Such an ordering scheme
was motivated by the desire to identify possible consistent patterns between the stage of
economic development (approximated by the level of per capita GDP), and the relative
contribution of the each of the different factors of production.
First consider the scale elasticities. On the average, medium income countries exhibited
higher scale elasticity than low income countries (1 23 versus 1.05 respectively). Furthermore,
medium to high income countries exhiuited higher scale elasticities than medium income
countries. High income countries exhibited the same level of scale elasticities as the medium
to high income ones (if one excludes Israel). One, therefore, would tend to conclude that higher
per capita GDP is consistent with higher scale elasticities.
The elasticity of labor stocks in the low income countries averaged 0.15 while for the
medium income it averaged 0.17. For medium to high and high income they averaged 0.23
and 0.44 respectively. One can observe a clear pattern here: As per capital GDP rises, the
relative contribution of I.-bor increases as well. The overall contribution of labor however, with
the exception of high income countries, was lower than the one of private capital as weli as the
one of human resource development capital.
We now turn to the private capital stocks. The average output elasticity with respect to
private capital for the low income group was 0.27, while for the other groups it was very much
comparable (it averaged to 0.37). In terms of overall performance, with the exception of high
6



income countries, private capital ranked second after human resource development capital, in
terms of its contribution in output growth.
The third capital stock used in tie analysis is infrastructure. As mentioned earlier, the
output elasticities of infrastructure capital were found to be relatively low. The only geographic
region which exhibited relatively high output elasticity is the LAT region (it averaged 0. 15). All
other regions exhibited elasticities of this type of capital lower than 0.05. It is important to
mention that the output elasticity of infrastructure estimates presented here are not directly
comparable to the ones obtained by Aschauer (1989), Munnell (1990) and Shah (1988, 1992),
since all those studies estimate output elasticities of infrastructure considering private output
rather than national output. The output elasticities of infrastructure using private output are
generally expected to be higher than the ones using national output (GDP).
Human resource development capital showed the highest or-vput elasticity compare"' to
all other types of capital stocks. For the low income countries, the average output elasticity of
human resource development capital was 0.58, twice as much compared to the one of the private
capital stocks. The same result holds for the medium income countries for which the average
human resource development capital elasticity with respect to output equals J.65, almost twice
its private capital counterpart. However, for the high income countries, the human resource
development capital elasticity of output was lower than the elasticity with respect to labor and
similar to the one with respect to private capital. Furthermore, the human resource development
capital elasticity of output for the high income counaries was the lowest among all four groups.
The last type of stock considered was military capital. In 13 countries the output
elasticity of military capital turned out to be negative.7 With the exception of Tanzania, in all
low income countries the military capital exhibited a negative output elasticity. On the other
hand, with the exception of Venezuela, the medium to high income countries were characterized
by positive elasticities. The other two groups presented rather mixed results.
S. CONCLUSIONS AND POLICY IMPLICATIONS
The objective of this paper was to examine the relationship between the different components
of public investment and the rate of economic growth. A translog production model was
estimnated, by expressing per capita GDP as a function of the following five types of capital
7



stocks: labor, private capital, infrastructure capital, human resource development capital, and
military capital. From a sample of 25 countries which covers the 1965-84 time period, with the
exception of four cases, the production functions of ail countries exhibited increasing returns to
scale. The highest output elasticity was wit.k respect to human resource development capital
followed by private capital and labor. Infrastructure capita' exhibited low output elasticity, while
military capital had negative output elasticity in slightly more than half of the cases considered.
'This suggests that reshaping public eypenditure priorities iP favor of human resource
development capital and away from miliiary spending will provide a positive stimulus for the
growth of the world economy.
8



ENDNOTES
1. A relatively large body of literature has utilized Cobb-Douglas specification in growth accounting models
e.g. Aschauer (1989), Antle (1983), Thomas and Wang (1392).
2. For a survey of models using linear regressions of growth on economic indicators see Levine
and Renelt (1992). They also examine the sensitivity of these models regarding the particular
selection of explanatory variables. Also, for the relationship between growth and technical change
see Hulten (1992). For some empirical work on the subject see Devarajan et al., 1993 and Landau,
1993 on the economic impact of military expenditures.
3. Note that the definition of returns to scale is f = 81n f(\K)/8In;\   , where K denotes the input vector
(McElroy, 1969). The definition introduced in the text is used for computational purposes.
4. Only capital spending data are utilized.
5. One should note here that the LAT region had the least satisfactory performance among the five regions
considered as the t-statistics indicate.
6. We have used the ICP measure of GDP which is based on the United Nations International Comparison
program (World Development Report, 1992).
7. The output elasticities with respect to inputs are unit free measures of the marginal products which are
expected to be positive in order for the production function to be well behaved. Therefore, the finding of
negative elasticity of output with respect i military capital contradicts the assumption the production function
is well behaved.
9



TABLE 1: Translog Production Function Parameter Estimates
AFrdCA      ASIA        EMENA        OECD        LAT
In(K,)                  -1.403      -1.755       8.482      -16.17       5.604
(-0.27)    (-1.66)     (4.45)       (-3.47)     (0.41)
In(K,                   -0.352       2.468       5.283       7.070       -1.532
(-0.23)    (3.24)       (2.64)      (4.46)      (-0.46)
In(K3)                  -0.889       0.152      -2.687       3.099       2.101
(-2.09)    (0.78)       (-4.01)     (2.73)      (1.63)
ln(K)                    2.295       0.342       6.308      -1.423      -6.911
(1.10)      (0.39)      (3.28)      (-0.82)     (-0.66)
In(K.)                   2.744      -1.935      -2.262       7.219       5.680
(1.55)      (-2.22)    (-1.68)      (2.52)      (0.75)
In(K,)ln(K,)            -0.088       0.255      -0.375      -0.966       -0.579
(-0.31)    (5.51)       (-3.32)     (-1.89)     (-0.59)
In(K,)ln(K)              0.739      -0.112       0.440       1.104        1.380
(5.39)      (-2.06)    (1.90)       (2.68)      (1.83)
In (K) ln (K3)           0.029      -0.025      -0.197       0.387       0.020
(0.51)      (-1.96)    (-2.97)      (2.48)      (0.14)
In (K)In (K)            -0.181      -0.133      -0.649       0.154       -1.451
(-0.47)    (-1.35)     (-2.94)      (0.59)      (-1.29)
In(K)ln (K)            -0.229       0.284       0.100      -0.119       0.163
(-1.10)    (3.01)       (1.10)      (-0.38)     (0.20)
In (K2) In (K)          -0.642      -0.194      -1.159      -0.315       -0.539
(-0.31)    (-3.08)     (-3.12)      (-2.71)     (-1.85)
in (K) In (K3)          -0.241      -0.01'2      1.272      -0.215       -0.571
(-1.14)    (-0.46)     (5.83)       (-2.55)     (-2.38)
In (K ln (K)            1.302       0.186       0.428       0.160        1.263
(2.66)      (3.14)      (0.84)      (1.16)      (1.70)
In (K iln (K)           -0.699      -0.043       0.401      -0.098       0.126
(-3.67)    (-0.91)     (1.13)       (-0.45)     (0.43)
10



TABLE 1: Translog Production Function Parameter Estimates (continued)
AFRICA     ASIA         EMENA       OECD        LAT
In (K)In (K)           -0.065      -0.001      -0.582      -0.012      0.296
(-1.98)    (-0.07)     (-4.13)     (-0.63)     (3.62)
In (K)ln (K1            0.374       0.009      -0.581      -0.047      -0.091
(1.59)     (0.26)      (-5.00)     (-1.84)     (-0.62)
In (K) In (K)           0.215       0.001      -0.493      -0.184      -0.398
(2.07)     (0.01)      (-4.93)     (-2.16)     (-2.62)
In (K4)in (K4)         -1.030      -0.166      -0.279      0.032       0.276
(-3.24)    (-3.51)     (-1.17)     (0.41)      (0.64)
In(K4)In (K)            0.373       0.151      0.437       -0.183      -0.374
(1.58)     (2.64)      (2.17)      (-1.07)     (-0.63)
In (K) In (K)           0.146      -0.021      -0.238      0.044       0.045
(2.16)     (-0.50)     (-2.48)     (0.45)      (0.25)
0.99       0.99        0.99        0.99        0.97
F-value                 767.6      2723.9      6262.4      16239.4      114.8
OBS                     80         160         100         80          80
NOTES: The numbers in parentheses denote t-statistics. Ki denotes the il capital stock: 1 =
labor; 2 = private capital; 3 = infrastructure capital; 4 = human resource development capital;
and 5 = military capital. As noted in the text, all models contain intercept and time trend
dummies which are suppressed here. The countries included in each of the regions are reported
in the Appendix. OBS denotes the number of observations. R2 is the adjusted R2.
11



Table 2A: Output and Scale Elasticities: Countries Arranged By Geographic Region
OUTPUT ELASTICITIES WITH RESPECT TO
LABOR     PRIVATE  INFRASTRUCTURE HUMAN RESOURCE MILITARY    SCALE
CAPITAL      CAPITAL    DEVEL. CAPITAL   CAPITAL  ELASTICITY
AFRICA
Egypt       0.12        0.33         0.03        0.75        -0.04        1.19
Kenya       0.15        0.22         0.04        0.85        -0.01        1.25
Tanzania    0.09        0.25         0.02        0.70         0.01        1.07
Zimbabwe   0.10         0.25         0.02        0.56        -0.03        0.90
ASIA
Indonesia   0.33        0.45         0.03        0.94        -0.13        1.62
India       0.30        0.35         0.02        0.57        -0.14        1.10
Korea       0.40        0.30         0.01        0.86         0.15        1.72
Sri Lanka    0.11       0.30         0.01        0.38         0.25        1.05
Malaysia    0.12        0.38         0.01        0.61         0.09        1.21
Philippines  0.20       0.30         0.01        0.60         0.04        1.15
Thailand    0.15        0.30         0.01        0.65         0.07        1.18
Pakistan    0.10        0.28         0.01        0.59        -0.02        0.96
EMENA
Greece      0.17        0.52        0.05         0.54         0.04        1.32
Iran        0.15        0.57         0.04        0.49         0.03        1.28
Israel      0.29        0.17         0.05        0.15         0.02        0.68
Portugal    0.16        0.47         0.02        0.45         0.09        1.19
Turkey      0.16        0.57         0.03        0.41         0.03        1.20
OECD
Canada      0.16        0.39         0.02        0.30         0.07        0.94
Japan       0.73        0.39         0.01        0.49        -0.01        1.61
U.S.A.      0.54        0.52        0.02         0.47        -0.09        1.46
Germany    0.46         0.44         0.02        0.41        -0.03        1.30
LAT
Bolivia     0.17        0.26         0.15        0.46        -0.01        1.03
Colombia   0.08         0.27         0.16        0.71        -0.08        1.14
Mexico      0.18        0.19         0.14        0.66        -0.03        1.14
Venezuela   0.18        0.24        0.13         0.52        -0.01        1.07
NOTES: The measures of elasticities have been calculated at sample means.
12



Table 2B: Output and Scale Elasticities -- Countries Arranged By 1990 per Capita GDP
(ICP Estimates)
OUTPUT ELASTICITIES WITH RESPECT TO
LABOR     PRIVATE  INFRASTRUCTURE HUMAN RESOURCE MILITARY    SCALE
CAPITAL      CAPITAL    DEVEL. CAPITAL   CAPITAL  ELASTICITY
PANEL A: Low Income
Tanzania    0.09        0.25        0.02        0.70         0.01        1.07
Kenya       0.15        0.22        0.04        0.85         -0.01       1.25
India       0.30        0.35        0.02        0.57         -0.14       1.10
Pakistan    0.10        0.28        0.01        0.59         -0.02       0.96
Bolivia     0.17        0.26        0.15        0.46         -0.01       1.03
Zimbabwe   0.10         0.25        0.02        0.56         -0.03       0.90
PANEL B: Medium Income
Philippines  0.20       0.30        0.01        0.60         0.04        1.15
Indonesia   0.33        0.45        0.03        0.94         -0.13       1.62
Sri Lanka    0.11       0.30        0.01        0.38         0.25        1.05
Egypt       0.12        0.33        0.03        0.75         -0.04       1.19
Iran        0.15        0.57        0.04        0.49         0.03        1.28
Thailand    0.15        0.30        0.01        0.65         0.07        1.18
Colombia   0.08         0.27        0.16        0.71        -0.08        1.14
Turkey      0.16        0.57        0.03        0.41         0.03        1.20
Malaysia    0.12        0.38        0.01        0.61         0.09        1.21
Mexico      0.18        0.19        0.14        0.66         -0.03       1.14
PANEL C: Medium to High Income
Venezuela   0.18        0.24        0.13        0.52        -0.01        1.07
Korea       0.40        0.30        0.01        0.86         0.15        1.72
Greece      0.17        0.52        0.05        0.54         0.04        1.32
Portugal    0.16        0.47        0.02        0.45         0.09        1.19
PANEL D: High Income
Israel      0.29        0.17        0.05        0.15         0.02        0.68
Germany    0.46         0.44        0.02        0.41         -0.03       1.30
Japan       0.73        0.39        0.01        0.49         -0.01       1.61
Canada      0.16        0.39        0.02        0.30         0.07        0.94
U.S.A.      0.54        0.52        0.02        0.47        -0.09        1.46
NOTES: The per capita GDP range is defined as: low is less than $US 3,000; medium is
between $US 3,000 and 6,000; medium high is between $US 6,000 and 9,000; and high is
greater than $US 9,000. The ICP measure of GDP is based on the United Nations International
Comparison program estimates.
13



REFERENCES
Antle, J. "Infrastructure and Aggregate Agricultural Productivity: International Evidence."
Economic Development and Cultural Change, (1983):609-619.
Aschauer, D.  "Is Government Spending Productive?"  Journal of Monetary Economics,
23(1989): 177-200.
Berndt, E.R. and B. Hansson. "Measuring the Contribution of Public Infrastructure Capital
in Sweden." Scandinavian Journal of Economics, v.94(1992):151-168.
Chrlistensen, L.R., D.W. Jorgenson, and L.J. Lau. "Transcendental Logarithmic Production
Frontiers. " Review of Economics and Statistics, 55(1973):28-45.
DeLong, B. and L. Sununers. "Equipment Investment, Relative Prices and Economic
Growth", mimeo, Harvard University, 1990.
Devarajan, S., V. Swaroop and H. Zou. "What Do Governments Buy?", Policy Research
Working Papers, The World Bank, WPS 1082, 1993.
Hulten, C. "Growth Accounting When Technical Change is Embodied in Capital." American
Economic Review, 82(1992):964-980.
Landau, D. "The Economic Impact of Military Expenditures". The World Bank Working Ppaer
Series NO. 1138, May 1993.
Levine, R. and D. Renelt. "A sensitivity Analysis of Cross-Country Growth Regressions."
American Economic Review, 82(1992):942-963.
Lucas, R. "On the Mechanics of Economic Development", Journal of Monetary Economics,
XXI, pp. 3-42, 1988.
Lynde, C and J. Richmond. "Public Capital and Total Factor Productivity", International
Economic Review, 34(1993),401-414.
McElroy, F.W. "Returns to Scale, Euler's Theorem and the Form of Production Functions."
Econometrica, 37(1969):275-279.
Munnell, A. "Why Has Productivity Growth Declined? Productivity and Public Investment."
New England Economic Review, Federal Reserve Bank of Boston, January/February,
1990, 3-22.
Murphy, K., A. Shielfer and R. Vishny "Industrialization and the Big Push", Journal of
Political Economy, v.97:5, pp. 103-126, 1989.
Richards, G.R. "Endogenous Technological Advance and Postwar Economic Growth: A
Production Function Analysis", Eastern Economic Journal, v. 18(1992): 315-332.
14



Shah, A. "Dynamics of Public Infrastructure, Industrial Productivity and Profitability. " Review
of Economics and Statistics, 74(1992):28-36.
Shah, A. "Public Infrastructure and Private Sector Profitability and Productivity in Mexico."
World Bank Policy, Planning and Research Working Paper No. 100, September, 1988.
Solow, R. "Technical Change and the Aggregate Production Function. " Review of Economics
and Statistics. 24(1957):312-320.
Summers, L.H. Keynote Address: Knowledge For Effective Action. Proceedings of the
World Bank Annual Conference on Development Economics, World Bank, Washington,
D.C., pp. 7-14.
Sunmers, L.H. "Keynote Address: Educating Girls." Proceedings of the 1992 Annual
Conference of the Pakistan Society of Development Economists, IslaTnabad, Pakistan.
Thomas, V. and Y. Wang. "Government Policies and Productivity Growth: Is East Asia an
Exception?" mimeo, The World Bank, 1992.
U.S. Congress, Congressional Budget Office. How Federal Spending for Infrastructure and
Other Public Investments Affects Economy. Washington, D.C. July 1991.
World Development Report. The World Bank, Oxford University Press, 1991.
15



Policy Resoarch Working Paper Series
Contact
Titbe                           Author                  Date              for paper
WPS1 164 Power, Distortions, Revolt, and  Hans P. Binswanger     July 1993          H. Binswanger
Reform In Agricutural Land Relations iKaus Deininger                      31871
Gershon Feder
WPS1 165 Social Costs of the Tranaition to  Branko Milanovic     August 1993        R. Martin
Capitalism: Poland, 1990-91                                               39026
WPS1 166 The Behavior of Russian Firms In    Simon Commander     August 1993       0. del Cid
1992: Evidence from a Survey    Leonid Liberman                           35195
Cecilia Ugaz
Ruslan Yemtsov
WPS1167 Unemployment and Labor Market    Simon Commander         August 1993        0. del Cid
Dynamics in Russia              Leonid Lberman                            35195
Ruslan Yemtsov
WPS1168 How Macroeconomic Projections    Rashid Faruqee          August 1993       N. Tannan
in Policy Framework Papers for the                                        34581
Africa Region Compare with Outcomes
WPS1169 Costs and Benefits of Debt and   Eduardo Femandez-Arias  August 1993       R. Vo
Debt Service Reduction                                                    33722
WPS1 170 Job Search by Emplyed Workers:   Avner Bar-Ilan         August 1993       D. Ballantyne
The Effects of Restrictions     Anat Levy                                 37947
WPS1 171 Finance and its Reform: Beyond  Gerard Caprib, Jr.      August 1993       P. Sintim-
Laissez-Faire                   Lawrence H. Summers                        Aboagye
38526
WPSI 172 Uberalizing Indian Agriculture:  Garry Pursil           September 1993    D. Ballantyne
An Agenda for Reform            Ashok Gulati                              37947
WPSI 173 Moroco's Free Trade Agreement with Thomas F. Rutherford  September 1993    N. Artis
the European Communitf          E. E. Rutstr8m                            38010
A Quantitative Assessment       David Tarr
WPS1 174 Asian Trade Barriers Against Primary Reed Safadi        September 1993    J. Jacobson
and Processed Commodities       Alexander Yeats                           33710
WPS1175 OECD Trade Barriers Faced by the   Bartomiel Kamnsid     September1993     J. Jacobson
Successor States of the Soviet  Alexander Yeats                           33710
Union
WPiS1176 Cash Social Transfers, Direct Taxes, Branko Mikanovic   September 1993    R. Martin
and Income Distribution In Late                                           39065
Socialism
WPS1 177 Environmental Taxes and Policies    Nell Bruce          September 1993    C. Jones
for Developing Countries        Gregory M. ElBr                           37699



Policy Research Working Paper Series
Contact
Title                           Author                  Date              for paper
WPS1 178 Productivity of Public Spending,    John Baffes         September 1993    C. Jones
Sectoral Allocation Choices, and  Anwar Shah                              37699
Economic Growth
WPS1179 How the Market Transition Affected   Bartlomiej Kaminski  September 1993   P. Kokila
Export Performance in the Central                                         33716
European Economies
IJ
-~~~   -                      C  -                  -~~~~~~~