Input–output multiplier analysis has been one of the workhorses
for ex-ante employment impact assessment. Its comprehensive
description of an economic structure enables one to map out
industry-level employment generation due to an external shock,
such as government fiscal expansion. Social accounting matrix (SAM)
as an extension of the input–output approach adds socio-economic
aspects, i.e. workers and their household characteristics, coherently
into the framework. Multiplier analysis based on SAM thus provides
an adequate simulation platform to analyze policy impacts on
disaggregated subgroups of households as well as industries. The
method however rests on the supposition that the technical coefficients
of production remain constant. Hence, modification of the
SAM is necessary if an intended simulation exercise entails, in one
form or another, a new technology requirement.
A government fiscal expansion, intended to provide jobs to a
specific subpopulation group who are otherwise marginalized in the
labor market, may require contractors to use a more labor-intensive
production technology than a conventional one would demand. This
type of intervention, in effect, changes the technology coefficients of
the recipient industry, and it should appear as a new industry in the
SAM for more accurate estimation. However, rebalancing the SAM can
be costly and sometimes does not augment any economic meanings.
This paper proposes a simple method to overcome the limitation in
context of a policy simulation for South Africa.
As is well known, South Africa is experiencing one of the highest
rates of unemployment among middle-income countries, reaching
25 to 30% over the last decade. To ameliorate the associated socioeconomic
pressures, in 2004 the government introduced a direct job
creation initiative, the Expanded Public Works Program (EPWP),
which has yielded some positive outcomes, but has been incommensurably
small to the scale of needed intervention. Currently the
scaling-up of the EPWP is under discussion and much research is
under way. It is within this context that a modeling exercise was
undertaken to assess the economy-wide, potential effect of a
substantial expansion of EPWP with focus on different labor factor
specifications that brought to light the theoretical and practical issues
discussed below.
There are four main EPWP sectors designated for job creation, one
of which is the EPWP social sector. This simulation exercise has
focused on scaling-up the home and community-based care (HCBC)
and early childhood development (ECD) programs, both of which are
part of the EPWP social sector. Besides enhancing income and
reducing unemployment, such social-sector job creation also results
in reducing women's burdens of unpaid care work. HCBC workers
perform a variety of tasks needed for the homebound and chronically
ill (including HIV/AIDS patients), while ECD workers provide support to childcare centers in tasks that range from sanitation and meal
preparation, to mental stimulation and psychological safety for
children aged 1–6 years. The original estimates on types and numbers
of proposed jobs—as well as associated implementation costs—are
from an extensive study by Friedman et al., (2007) of the Health
Systems Trust.1
EPWP consists of job opportunities provided to unskilled,
unemployed, and marginalized poor individuals who work in projects
that are labor intensive. They are hired at a minimum wage and, while
receiving training and accreditation, they provide services for their
communities. These projects are therefore not typical in comparison
to the existing South African economic structure and cannot be
represented by production conditions of similar sectors in the private
or public domain. Along with employment targeting, the effectiveness
of the program mandates that technologies be used to maximize the
labor content. Any multiplier analysis of such a program should
not rely on simulating an injection of public funds in sectors whose
production technology is not subject to this mandate. Rather, the
injection should introduce the new particularities and features of this
government intervention. Hence, the EPWP technology, represented
as more labor-intensive input composition in our study, must be
introduced anew. Moreover, job targeting requires a separate, new
account that is not governed by the existing employment distribution
structure of South Africa. Therefore, to integrate these two technical
requirements, modification of the existing SAM is required.
A simple hypothetical integration method is developed in this paper
to circumvent a rebalancing of the SAMwithout sacrificing the accuracy
ofmultiplier-effect analysis.Without a balanced SAM, simulation results
violate the accounting identity principle in that total revenue does not
equal to total expenditure for each account. The imbalance contradicts
the underlying equilibrium concept in the multiplier analysis. Many
examples of previous research that have required SAMmodification can
be found in the literature. For instance, Khan and Thorbecke, (1989)
subdivided sectors (mainly agriculture) into modern and traditional
ones to make evident technological dualism, namely the difference in
technologies used. Cella, (1984),Milana, (1985), Clements, (1990), and
Dietzenbacher et al., (1993), in order to estimate the value of a sector,
engaged in hypothetical extraction by replacing the sector's domestic
use and supply of goods with imports, thus eliminating the existing
sector's linkages to the rest of the economy. The hypothetical extraction,
combined with sectoral decomposition as in Khan and Thorbecke,
(1989), may seem an alternative way to evaluate the impact of the
proposed expansion of the public care sector.However, the unique labor
requirement of the EPWP does not resemble the existing care sectoreducation
and health care- in the SAM. Therefore, the extraction
approach based on the decomposition of the sector is not feasible.
This paper uses an integration of a new hypothetical sector, called
EPWP social sector (or EPWP in short) from an exogenous injection
into the SAM by modifying the scale of the new sector. The scalingdown
generates insignificant values for new accounts associated with
the sector and, hence, may not violate an acceptable margin of error
used in a conventional technical balancing. The insignificant values
however preserve backward linkages that generate multiplicative
effects of the intervention on the sector. The method is also flexible
enough to incorporate policy exercises (in this study, employment
targeting for the poor) into the SAM.
The usual practice of SAM rebalancing does not apply in this study,
as a prior information basis on which minimum entropy method
relies does not exist. The maximum entropy approach that does not
require the prior information could be used for rebalancing, but at the
cost of abandoning some useful prior information, such as a SAM from
a previous time period. Moreover, technical balancing without any
reference to compare before-and-after balancing (to evaluate the
success of balancing) does not yield valuable knowledge upon which
to analyze the impacts of the sector, especially when it comes to the
hypothetical sector.
The structure of this paper is as follows: Section 2 provides a
general description of the SAM structure and specific features of the
South African SAM (SAM-SA) used in this paper; Section 3 describes
the reformulation of SAM-SA for this exercise; an introduction to the
fixed-price multiplier appears in Section 4; and the results and comparative
analyses of the simulation are presented to draw attention
on the need for reformulation in Section 5.
We simulate and analyze the distributional impact of scaling-up of
social service delivery under the Expanded Public Works Program in
South Africa (EPWP). The social service includes home and communitybased
care and early childhood, and the scaling up expect to provide jobs
to unemployed workers in poverty under the auspices of the EPWP.
To depict the distributional impact, we use the Social Accounting
Matrix (SAM)-based multiplier analysis. The method elucidates the
circular flow of funds through which supply and demand of products and
factors augment income and expenditure, and in turn expand the
economy. The multiplier effect of direct job provision to the poor
alleviates high degree of income inequality from racially and spatially
biased unemployment. The positive impact however does not come
into effect in the analysis, because the SAM is a balanced matrix that
delineates only an existing flow of funds, like an accounting table.
Simulation of the policy that features a new channel of flows to the poor
requires modification of the matrix, preferably without a technical hassle.
Reformulation can be simple without compromising the balance
of the matrix. A hypothetical input composition—intensive use of
unskilled poor workers, for instance—embedded in the policy can add
to the matrix as a new sector. The hypothetical integration establishes
new channels of flow of funds to poor households from job targeting,
and thus increases their employment, and hence, income more than it
would be otherwise under the existing unequal income structure. Our
comparative analysis demonstrates the pro-poor nature of the job
targeting. The method proposed in this paper allows flexible use of the
SAM in a policy simulation exercise, in particular, for a large-scale
socio-economic policy.
