Abeberese, 2020

Journal of African Economies, 2019, 1–17
doi: 10.1093/jae/ejz013
Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Article
The Effect of Electricity Shortages on Firm

Investment: Evidence from Ghana
Ama Baafra Abeberese*
Department of Economics, Wellesley College, 106 Central Street, Wellesley 02481, USA
*Corresponding author: Ama Baafra Abeberese. E-mail: aabebere@wellesley.edu
Abstract
Access to electricity has been widely acknowledged as playing an important role in
economic growth. However, there has been relatively little research on how access to
electricity affects the decisions of firms. Electrification rates in developing countries
have increased but electricity access remains plagued by outages. This paper exam-
ines the impact of electricity shortages on firm investment. I identify this impact by
studying an electricity rationing programme in Ghana, which placed significant con-
straints on electricity supply to firms. Using data on Ghanaian manufacturing firms,
I find a decline in investment in plant and machinery during the electricity rationing
period, with a more pronounced decline for firms in electricity-intensive sectors. This
result suggests that at least part of the reduction in investment during the electricity
rationing period was due to the constraints on the availability of electricity. These find-
ings highlight the potentially negative impact of the inadequate provision of electricity
that frequently plagues developing countries. These electricity constraints can hinder
growth in these countries by curbing investment by firms.
Key words: investment, infrastructure, electricity supply, manufacturing
JEL classification: D25, H54, O13, H54, O14
1. Introduction
Lack of electricity is a pervasive problem in developing countries. An estimated 1.3 billion
people worldwide are without electricity, over 95% of whom live in developing countries.
Even in the areas where there is provision of electricity, this service is plagued by frequent
interruptions. In the 2007 World Bank Enterprise Survey, firms in Ghana, the setting for
this paper, reported an average of approximately 10 power outages in a typical month and
almost 50% of the firms surveyed ranked electricity as the most severe obstacle to invest-
ment (World Bank, 2007).1 Figure 1 shows the top ten constraints to firm investment in
1 In this survey, firms were asked to indicate which element posed the biggest constraint out of a list
of 15 elements including electricity, access to finance, corruption, an inadequately educated work-
force, labour regulations, political instability, corruption and courts.
© The Author(s) 2019. Published by Oxford University Press on behalf of the Centre for the Study of African
Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com
2 Ama Baafra Abeberese

Figure 1: Top 10 Constraints to Firm Investment in Ghana. Notes: Data are from World Bank (2007).
The numbers indicate the percentage of firms in the 2007 World Bank Enterprise Survey that selected
the indicated constraint as the main obstacle to their investment.
Ghana as reported in the survey. The goal of this paper is to determine if electricity outages
do indeed constrain firm investment by using data on manufacturing firms in Ghana and
exploiting an electricity rationing programme that took place in the country.
A growing literature within development assesses the impact of electricity provision on
firms (see, for example, Reinikka and Svensson (2002), Rud (2012a), Rud (2012b), Zuberi
(2012), Alby et al. (2013), Fisher-Vanden et al. (2015) and Allcott et al. (2016)).2 This paper
departs from the existing literature on the impact of electricity provision on firms in two ways.
First, investment has rarely been studied as an outcome variable in understanding how
electricity shortages affect firms.3 Since the low level of firm investment is often cited as one
of the causes of the slow growth of developing countries, it is important to understand how
the frequent electricity shortages in these countries affect investment by firms. Second, few
studies have examined the effect of electricity rationing programmes. Most studies make no
distinction between the types of outages (that is, scheduled versus unscheduled) or analyse
the effect of electricity outages which tend to be unexpected and intermittent in nature and
last for only a few hours at a time. Such outages are in contrast with electricity rationing
programmes which tend to be pre-announced and persist for an extended period of time,
with outages lasting for several hours at a time. These electricity rationing programmes are
widespread and can last for as long as a year or more.4
Both types of outages – scheduled (as in power rationing programmes) and unscheduled
outages – can have detrimental effects on firms but the margins of adjustment available to
firms may differ between the two. With rationing programmes, unlike with unscheduled
2 Some recent macroeconomic evidence also suggests that electricity outages in Sub-Saharan
Africa have resulted in a substantial drag on economic growth (Andersen and Dalgaard, 2013).
3 An exception is Reinikka and Svensson (2002).
4 Countries that have experienced recent episodes of electricity rationing include Bangladesh, Brazil,
Chile, China, Colombia, Ethiopia, Ghana, Kenya, Indonesia, the Philippines, South Africa, Tanzania,
Togo, Uganda and Vietnam (Maurer et al., 2005; Heffner et al., 2010).
Effect of Electricity Shortages on Firm Investment 3
outages, the timing of outages is mostly predictable and the duration of outages is much

longer allowing firms to plan in advance to re-optimise among inputs or adjust production
times. Therefore, firms may be able to avoid large impacts on their output and productivity.
For instance, Fisher-Vanden et al. (2015) find that firms in China re-optimised among
inputs in response to electricity scarcity and, hence, were able to avoid substantial product-
ivity losses. Some of the Chinese cities had power rationing programmes and, thus, the abil-
ity of firms to re-optimise among inputs may have been due to the scheduled nature of
some of the outages. Although firms may be able to avoid large output and productivity
losses in the short term with rationing programmes, these programmes may affect their
investment behaviour, which may have repercussions for their long-term growth, as re-
optimisation among inputs will imply a movement away from electricity-dependent capital
towards less electricity-dependent inputs.
The electricity rationing programme I study in this paper occurred in Ghana starting in
late 1997. Under this programme, consumers were supplied with electricity for only a por-
tion of the day. The onset of this rationing programme was unexpected and there was
uncertainty about when it would come to an end. The rationing programme therefore had
the potential to influence the long-term outlook of firms and, hence, their investment deci-
sions. Using data on Ghanaian manufacturing firms, I find that investment in plant and
machinery was halved for the average firm during the rationing period. A challenge with
interpreting this result as being attributable to electricity constraints is that other factors
besides the availability of electricity may have changed during the rationing period. I find
that the decline in investment was more pronounced for firms in electricity-intensive sectors,
suggesting that, at least part of the reduction in investment was driven by the constraints
on the availability of electricity during the rationing period. I also perform robustness tests
to explore other factors that may have potentially contributed to the fall in investment but
find that they cannot fully explain the fall in investment.
Further, while the electricity rationing programme officially ended in 1998, I find that
firms in electricity-intensive industries continued to experience depressed levels of invest-
ment after 1998, suggesting that such crises may have long-lasting effects. This inertia in
investment may be explained by the theoretical literature on investment under uncertainty
which suggests that, given the irreversibility of investment, firms may not resume invest-
ment until they can ascertain that a change in economic conditions is not transitory. Given
the unexpected onset of the rationing programme, it is possible that firms may have been
unsure if the conclusion of the programme was permanent.
In line with evidence in recent studies (see, for example, Fisher-Vanden et al. (2015) and
Allcott et al. (2016)), I also find that firms did not experience significant productivity losses
as a result of the electricity shortages. This is consistent with the argument that firms may
be able to re-optimise among inputs and, hence, avoid productivity losses. However, as dis-
cussed above, if this re-optimisation implies a movement away from electricity-dependent
capital towards less electricity-dependent inputs, as suggested by the fall in investment, this
re-optimisation may have implications for firms’ long-term growth. In fact, I find some sug-
gestive evidence that electricity-dependent firms were less likely to reduce their use of labour
relative to other firms during the crisis, reinforcing the argument that firms may have
moved away from electricity-dependent capital to less electricity-dependent inputs. I also
find that electricity-dependent firms experienced slower increases in growth relative to other
firms during the crisis.
The results of the paper indicate that, although firms may be able to avoid immediate

output and productivity losses by re-optimising among inputs, prolonged interruptions to
electricity supply can be detrimental for firm investment and have persistent effects. A cav-
eat here is that, since the analysis is based on an electricity rationing programme, it is not
immediately evident that occasional power outages would have the same effect on invest-
ment. Nonetheless, the results suggest that increasing electricity access by reducing the inci-
dence of electricity rationing programmes, which have become increasingly common in
developing countries, can help increase firm investment rates.
The rest of the paper is organised as follows. Section 2 provides a brief description of
the electricity crisis in Ghana. Section 3 briefly discusses how the theoretical literature on
the response of investment to shocks applies to a setting with electricity shortages.
Section 4 describes the data and provides some summary statistics. Section 5 outlines the
empirical strategy and presents the results. Section 6 concludes.
2. Ghana’s electricity crisis

The main source of electrical power in Ghana, until the late 1990s, was from government-
owned hydro-electric power plants at the Akosombo and the Kpong dams, located on the
Volta River in the southeastern part of the country. Together, they provided 1,072 MW of
installed generation capacity.5 Total domestic consumption was steady at about 350 MW
until the early 1990s when the government embarked on nationwide electrification projects
aimed at connecting more communities to the national electricity grid. By 1997, annual
demand for electrical consumption was growing at 15% per year. Of the total electricity
output, about one-third was consumed by aluminium smelter, Volta Aluminum Company
(VALCO), based on existing agreements with the government. Another 20% was con-
sumed by the mining industry, leaving half the production capacity for residential, non-
residential and industrial use.
Currently, an estimated 79% of Ghanaians have access to grid electricity.6 All the
regional capitals are connected to the public power grid. The firms in the dataset used in
this study are all located in Accra, Kumasi, Takoradi and Cape Coast, which are all
regional capitals.
Starting in late 1997 there was a severe drought in Ghana which caused the water level
in the Akosombo reservoir to fall below the levels needed to generate enough electricity to
meet the increasing demand. Between 1997 and 1998, annual rainfall for the country fell
by about 20% (University of East Anglia, 2013). The water volume of the Akosombo reser-
voir fell by about 18% in this same period (Volta River Authority, 1997, 1998). In the first
quarter of 1998, total output from the hydro power plant and other sources had fallen by
about 40% to 600 MW. The drought and increased demand for electricity, coupled with
the government’ s failure to retrofit existing power plants to produce at near-installed cap-
acities, led to a severe energy crisis.
The government subsequently embarked on a nationwide electricity rationing pro-
gramme, which started in September 1997, to avoid a complete shutdown of the
5 See Ghana’ s Volta River Authority at http://www.vra.com/ (last accessed March 21, 2016).
6 World Development Indicators, The World Bank (available at http://data.worldbank.org/data-
catalog/world-development-indicators) (last accessed June 20, 2018).

Figure 2: Electricity Consumption and Production (kWh per capita). Notes: Data are from the World
Bank’s World Development Indicators database available at http://data.worldbank.org/data-catalog/
world-development-indicators (last accessed April 21, 2013). The electricity production figures
represent total gross electricity production as measured at the power plants and do not exclude
transmission and distribution losses or the amount of electricity used in the power plants. Ghana
also imports electricity from neighbouring countries, causing electricity consumption to exceed pro-
duction in some years.
Akosombo hydro-electric plant. Initially, the rationing schedule was such that households
would be provided with 24 h of electricity every other day. Industries would receive three con-
tinuous days of power supply and one day of no power. These schedules were later redesigned
to be 12 h with power and 12 h without power each day for both households and industrial
consumers. With these schedules, consumers would have access to electricity from 6 a.m. to
6 p.m. and then no access from 6 p.m. to 6 a.m. on some days. On other days they would
have no access from 6 a.m. to 6 p.m. and have access from 6 p.m. to 6 a.m.
The beginning of this rationing programme was largely unexpected. There was also uncer-
tainty regarding when the programme would be terminated. Although, the government offi-
cially announced the end of the electricity rationing programme in November 1998, due to
erratic rainfall, there were still occasional periods of electricity rationing as of 2002, though
none were as prolonged as the electricity rationing programme in 1998 (Government of
Ghana, 2001, 2002). Figure 2 plots the production and consumption of electricity per capita
in Ghana over the period 1991 to 2002.7 The figure shows a clear dip in both electricity pro-
duction and consumption in 1998. Prior to 1998, electricity consumption had been close to
400 kilowatt-hours per capita, but had dropped by about 30% to a low of 280 kilowatt-
hours per capita in 1998 and had not returned to the pre-crisis level by 2002.
3. Theories of the response of investment to shocks

Standard economic theory predicts that firms find it profitable to invest if the marginal
product of capital exceeds the cost of capital. Therefore, all else equal, a reduction in the
7 Note that the electricity production figures represent total gross electricity production as measured
at the power plants and do not exclude transmission and distribution losses or the amount of elec-
tricity used in the power plants.
marginal product of capital is predicted to lead to a fall in investment. As electricity is a

critical input for most forms of capital, a reduction in electricity supply induced by the elec-
tricity rationing programme would be expected to reduce the productivity of electricity-
dependent capital and, hence, the incentive to invest.
In addition, the uncertainty about when the electricity rationing programme would end
could further influence the incentive of firms to invest. A large literature spurred by earlier
work by Bernanke (1983) and Pindyck (1991) has focused on uncertainty as an important
determinant of investment. Two characteristics of investment are emphasised in models of
investment under uncertainty. One, investments tend to be irreversible because of sunk costs
and, two, investments can be delayed. If firms delay investment, they lose the potential returns
from investing today. However, if firms invest today, they give up the option of waiting for
new information about economic conditions that can allow them to make better-informed
decisions about the desirability and timing of investment. Thus, these models predict that,
given the irreversibility of and potential to delay investment, investment can fall if uncertainty
about the future returns to investment is high (see Carruth et al. (2000) for a review).
In the context of energy, recent models (see Elder and Serletis (2010) for a review) show
that uncertainty about oil prices will tend to reduce investment. Analogously, firms facing
electricity outages can be considered to be facing an infinite electricity price. Therefore,
uncertainty about electricity outages, and in effect, electricity prices, could adversely affect
firm investment.
Further, models on investment under uncertainty suggest that there may be optimal iner-
tia in investment in that, even when the return to investment is much higher than the cost
of capital, it may be optimal for firms to continue to wait to ensure that the current condi-
tions are not transitory before resuming investment (see Dixit, 1992).
In the empirical analysis below, I examine whether, as predicted by these theories, firm
investment responded negatively to electricity rationing and whether this response extended
beyond the official end of the rationing period.
4. Data
The results in this paper are based on data from the Ghanaian Manufacturing Enterprise
Survey (GMES), which covers the period 1991 to 2002.8 This dataset contains unbalanced
panel data on a sample of firms in the Ghanaian manufacturing sector. The earlier rounds
of the survey were part of the World Bank’s Regional Programme on Enterprise
Development (RPED). The later rounds were conducted by the University of Oxford
together with the University of Ghana and the Ghana Statistical Office.
The original sample of firms, first surveyed in 1992, was randomly drawn from Ghana’ s
1987 Census of Manufacturing Firms. The sample was chosen so as to be representative of
the size distribution of firms across the major sectors of the Ghanaian manufacturing indus-
try. The sectors include the food, beverages, bakery, textile, garment, wood, furniture,
machinery and metal sectors. The firm size ranges from less than five employees to over
8 The dataset is made available by the Center for the Study of African Economies (CSAE) at the
University of Oxford. Financing for the surveys was from the Overseas Development Administration
(now the Department for International Development). The dataset and the questionnaires used for
the survey are available at https://www.csae.ox.ac.uk/data (last accessed June 20, 2018).
Table 1: Summary Statistics by Investing Status

Firms with zero investment Firms with non-zero investment All firms
Investment rate 0.07 0.03

<0.01> <0.002>
Output (millions of cedis) 129.0 617.0 340.0
<20.4> <64.2> <30.6>
Number of workers 41.1 97.0 65.2
<5.3> <5.3> <2.8>
Weekly hours per worker 46.4 45.9 46.2
<0.3> <0.3> <0.2>
Labour productivity 2.1 3.7 2.8
<0.1> <0.2> <0.1>
log(Total Factor 10.9 11.7 11.2
Productivity) <0.04> <0.06> <0.04>
Electricity Intensity 0.014 0.014 0.014
<0.0003> <0.0003> <0.0002>
Number of observations1 968 739 1,707
Note: Standard errors of the means are in parentheses. All monetary values are in 1991 cedis. The investment
rate is defined as the ratio of a firm’s investment in plant and machinery to its output. Total factor productivity
is calculated using the approach in Levinsohn and Petrin (2003). Electricity intensity is defined as the average
ratio of electricity costs to output value for firms in a sector.
1
The number of observations for firms with non-zero investment is 737 for number of workers and labour
productivity. The numbers of observations for firms with zero investment and non-zero investment are 864
and 704, respectively, for weekly hours per worker and 725 and 547, respectively, for log(total factor
productivity).
1,000 employees. The firms are located in Accra, Kumasi, Takoradi and Cape Coast. Firms
that exited were replaced by firms of similar size and in the same sector and location.9
Approximately 200 firms were sampled in each year. The firm-level variables available in
the dataset include output, investment, capital stock and electricity expenditure.10 To reduce
the influence of outliers, I ‘winsorize’ the firm-level variables within each year by setting values
below the 1st percentile to the value at the 1st percentile and values above the 99th percentile
to the value at the 99th percentile.11 All monetary values are deflated to 1991 cedis.
Table 1 reports some summary statistics for the period 1991 to 2002 for the firms in the
dataset.12 As is typical in most developing countries, investment rates are low. On average,
only about 40% of firms invested in plant and machinery in a given year. The average
9 Appendix Table A1 shows the number of firms in each sector by year.

10 The data on electricity expenditure are available from 1992 onwards.
11 Winsorizing affects 30 observations, which represent 1.8% of the total number of observations.
Winsorizing the data does not change the conclusions. The coefficients from the raw data are
slightly larger in magnitude but there are no differences in statistical significance.
12 Total factor productivity is calculated using the approach in Levinsohn and Petrin (2003) which
addresses the endogeneity of input choices by using the firm’s raw materials as a proxy for its
unobserved productivity shock. I implement this approach using the levpet command in Stata.
investment rate (defined as the ratio of a firm’s investment in plant and machinery to its

output) was 0.03. For those firms that invested, the average investment rate was 0.07.
5. Econometric analysis
5.1 Empirical strategy
I identify the effect of the availability of electricity supply from the public grid by looking
at how firm investment changed in the electricity crisis period relative to the other periods.
The main estimating equation is
investmentrateijct = β0 + β1 T + λ i + δc t + πj t + εijct (1)
where i, j, c and t index firm, sector, city and year, respectively. The investment rate is
defined as the ratio of a firm’s investment in plant and machinery to its output. T is a dum-
my variable equal to one if the year is greater than or equal to 1998, corresponding to the
electricity crisis period, and zero otherwise. I include firm fixed effects, λi, to capture time-
invariant firm characteristics that may affect investment decisions. I also include city time
trends, δct, and sector time trends, πjt, to account for unobserved differential trends across
cities and sectors, respectively. To allow for serial correlation within firms, I cluster stand-
ard errors by firm.
In the empirical analysis below, to determine if investment continued to be affected after
the official end of the rationing programme in November 1998, I replace the dummy vari-
able, T, with two dummy variables: a dummy variable equal to one in 1998 and a dummy
variable equal to one after 1998.
A concern with the empirical strategy is that there could have been other events in the
crisis period, unrelated to electricity supply, that influenced firm investment. However, if
any changes in investment during the crisis period are indeed due to the constraints on elec-
tricity supply, then these changes should be more evident for firms in sectors that are most
reliant on electricity. To test this, I estimate the following equation which includes an inter-
action between the crisis period dummy variable and the electricity intensity of the firm’s
sector.
investmentrateijct = α0 + α1 T + α2 T * log (electricityintensity )j + λ i + δc t + πj t + εijct

(2)
I define the electricity intensity of a sector as the average ratio of electricity expenditure
to output for firms in that sector.13 The coefficient α1 may capture effects from conditions
during the crisis period other than the electricity shortage. Therefore, to isolate the impact
of the electricity shortage, the coefficient of interest is α2, which is the estimate of the impact
of the electricity crisis on investment for firms in the sectors that are most reliant on
electricity.
13 I define the electricity intensity using data from 1992, the first year for which data on electricity
expenditure are available, to avoid confounding effects from endogenous changes in sectors’
electricity intensities over time. For the beverages sector, I use data from 1996 since this is the
first year in which data on firms in this sector appear in the dataset.

Figure 3: Investment Rate. Notes: The investment rate is defined as the ratio of a firm’s investment in
plant and machinery to its output.
5.2 Effect of electricity rationing on investment

I first show graphically how investment changed over the period. Figure 3 plots the invest-
ment rate over time and shows a sharp decline in the investment rate in 1998. The invest-
ment rate following the crisis remained much lower than the rate before. Table 2 presents
estimates of equations (1) and (2). Column 1 of Table 2 reports estimates of equation (1)
excluding city and sector time trends. The coefficient on the crisis period dummy is negative
suggesting that firm investment fell during the crisis period. The average firm investment
rate is 0.03. Therefore, the coefficient of negative 0.014 indicates that investment fell by
almost half during the rationing period. This negative coefficient is robust to the introduc-
tion of city and sector time trends in Column 2 of Table 2. In Column 3, I also control for
GDP per capita growth rate and unemployment rate14 to capture general economic condi-
tions that could affect investment. The coefficient on the crisis period dummy remains
essentially unchanged.
The estimated fifty percent decline in investment is economically large. It should how-
ever be noted that, as discussed in Section 5.1, the decline in investment during the crisis
period may be partially attributable to changes in factors other than electricity supply. It is
plausible that part of this decline was, for instance, driven by the drought that induced the
electricity crisis. This estimate of the decline in investment should, therefore, be interpreted
as an upper bound of the effect of the electricity crisis on investment. In the subsequent ana-
lyses in this section and the sections below, I argue that part of the decline in investment
can be attributed to the electricity crisis itself, that is, that the lower bound of the magni-
tude of the effect of the electricity crisis on investment is greater than zero.
If the drop in investment in the crisis period is due to electricity constraints, then we
should observe a larger decline for sectors with higher electricity dependence. To investigate
14 The data on GDP per capita growth rate and unemployment rate are from the World Bank’s World
Development Indicators database (http://databank.worldbank.org/data/reports.aspx?source=world-
development-indicators), accessed on June 20, 2018.
Table 2: Effect on Investment

(1) (2) (3) (4) (5) (6)
Dep. Var.: investment rate
Crisis dummy −0.014*** −0.016* −0.015* −0.015*** −0.017* −0.015*

[0.005] [0.009] [0.009] [0.005] [0.009] [0.009]
Crisis dummy*log(sector −0.017*** −0.026** −0.026**
electricity intensity) [0.006] [0.010] [0.010]
No. of observations 1,707 1,707 1,707 1,707 1,707 1,707
No. of firms 207 207 207 207 207 207
Firm effects x x x x x x
City time trend x x x x
Sector time trend x x x x
Controls x x
Note: This table reports the coefficients from regressions for investment rate. The investment rate is defined as
the ratio of a firm’s investment in plant and machinery to its output. The crisis dummy is a variable equal to
one if the year is greater than or equal to 1998 and zero otherwise. Sector electricity intensity is defined as the
average ratio of electricity costs to output value for firms in that sector and has been deviated from the overall
mean. Control variables include GDP per capita growth rate and unemployment rate. Robust standard errors,
in square brackets, are clustered at the firm level. *** indicates statistical significance at the 1% level, ** at the
5% level, and * at the 10% level.
this, I run regressions which include an interaction between the crisis period dummy and
the log of the electricity intensity of the firm’s sector. The results from these regressions are
reported in Columns 4 through 6 of Table 2. The log of sector electricity intensity has been
deviated from the overall mean. Therefore, the coefficient on the uninteracted term repre-
sents the change in investment for a firm in a sector with average electricity intensity. In
line with the hypothesis that the fall in investment during the crisis period was a result of
the electricity crisis, the coefficient on the interaction between the crisis period dummy and
the log of sector electricity intensity is negative and statistically significant. Firms in more
electricity-intensive sectors had greater reductions in investment during the crisis period.
The average pre-crisis investment rate of the most electricity-intensive industry, wood, was
0.07 and that of the industry with electricity intensity closest to the overall average, furni-
ture, was 0.04. The coefficients in Column 6 of Table 2 imply that, as a result of the crisis
period, the investment rate fell to about 0.027 for wood and to about 0.029 for furniture,
eliminating the pre-crisis gap in investment between these sectors. In Table 3, I present esti-
mates from regressions where the crisis dummy is interacted with a dummy variable for
each sector. The excluded sector is beverages. In line with the results in Table 2, the decline
in investment is largest for the wood and food sectors, which are the sectors with the high-
est electricity intensities. In contrast, the coefficient is least negative for the bakery sector,
which has the lowest electricity intensity.
As described in Section 2, the electricity rationing programme was officially in place till
November 1998, when the government announced that it had come to an end. However,
as of 2002, there was still intermittent electricity rationing, though not as prolonged as the
rationing programme that had been in place in 1998. As discussed in Section 3, if firms
were uncertain that the electricity rationing programme had permanently ended, they may
Table 3: Effect on Investment by Sector

(1) (2)
Crisis dummy*wood −0.036** −0.035**

[0.017] [0.017]
Crisis dummy*food −0.058** −0.056**
[0.023] [0.024]
Crisis dummy*garment −0.011 −0.01
[0.010] [0.010]
Crisis dummy*furniture −0.014 −0.013
[0.012] [0.012]
Crisis dummy*machinery −0.005 −0.004
[0.023] [0.023]
Crisis dummy*textile 0.008 0.009
[0.009] [0.009]
Crisis dummy*metal 0.005 0.006
[0.013] [0.013]
Crisis dummy*bakery 0.014 0.015
[0.013] [0.013]
No. of observations 1,707 1,707
No. of firms 207 207
Firm effects x x
City time trend x x
Controls x
the ratio of a firm’s investment in plant and machinery to its output. The crisis dummy is a variable equal to
one if the year is greater than or equal to 1998 and zero otherwise. Control variables include GDP per capita
growth rate and unemployment rate. Robust standard errors, in square brackets, are clustered at the firm level.
*** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.
have continued to postpone investment even after 1998. To assess this, I estimate equa-
tions (1) and (2) with the crisis dummy replaced by a dummy variable equal to one in 1998
(during dummy) and a dummy variable equal to one after 1998 (post dummy). The esti-
mates are presented in Table 4. Columns 1 and 2 indicate a statistically significant decline
in investment in 1998 but not afterward. These estimates, however, mask variation across
sectors. The estimates in Columns 3 and 4 indicate that firms in electricity-intensive sectors
continued to experience depressed levels of investment even after 1998, which is consistent
with the theoretical literature on investment under uncertainty. Thus, the electricity ration-
ing programme had a persistent negative effect on investment for firms that were most reli-
ant on electricity.15
15 Since access to credit can help firms fund investment, and in particular, generators, a question
that may arise is whether the investment behaviour of credit-constrained firms was differentially
affected during the crisis. Using data from the survey, I define a firm as credit-constrained if it had
a liquidity problem but did not get a loan from either an informal or a formal source (either
because the firm’s loan application was rejected or because the firm did not apply for a loan for
Table 4: Effect on Investment – During and Post 1998

(1) (2) (3) (4)
During dummy −0.016* −0.016* −0.016* −0.016*

[0.009] [0.009] [0.009] [0.009]
Post dummy −0.017 −0.015 −0.017 −0.015
[0.011] [0.011] [0.011] [0.011]
During dummy*log(sector electricity intensity) −0.027*** −0.027***
[0.009] [0.009]
Post dummy*log(sector electricity intensity) −0.025** −0.024*
[0.012] [0.013]
No. of observations 1,707 1,707 1,707 1,707
No. of firms 207 207 207 207
Firm effects x x x x
Controls x x
Note: This table reports the coefficients from regressions for investment rate. The investment rate is defined as the
ratio of a firm’s investment in plant and machinery to its output. The during dummy is a variable equal to one if
the year is equal to 1998 and zero otherwise. The post dummy is a variable equal to one if the year is greater than
or equal to 1999 and zero otherwise. Sector electricity intensity is defined as the average ratio of electricity costs to
output value for firms in that sector and has been deviated from the overall mean. Control variables include GDP
per capita growth rate and unemployment rate. Robust standard errors, in square brackets, are clustered at the
firm level. *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.
5.3 Effect of electricity rationing on other outcomes

In this section, I investigate whether the electricity rationing programme affected firms
along other dimensions. The results are presented in Table 5. In Column 1, there is no evi-
dence of a decline in output as a result of the crisis. In Column 2, there appears to be a
reduction in labour but not more so for firms in electricity-intensive industries. In Column 3,
the dependent variable is the average number of hours worked per worker in a week. The
estimates for this variable, although only marginally significant, suggest that electricity-
dependent firms relatively increased the amount of hours of labour used. This would be con-
sistent with a scenario in which labour is a substitute for capital and firms re-optimise among
inputs by shifting from electricity-dependent capital towards labour.
The estimates in Columns 4 and 5 of Table 5 indicate that there were no significant
impacts on labour productivity (defined as output per worker) and total factor productivity.
As discussed in Section 1, in an environment with electricity rationing where shortages are
expected and are of long duration, firms may re-optimise among inputs and, hence, may be
able to avoid substantial productivity losses from electricity shortages. The results for output
and productivity are in line with this hypothesis and are consistent with similar evidence in
recent papers (see, for example, Fisher-Vanden et al. (2015) and Allcott et al. (2016)).
reasons other than not needing one) in a baseline year, 1995. I find that credit-constrained firms
were not differentially affected during the crisis.
Table 5: Effect on Other Outcomes

(1) (2) (3) (4) (5) (6)
Dependent log log log(average log(labour log(total factor output
variables (output) (workers) weekly hours) productivity) productivity) growth
Crisis dummy −0.082 −0.148*** −0.038 0.066 0.05 0.283***

[0.069] [0.040] [0.024] [0.072] [0.072] [0.040]
Crisis dummy* −0.0002 −0.024 0.069* 0.022 0.024 −0.100**
log(sector [0.078] [0.044] [0.036] [0.078] [0.102] [0.043]
electricity
intensity)
No. of 1,707 1,705 1,564 1,705 1,266 1,475
observations
No. of firms 207 207 199 207 175 189
City time trend x x x x x x
Sector time trend x x x x x x
Note: This table reports the coefficients from regressions for output, labour, productivity and growth. The cri-
sis dummy is a variable equal to one if the year is greater than or equal to 1998 and zero otherwise. Sector elec-
tricity intensity is defined as the average ratio of electricity costs to output value for firms in that sector and has
been deviated from the overall mean. Total factor productivity is calculated using the approach in Levinsohn
and Petrin (2003). *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10%
level.
In Column 6 of Table 5, I examine the impact of the electricity crisis on output

growth. Interestingly, while firms were able to increase growth, electricity-intensive firms
were less able to do so as indicated by the statistically significant negative coefficient on
the interaction between the crisis period dummy variable and the electricity intensity of
the firm’s sector. These results suggest that while electricity-dependent firms were able to
avoid immediate output and productivity losses, the reduction in investment had a nega-
tive impact on growth. Table 6 presents results repeating the analysis from Table 5 with
controls for GDP per capita growth rate and unemployment rate. The conclusions
remain the same.
5.4 Alternative hypotheses and robustness checks

The analysis above provides evidence of a fall in investment during the crisis period.
However, one might argue that there may have been several factors during the crisis period,
besides the electricity shortage, which may have driven the observed reduction in invest-
ment. The estimate of the coefficient on the interaction between the crisis period dummy
and the electricity intensity of the firm’s sector indicates that this reduction was more acute
for firms in electricity-intensive sectors, suggesting that part of the observed effect on invest-
ment was due to constraints on the availability of electricity. Nonetheless, I perform further
analyses in this section to verify that the electricity constraint was one of the underlying
causes of the fall in investment.
The fact that the electricity crisis was induced by a drought naturally raises the concern
that the drought itself may have had effects, besides the electricity shortage, which could
have resulted in a change in the investment behaviour of firms. For instance, the drought
Table 6: Effect on Other Outcomes (with Controls)

(1) (2) (3) (4) (5) (6)
Dependent log log log(average log(labour log(total factor output
variables (output) (workers) weekly hours) productivity) productivity) growth
Crisis dummy −0.076 −0.141*** −0.048** 0.064 0.058 0.313***

[0.070] [0.042] [0.024] [0.074] [0.075] [0.042]
Crisis dummy* −0.004 −0.025 0.068* 0.019 0.025 −0.095**
log(sector [0.078] [0.044] [0.036] [0.078] [0.101] [0.042]
electricity
intensity)
No. of 1,707 1,705 1,564 1,705 1,266 1,475
observations
No. of firms 207 207 199 207 175 189
City time trend x x x x x x
Sector time trend x x x x x x
Controls x x x x x x
Note: This table reports the coefficients from regressions for output, labour, productivity and growth. The cri-
sis dummy is a variable equal to one if the year is greater than or equal to 1998 and zero otherwise. Sector elec-
tricity intensity is defined as the average ratio of electricity costs to output value for firms in that sector and has
been deviated from the overall mean. Total factor productivity is calculated using the approach in Levinsohn
and Petrin (2003). Control variables include GDP per capita growth rate and unemployment rate. *** indi-
cates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.
may have led to a fall in household incomes and, hence, in demand for firms’ products, espe-
cially since a significant percentage of employment in Ghana is in the agricultural sector
(approximately 55% in 1998–1999 (Ghana Statistical Service, 2000)). In fact, however, the
country did not experience a recession during the electricity crisis due to export-led growth dri-
ven by high world prices for cocoa and high export volumes for gold, the top export earners
for the country (Government of Ghana, 1999). Value-added in agriculture continued to grow
at a rate of 5.1% in 1998 (Choudhary et al., 2016). Nonetheless, one might expect that any
contraction of agricultural output could have directly affected firms that relied on agricultural
inputs irrespective of limitations on electricity supply. However, the following arguments pro-
vide some support for the hypothesis that the fall in firm investment was driven in part by the
electricity shortage faced by firms rather than by only other potential effects of the drought.
First, if the fall in investment was only due to some impact of the drought unrelated to
electricity shortages, it is less plausible that the most affected firms would have been
electricity-intensive firms, as the results show.
Second, I explore the idea that the results may be driven by some contraction of agricul-
tural output by redoing the analysis with a sample that excludes firms who are most likely
to be reliant on agricultural inputs. The firms I exclude are those in the food, beverages,
and bakery sectors. The results from the analysis with this restricted sample are reported in
Table 7. In Columns 3 and 4 of Table 7, the coefficients on the interaction between the cri-
sis period dummy and the log of sector electricity intensity are negative and statistically sig-
nificant, implying a fall in investment for firms in electricity-intensive sectors. Thus, even
with a sample that excludes firms who are most likely to be reliant on agricultural inputs,
Table 7: Effect on Investment – Excluding Food Sectors

(1) (2) (3) (4)
Crisis dummy −0.008 −0.008 −0.004 −0.004

[0.009] [0.010] [0.009] [0.009]
Crisis dummy*log(sector electricity intensity) −0.048** −0.048**
[0.019] [0.020]
No. of firms 159 159 159 159
Controls x x
Note: This table reports the coefficients from regressions for investment rate for a sample excluding firms in
the food, beverages and bakery sectors. The investment rate is defined as the ratio of a firm’s investment in plant
and machinery to its output. The crisis dummy is a variable equal to one if the year is greater than or equal to
1998 and zero otherwise. Sector electricity intensity is defined as the average ratio of electricity costs to output
value for firms in that sector and has been deviated from the overall mean. Control variables include GDP per
capita growth rate and unemployment rate. Robust standard errors, in square brackets, are clustered at the firm
level. *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.
there is still evidence that the reduction in investment during the power crisis period was
concentrated among firms in electricity-intensive sectors.
Third, there was an even more severe drought in Ghana in 1992. Between 1991 and
1992, annual rainfall for the country fell by about 30% (University of East Anglia, 2013).
Value-added in agriculture contracted by 1.2%, whereas in 1998 value-added in agriculture
continued to grow at a rate of 5.1% (Choudhary et al., 2016). This drought, however, was
not accompanied by an electricity crisis. Therefore, if the results above are driven in part by
the electricity crisis rather than the only the drought, we should not observe the decline in
investment seen above in 1992. In Table 8, I present results from regressions that assume
that 1992 was the crisis period. I observe no statistically significant impacts on investment
suggesting that the results seen above were driven in part by the electricity crisis.
Finally, to my knowledge, the only major change in the economic environment in
Ghana during the crisis period, other than the electricity crisis, was the reintroduction of
the value-added tax (VAT) in December 1998 to increase the tax base and government rev-
enue. This reintroduction followed a failed initial attempt in 1995. The initial introduction
failed mainly because the public had not been adequately informed and educated about the
new tax system, resulting in public riots, and because of delays in the passage of legislation
related to the VAT. To avoid another failure, prior to the reimplementation of the VAT in
December 1998, there were intensive and comprehensive education campaigns. Also, the
legislation was enacted 10 months before implementation (World Bank, 2001). The reintro-
duction of the VAT in 1998 was widely expected and, hence, was not a sudden shock. The
reimplementation of this tax system should therefore not cause the decline in investment
that is observed in the results. In any case, even if one were to argue that the VAT was in
some way responsible for the drop in investment, it is implausible that it should have had a
stronger effect on firms in more electricity-intensive sectors.
Table 8: Effect on Investment – Placebo with 1992 as Crisis Period

(1) (2) (3) (4)
1992 dummy 0.015 0.017 0.014 0.016

[0.011] [0.011] [0.012] [0.012]
1992 dummy*log(sector electricity intensity) −0.002 −0.002
[0.010] [0.010]
No. of firms 207 207 207 207
Controls x x
the ratio of a firm’s investment in plant and machinery to its output. Sector electricity intensity is defined as the
average ratio of electricity costs to output value for firms in that sector and has been deviated from the overall
mean. Control variables include GDP per capita growth rate and unemployment rate. Robust standard errors,
in square brackets, are clustered at the firm level. *** indicates statistical significance at the 1% level, ** at the
5% level, and * at the 10% level.
6. Conclusion
This paper uses data on Ghanaian manufacturing firms to examine the effect of electricity
constraints on firm investment. Given the pervasiveness of inadequate access to electricity
in developing countries, it is important to understand how this constraint can affect firms
and, hence, growth. I provide empirical evidence that investment in plant and machinery
fell for firms in more electricity-intensive sectors during an electricity crisis that took place
in Ghana starting in late 1997. This result is in line with the argument that the restrictions
on the availability of electricity during the electricity crisis led firms to curtail their
investment.
I also find that, while firms were able to avoid significant output and productivity losses,
firms in electricity-dependent sectors were less able to increase growth during the electricity
crisis.
The findings of this paper are indicative of a potential role of infrastructure constraints
in impeding growth in developing countries. In particular, electricity outages, which are
common in developing countries, may hamper firm investment and prevent these countries
from spurring economic growth through private sector investment.
Supplementary material
Supplementary material is available at Journal of African Economies online.
References
Alby P., Dethier J.-J., Straub S. (2013) ‘Firms Operating under Electricity Constraints in
Developing Countries’, World Bank Economic Review, 27 (1): 109–32.
Allcott H., Collard-Wexler A., O’Connell S. D. (2016) ‘How Do Electricity Shortages Affect

Industry? Evidence from India’, American Economic Review, 106 (3): 587–624.
Andersen T. B., Dalgaard C.-J. (2013) ‘Power Outages and Economic Growth in Africa’, Energy
Economics, 38 (2013): 19–23.
Bernanke B. (1983) ‘Irreversibility, Uncertainty, and Cyclical Investment’, Quarterly Journal of
Economics, 98 (1): 85–106.
Carruth A., Dickerson A., Henley A. (2000) ‘What Do We Know About Investment Under
Uncertainty?’, Journal of Economic Surveys, 14 (2): 119–54.
Choudhary V., Christienson G., D’Alessandro S. P., Josserand H. P. (2016) Ghana Agricultural
Sector Risk Assessment. Washington, DC: World Bank.
Dixit A. (1992) ‘Investment and Hysteresis’, Journal of Economic Perspectives, 6 (1): 107–32.
Elder J., Serletis A. (2010) ‘Oil Price Uncertainty’, Journal of Money, Credit and Banking, 42 (6):
1137–59.
Fisher-Vanden K., Mansur E. T., Wang Q. J. (2015) ‘Electricity Shortages and Firm Productivity:
Evidence from China’s Industrial Firms’, Journal of Development Economics, 114 (2015):
172–88.
Ghana Statistical Service (2000) Ghana Living Standards Survey: Report of the Fourth Round
(GLSS 4). Accra: Ghana Statistical Service.
Government of Ghana (1999) The Budget Statement and Economic Policy of the Government of
Ghana. Accra: Government of Ghana, Ministry of Finance.
Government of Ghana (2001) Public Utilities Regulatory Commission Annual Report 2001.
Accra: Government of Ghana, Public Utilities Regulatory Commission.
Government of Ghana (2002) Public Utilities Regulatory Commission Annual Report 2002.
Accra: Government of Ghana, Public Utilities Regulatory Commission.
Heffner G., Maurer L., Sarkar A., Wang X. (2010) ‘Minding the Gap: World Bank’ s Assistance
to Power Shortage Mitigation in the Developing World’, Energy, 35 (4): 1584–91.
Levinsohn J., Petrin A. 2003. ‘Estimating Production Functions Using Inputs to Control for
Unobservables’, Review of Economic Studies, 70 (2): 317–14.
Maurer L., Pereira M., Rosenblatt J. 2005. ‘Implementing Power Rationing in a Sensible Way:
Lessons Learned and International Best Practices.’ ESMAP-World Bank.
Pindyck R. (1991) ‘Irreversibility, Uncertainty, and Investment’, Journal of Economic Literature,
29 (3): 1110–48.
Reinikka R., Svensson J. (2002) ‘Coping with Poor Public Capital’, Journal of Development
Economics, 69 (1): 51–69.
Rud J. P. (2012a) ‘Electricity Provision and Industrial Development: Evidence from India’,
Journal of Development Economics, 97 (2): 352–67.
Rud J. P. (2012b) ‘Infrastructure Regulation and Reallocations within Industry: Theory and
Evidence from Indian Firms’, Journal of Development Economics, 99 (1): 116–27.
University of East Anglia, Climatic Research Unit. 2013. CRUCY v3.22 Country Averages: PRE.
Available online at http://www.cru.uea.ac.uk/cru/data/hrg/cru_ts_3.22/crucy.1406251334.v3.
22/countries/pre/ (accessed July 25, 2015).
Volta River Authority (1997) 36th Annual Report and Accounts. Accra: Volta River Authority.
Volta River Authority (1998) 37th Annual Report and Accounts. Accra: Volta River Authority.
World Bank. 2001. ‘Introducing a Value Added Tax: Lessons from Ghana.’ World Bank.
World Bank. 2007. ‘Enterprise Surveys.’ World Bank. http://www.enterprisesurveys.org (accessed
November 14, 2008).
Zuberi J. 2012. ‘Estimating the Cost of Power Outages for Large Scale Manufacturing Firms.’
Unpublished manuscript, Department of Economics, University of California, Berkeley.

Abeberese, 2020

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Abeberese, 2020

Uploaded by

Copyright:

Available Formats

Journal of African Economies, 2019, 1–17

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

The Effect of Electricity Shortages on Firm

*Corresponding author: Ama Baafra Abeberese. E-mail: aabebere@wellesley.edu

Key words: investment, infrastructure, electricity supply, manufacturing

JEL classiﬁcation: D25, H54, O13, H54, O14

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

2. Ghana’s electricity crisis

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

3. Theories of the response of investment to shocks

marginal product of capital is predicted to lead to a fall in investment. As electricity is a

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Table 1: Summary Statistics by Investing Status

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Investment rate 0.07 0.03

9 Appendix Table A1 shows the number of ﬁrms in each sector by year.

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

investmentrateijct = β0 + β1 T + λ i + δc t + πj t + εijct (1)

investmentrateijct = α0 + α1 T + α2 T * log (electricityintensity )j + λ i + δc t + πj t + εijct

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

5.2 Effect of electricity rationing on investment

Table 2: Effect on Investment

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Crisis dummy −0.014*** −0.016* −0.015* −0.015*** −0.017* −0.015*

Table 3: Effect on Investment by Sector

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Crisis dummy*wood −0.036** −0.035**

Table 4: Effect on Investment – During and Post 1998

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

During dummy −0.016* −0.016* −0.016* −0.016*

5.3 Effect of electricity rationing on other outcomes

Table 5: Effect on Other Outcomes

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Crisis dummy −0.082 −0.148*** −0.038 0.066 0.05 0.283***

In Column 6 of Table 5, I examine the impact of the electricity crisis on output

5.4 Alternative hypotheses and robustness checks

Table 6: Effect on Other Outcomes (with Controls)

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Crisis dummy −0.076 −0.141*** −0.048** 0.064 0.058 0.313***

Table 7: Effect on Investment – Excluding Food Sectors

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

Crisis dummy −0.008 −0.008 −0.004 −0.004

Table 8: Effect on Investment – Placebo with 1992 as Crisis Period

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

1992 dummy 0.015 0.017 0.014 0.016

Downloaded from https://academic.oup.com/jae/advance-article-abstract/doi/10.1093/jae/ejz013/5433369 by EKU Libraries user on 26 November 2019

You might also like

Crisis dummy*wood −0.036 −0.035

Crisis dummy −0.082 −0.148* −0.038 0.066 0.05 0.283*

Crisis dummy −0.076 −0.141* −0.048 0.064 0.058 0.313***