Access to this full-text is provided by Tech Science Press.
Content available from Energy Engineering
This content is subject to copyright. Terms and conditions apply.
Impact of Electricity Shortages on Productivity: Evidence from Manufacturing
Industries
Xinhong Xue
1
and Zhongcheng Wang
2
,
*
1
School of Finance, Anhui University of Finance & Economics, Bengbu, 233030, China
2
School of International Trade and Economics, Anhui University of Finance & Economics, Bengbu, 233030, China
*Corresponding Author: Zhongcheng Wang. Email: 120180005@aufe.edu.cn
Received: 11 October 2020 Accepted: 30 November 2020
ABSTRACT
Electricity shortages present a significant constraint on manufacturers, who rely on electricity as an important
input into production. In China, electricity supply has been growing rapidly. However, the rapid industrialization
sometimes makes the power supply still unable to meet the demand. Using a survey of 1673 Chinese manufactur-
ing firms, this paper explores how firms response to electricity shortages and its impact on productivity. We find
that self-generation of electricity and investment in Research and Development (R&D) have significant positive
relationships with electricity shortages. Further investigations reveal that self-generation is the most common way
to deal with electricity shortages. However, it aggravates productivity loss. Though investment in Research and
Development can promote productivity, it cannot offset the negative impact of electricity shortages on produc-
tivity. Analyses on subsamples show heterogeneity in the impacts of electricity shortages across firms. In parti-
cular, large firms are more inclined to invest in R&D than small and medium-sized firms are. They are also
the one who suffer significant productivity loss due to self-generation of electricity. Though it is possible for
medium-sized firms to reduce productivity loss through R&D, they are not likely to invest in R&D as a response
to electricity shortages. This study illustrates that firms can hardly have effective solutions to electricity shortages,
and policy makers should take great efforts to increase electricity supply.
KEYWORDS
Electricity shortages; research and development; productivity; electricity efficiency
1 Introduction
Infrastructure is widely perceived to be basic and important for economic growth. For some resources,
like electricity, reliable and economic storage is hardly possible. Unreliable supply of electricity requires
firms to respond in other ways to ensure their performance.
In the past decades, China has spent greatly on power sector. However, electricity supply remains as an
important issue for China’s economic and social development [1–3]. According to the prediction by China
Electricity Council (CEC), in the second half of 2020, there are still power shortages in Hunan, Jiangxi,
Guangdong, and Western Inner Mongolia during the peak period. Production and operation are suspended
when electivity is cut off. Though the cutoff is supposed to be orderly arranged, and enterprises can
adjust their production plan accordingly, losses are unavoidable. Responses, such as self-supply of
This work is licensed under a Creative Commons Attribution 4.0 International License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original
work is properly cited.
DOI: 10.32604/EE.2021.014613
ARTICLE
ech
T
PressScience
electricity with generators, are taken to reduce the losses as much as possible. However, these responses can
be costly by themselves.
In this paper, we study how firms respond to electricity shortages and the impact on productivity. First,
we try to find out whether there are effective measures to deal with power failure. To this end, we examine
firms’investment in technology, labor and generator. Although the impact of energy prices on productivity
and output has been recognized by the existing research, little discussion has been done about the impact of
electricity shortages on the investment in technologies. As most innovations in production processes are
reliant on electricity, we can expect that the electricity intensity of an industry is positively associated
with its technology intensity. In this case, electricity shortages would result in less use of new
technologies. An alternative is to invest in more electricity-efficient technologies, which may offset the
impact of electricity shortages on productivity. In this paper, we will explore how firms respond to
electricity shortages from the perspective of investment in technology.
Second, we also identify other possible partial solutions: Use of temporary workers and generating
electricity as site, namely “self-generation”. Although policymakers are well aware that electricity
shortages can be obstacles for economic growth, it is difficult to increase electricity supply in a short
period, because the construction of power infrastructure normally will take several years. The fluctuation
of electricity demand and the delay of electric power investment will inevitably result in an unbalance
between electricity demand and supply. Quantifying the distortions of firms’activities due to electricity
shortages will motivate policymakers to advance the plan for electric power investment.
Third, we examine whether firms’responses can reduce the productivity loss caused by electricity
shortages. If the productivity loss due to electricity shortages cannot be reduced at firm level, increasing
electricity supply at the country level will be the best solution. For the policymakers, it is important to
increase power supply, and to do it in environmental-friendly way. This study will underline the
important role government plays in developing clean energy and promoting economic growth.
The remainder of the paper is organized as following. Section 2 is literature review; Section 3 represents
our empirical strategy, including the descriptions of data, the empirical model, and the variables we will use.
In Section 4, we will represent our results and discussion and Section 5 will be the conclusions.
2 Literature Review
This paper is part of the extensive literature that evaluates the economic effects of investment in
infrastructure. Based on a panel of 18 OECD countries during 1870–2009, Farhadi [4] found that the
growth of infrastructure could positively influence the growth of labor productivity and total factor
productivity. Similarly, Banerjee et al. [5] found that access to transportation networks had a positive
causal effect on per capita GDP. Yu et al. [6] found positive spatial spillover effects of transport
infrastructure in Chinese regions during 1978–2009. Dimitriou et al. [7] studied the economic effects of
mega infrastructure project. Applying the input-output (IO) method, Dimitriou et al. [8] analyzed the
economic footprint of air transport in Greece, and found that the air transport sector contributed 10.8% of
the national GDP. When assessing the macroeconomic impact of Niger’s Kandaji Dam project, Beguy
et al. [9] found that construction of the project spurred domestic production and increased GDP by 0.25.
The econometric estimations based on panel surveys of the households in rural Vietnam suggested that
grid electrification benefited not only the households in income, expenditure, and educational outcomes,
but also the rural economy as a whole [10]. Some studies have focused on the disproportionate impact of
public infrastructure deficiencies. For example, Hallward-Driemeier et al. [11] and Dollar et al. [12]
revealed that in developing countries enterprises’access to infrastructure services varied by type of
service and firm size, with smaller firms shouldering disproportionate burdens.
996 EE, 2021, vol.118, no.4
Our paper is also part of the subset studies concerning about the impact of energy, especially electricity
supply, on economic growth and performance. Chakravorty et al. [13] examined the effects of connecting a
household to the grid and daily electricity supply on non-agricultural incomes. Using a panel of more than
10,000 households during the period between 1994 and 2005, they found that a grid connection could
increase non-agricultural incomes of rural households by about 9 percent. Lipscomb et al. [14] examined
the long-term effects of electrification on local income and population density in Brazil. They found large
positive effects of improved electricity supply on in-migration of people and firms. Based on an
econometric analysis with data from 1953 to 2010, Cheng [15] revealed that China’s electricity
generation growth way-one Granger caused her GDP growth. Ou et al. [16] used a CGE model to
simulate the impacts of electricity shortages on China’s Economy. They found that electricity shortages
affected the macro-economy negatively, and the light industry, the coal production industry, and the
heavy industry were most affected. Study of Pakistan’s agriculture sector, industrial sector, and service
sector also revealed that electricity shortages deteriorated output [17]. Moreover, power shortages led to
the closure of factories and the losses of jobs [18].
Some studies focused on the effects of electricity shortages on firms’activities. To offset the negative
impacts of electricity shortages on operation, industrial firms often opt for self-generation of electricity.
A review of 25 sub-Saharan countries by Foster et al. [19] showed that in-house generation accounted for
more than 25 percent of the installed generating capacity, though self-generation is over 300 percent more
expensive than electricity supplied from the grid. Similarly, the study on developing countries by Alby
et al. [20] revealed that unreliable electricity supply induced firms to invest in costly generators or to
settle for second-best technologies in sectors whose first-best technology depended heavily on electricity.
In these sectors, small firms were squeezed out of the financial market by a high prevalence of power
outages and unable to borrow to expand production. Hallward-Driemeier et al. [11] and Dollar et al. [12]
also documented patterns of access to infrastructures in developing countries and show that electricity
were often the largest barrier, especially for small firms.
Besides the more adverse impact on small firms, unreliable electricity supply also has negative effects on
industry choices and performance. Abeberese [21] found that Indian firms switched to less electricity-
intensive production processes when there was an increase in electricity price, which led to lower output
and productivity growth rates. On the contrary, electricity network expansion could lead to an increase in
manufacturing output [22]. Allcott et al. [23] found that electricity shortages reduced the average Indian
firm’s revenues and producer surplus. Fisher-Vanden et al. [1] estimated the impacts of electricity
shortages in the early 2000s on Chinese manufacturing firms. They found that firms purchased more
electricity-intensive inputs in response to electricity shortages. This shift from “make”to “buy”, namely
“outsourcing”, could be costly, but helped firms to avoid substantial productivity losses. A recent study
by Elliott et al. [24] revealed that as a result of rising electricity costs, Chinese manufacturing firms were
more likely to switch to a less energy intensive industry.
3 Methodologies
3.1 Data and Information on Chinese Firms
The material we use for empirical study is the World Bank Enterprise Survey (WBES) of Chinese
manufacturing firms. The WBES conducted in 2012 provides the information on a random sample of
1,673 manufacturing firms over the fiscal year 2011 in 25 Chinese cities
1
. Most of these cities are in
eastern China and are either capital cities or main cities. The experience of these typical cities in
2011 could well represent the effects of energy supply on Chinese economic development in the past
decades of years.
1
The samples are from the following 25 cities: Hefei, Beijing, Guangzhou, Shenzhen, Foshan, Dongguan, Shijiazhuang, Tangshan,
Zhengzhou, Luoyang, Wuhan, Nanjing, Wuxi, Suzhou, Nantong, Shenyang, Dalian, Jinan, Qingdao, Yantai, Shanghai, Chengdu,
Hangzhou, Ningbo, Wenzhou.
EE, 2021, vol.118, no.4 997
Firms’locations and sizes can change their access to electricity. Capital cities or main cities, especially
those in eastern provinces, have priorities in electricity consumption, as they are the economic centers.
Important economic position makes it possible that these cities experience less power outages. Therefore,
it is plausible to believe that firms in these cities are less affected by electricity shortages. This may be a
disadvantage, because the impact of electricity shortages would be underestimated. Another important
factor is firms’sizes. Large firms play an important role in local economic development. Though enjoying
more bargaining power over resources access, they are also vulnerable as they may use more machines in
production process. In fact, about 12% of the large firms in our investigation reported that they
experienced at least one power outage in a typical month in 2011.
Our work provides firm-level distribution of energy shortages. Based on the firm-level data on power
outages in a typical month over the fiscal year 2011, we build up two indicators of electricity shortages.
The measure is to divide firms into three groups. The first group consists of firms who reported
experiencing no power outage. The second group includes firms who reported experiencing one power
outage in a typical month. The third group consists of firms who experienced two or more outages a
month. Among the 232 firms who experienced power outages 51 firms reported that they experienced
more than two times of outages a month, and 181 firms reported only one outage a month. Fig. 1 shows
the distribution of power outages.
The WBES dataset also provides information on sales, employees, investment in Research and
Development and so on over the fiscal year 2011. Tab. 1 is a summary of the information including the
number of firms, power outages, the average number of permanent and temporary employees, and the
number of firms who used licensed technology, invested in research and development (R&D), and owned
or shared a generator. According to the number of permanent, full-time employees at the end of 2011,
there are 228 firms with employees less than 20, and 724 firms with more than 100 employees. Firms
with employees between 20 and 100 are most likely to suffer from power outages. There are totally
210 firms owning or sharing a generator, and almost half of them are large firms. The large firms also
have the largest average number of temporary, full-time workers, the highest ratio of investing in R&D
and using licensed technology from a foreign company.
Figure 1: Distribution of power outages in a typical month and firms’permanent, full-time workers
998 EE, 2021, vol.118, no.4
There seems to be positive relationships between electricity shortages and self-generation, use of
temporary workers, and investment in new technology including using licensed technology from foreign
companies and investing in R&D. When production is significantly affected by electricity shortages, firms
will respond in a number of ways to offset the effect. For example, a firm may choose to invest in more
electricity-efficient technologies either by investing in R&D or by buying technology from other
companies, which will increase output when electricity is available. Owning or sharing a generator will
provide self-generating electricity during blackout. If power outages are seasonal or scheduled, using
temporary workers to produce more when there are no blackouts are plausible.
In the following empirical analysis, we first explore the ways Chinese manufacturing firms respond to
electricity shortages. Their responses have important policy implications. If there is a significant positive
relationship between electricity shortages and investing in R&D, it implies that the investment in R&D
might be able to improve electricity efficiency. Such technology will improve productivity, and at the
same is environmental friendly. Using licensed technology from foreign companies may have the same
implication. If a significant positive relationship is found between electricity shortages and owning or
sharing a generator, it indicates that Chinese manufacturing firms turn to self-generation and might suffer
productivity loss, becasue the price of self-generation is higher than that of grid. If the production is
electricity-intensive, it is less likely to offset the effect of electricity shortages by using more labors as
most of the production processes depend on machines. A significant positive relationship between
electricity shortages and temporary, full-time workers might imply that the production is less dependent
on electricity and more labor-intensive.
The second issue we are interested in is whether the above possible responses are able to relieve or offset
the impact of electricity shortages on firms’productivity. Take R&D investment for example. Technology
innovation is closely related to productivity improvement. If productivity-enhancing technologies are
those that are electricity-intensive, technology development will make production more relied on
electricity. In this case, only more electricity-efficient technology will bring higher productivity when
there are electricity shortages.
According to the discussion above, we develop three testable hypotheses:
1. Investment in technology: Firms may invest in R&D or licensed technology from foreign-owned
companies when facing electricity shortages, implying a potential productivity improvement.
2. Self-Generation: Firms may turn to self-generation during power outages, resulting in high
production cost and low productivity.
3. Temporary workers: Firms may substitute away from machines toward labors to ensure production,
implying switching to less electricity-intensive industries.
Table 1: Descriptions of electricity shortages and other factors
Firm size Number
of firms
Power outages Generator Employees Temporary
workers
Licensed
technology
R&D
One Two or more
small 228 24 (10.6%) 3 (1.3%) 19 (8.3%) 13 3 21 (9.2%) 46 (20.2%)
medium 721 94 (13.2%) 19 (2.7%) 90 (12.5%) 55 8 148
(20.5%)
294
(40.8%)
large 724 63 (8.8%) 29 (4.0%) 101
(14.0%)
609 18 231
(31.9%)
379
(52.3%)
all firms 1673 181
(10.9%)
51 (3.1%) 210
(12.6%)
289 12 400
(23.9%)
719
(43.0%)
Note: Firms with employees less than 20 are small; firms with 20 to 100 employees are medium; firms with over 100 employees are large.
EE, 2021, vol.118, no.4 999
3.2 Empirical Model
To test the three hypotheses presented in Section 3.1, we start with examining the relationship between
electricity shortages and the possible responses: Investment in technology, self-generation and using
temporary, full-time workers. To this end, we use the following empirical Model 1:
responsei¼a0þa1outageiþa2outagesiþcontfirmiþcont regionrþgrþgjþei(1)
where idenotes firm, rdenotes region, and jdenotes industry. response
i
denotes firm’s responses to
electricity shortages, including investing in R&D, using licensed technology from foreign companies,
owning or sharing a generator, and using temporary, full-time employees. outage
i
and outages
i
are
indicators of power outages the firms experienced in a typical month. cont_firm
i
is a set of firm-level
control variables, including labor-productivity, human capital, size, ownership and exporting experience.
cont_region
r
consists of two region-level control variables, GDP and population of the cities where the
firms locate. η
r
and η
j
are fixed effects for city and industry, respectively, to control bias due to
unobserved factors. ε
i
is the error term that includes all the unobservable factors.
Another important issue is whether these responses can offset the impact of electricity shortages on
productivity. As most of the production processes are reliant on electricity, we might expect it to be the
case that these responses are to relieve the impact of electricity shortages on productivity. The
identification of the effects is achieved through the presence of interaction terms between the indicators of
power outages and the indicators of the responses. A positive and significant coefficient indicates that a
certain response can reduce productivity loss due to electricity shortages. To this end, we use the
following empirical Model 2:
laborprodi¼a0þa1outageiþa2outagesiþa3outageiresponseiþa4outagesireponsei
þ0cont firmiþcont regioniþgrþgjþei
(2)
where i,jand rdenote firm, industry and region, respectively. laborprod
i
represents labor productivity. All
the other symbols have the same meaning as in Model 1.
3.3 Measurements
1. Electricity shortages: outage
i
and outages
i
. In WBES dataset, the firms reported how many power
outages they experienced in a typical month. For firms who reported one power outage in a
typical month, variable outage
i
will be assigned value of one. Otherwise, it will be assigned zero.
outages
i
is one for firms who reported two or more outages. Otherwise, it is zero.
2. Investment in technology: R&D
i
and licensed_tech
i
. We use two dummies to indicate whether the
firms had invested in new technology. Dummy R&D
i
is equal to one for firms who had spent on
R&D. A second variable to measure investment in new technology, licensed_tech
i
, is based on the
report on whether using technology licensed by a foreign-owned company, excluding office
software. For the firms who reported “Yes ”,licensed_tech
i
is equal to one.
3. Self-generation: self_g
i
.self_g
i
is equal to one if a firm owned or shared a generator.
4. Temporary worker: temp_w
i
. It is the number of full-time seasonal or temporary workers.
5. Control variables: cont_firm
i
(firm-level control variables) and cont_region
r
(region-level control
variables). laborprod
i
denotes firms’productivity. As there is no information on total output for
the calculation of total factor productivity, we use sales per capita as a proxy, where sales are
total annual sales for all products and services. worker_edu
i
, denoting human capital, is the ratio
of employees who completed second school to the total of full-time permanent workers. exporter
i
,
a dummy, is equal to 1 if more than 10% of the products is exported directly. According to the
definition of foreign direct investment in Chinese Statistical Yearbook, we define foreign
i
as a
1000 EE, 2021, vol.118, no.4
dummy, equal to 1 if private foreign individuals, companies or organizations own over 10% share of
the firms. laborprod
i
,worker_edu
i
,exporter
i
, and foreign
i
are all firm-level control variables. The
region-level control variables include city_GDP
j
and city_pop
j
, denoting GDP and resident
population of the cities where the firms are located.
The definitions and sources of the variables are summarized in Tab. 2. The descriptive statistics are
presented in Tab. 3.
Table 2: Variables, their definitions and sources
Variable
type
Variable Definition Data sources
Electricity
shortages
outage Dummy, One power outage in a typical month World Bank
Enterprise Survey
outages Dummy, two or more power outages in a typical month
Possible
responses
R&D Dummy, investing in R&D World Bank
Enterprise Survey
licensed_tech Dummy, use licensed technology from a foreign-owned
company, excluding office software
self_g Dummy, own or share a generator
temp_w Number of full-time seasonal or temporary workers
cont_firm:laborprod labor productive: Logarithm of sales per capita World Bank
Enterprise Survey
worker_edu Human capital: Ratio of full-time permanent workers who
completed second school
exporter Dummy, more than 10% of the products is exported directly
foreign Dummy, over 10% share of the firms are owned by private
foreign individuals, companies or organizations
cont_region: city_GDP Logarithm of city GDP Statistical
yearbooks of each
city
city_pop Logarithm of resident population of the cities
Table 3: Descriptive statistics
Variable Mean SD Min Max
outage 0.109 0.311 0 1
outages 0.030 0.172 0 1
R&D 0.433 0.495 0 1
licensed_tech 0.241 0.428 0 1
self_g 0.125 0.331 0 1
temp_w 11.979 38.400 0 600
laborprod 12.458 1.034 8.911 17.770
worker_edu 50.285 28.205 0 100
exporter 0.198 0.398 0 1
foreign 0.075 0.264 0 1
city_GDP 8.646 0.451 7.701 9.862
city_pop 6.735 0.261 6.377 7.761
EE, 2021, vol.118, no.4 1001
4 Results and Discussion
4.1 Response of Chinese Manufacturing Firms as a Whole to Electricity Shortages
Our first step is to investigate firms’responses to electricity shortages by estimating the Model 1. Tab. 4
reports the results. We omit the firm-level and region-level control variables in odd-numbered columns (1),
(3), (5) and (7). Our results remain qualitatively unchanged when these variables are controlled for in the
even-numbered columns, indicating that our results are robust to a certain extent.
In column (1) and column (2), the coefficients on outage and outages are all positive and significant
when the dependent variable is R&D. In column (2), the coefficients on outage and outages are
0.373 and 0.471, respectively. Electricity shortages increase the probability of investing in R&D. One
power outage a month increases the probability of investing in R&D by approximately 13.6%. The
probability would be even higher, namely 17.2%, if there are two power outages in a typical month.
The coefficients on control variables, except foreign and city_GDP, are positive and significant. The
significant positive relationship between electricity shortages and investment in R&D suggests that when
facing electricity shortages, firms are likely to adopt new and advantage technologies or machines. We
will investigate in Sector 4.4 whether R&D investment aims at relieving the impact of electricity
shortages on productivity.
Table 4: Estimations for firms’response and electricity shortages
Dependent Variable: response
Independent
Variables
response =R&D response =licensed_tech response = temp_w response =self-g
(1) (2) (3) (4) (5) (6) (7) (8)
outage 0.196*
(0.101)
0.373***
(0.106)
−0.069
(0.110)
0.075
(0.119)
2.972
(4.008)
3.932
(4.299)
1.187***
(0.106)
1.153***
(0.110)
outages 0.478***
(0.183)
0.471**
(0.190)
0.024
(0.197)
−0.122
(0.201)
14.518
(10.278)
13.862
(10.293)
1.337***
(0.183)
1.175***
(0.185)
city_pop 0.386**
(0.171)
0.460***
(0.175)
12.623***
(4.304)
−0.366
(0.228)
city_GDP −0.477***
(0.094)
0.193*
(0.102)
−6.633**
(2.820)
0.256**
(0.122)
laborprod 0.198***
(0.032)
0.076**
(0.035)
1.893*
(1.047)
0.100**
(0.041)
worker_edu 0.007***
(0.001)
0.006***
(0.001)
0.015
(0.031)
−0.005***
(0.001)
foreign 0.203
(1.127)
0.719***
(0.125)
1.251
(4.554)
0.212
(0.162)
exporter 0.427***
(0.084)
0.437***
(0.089)
6.017**
(2.995)
0.360***
(0.106)
constant −0.088
(0.088)
−1.393
(1.013)
−0.364***
(0.097)
−6.614***
(1.057)
18.858***
(2.856)
−34.702
(27.646)
−1.505***
(0.125)
−2.370*
(1.362)
City fixed effects & Industry fixed effects are controlled
observations 1647 1621 1645 1617 1631 1609 1658 1631
R
2
0.01 0.07 0.03 0.12 0.01 0.02 0.13 0.16
Notes: (1) *, **, ***: Coefficients are statistically significant at the levels of 10%, 5% and 1%, respectively. (2) In the parentheses are robust standard
error.
1002 EE, 2021, vol.118, no.4
Though R&D is an important source for technology development, the failure of R&D will result in
nothing of any new or advanced technologies. Besides, self-reported R&D investment could be noisy and
subjective because it is based on the firms’self-assessment about their product or process innovations.
Therefore, following Montalbano et al. [25], we use variable licensed_tech as an indicator of investment
in advanced technology. The results are presented in column (3) and column (4). To our surprise, the
coefficients on outage are not significant. Neither are the coefficients on outages. In column (5) and
column (6) are the estimations for using temporary, full-time workers. The coefficients on outage and
outages are positive, but not significant. This means using more temporary labors is not a popular way to
deal with electricity shortages.
Column (7) and column (8) report the results for owning or sharing a generator. The coefficients on the two
indicators of electricity shortages are positive and significant at the 1% level. In particular, for firms who reported
one outage a month, the probability of owning or sharing a generator is 19.6 percent higher. For firms who
experienced two or more outages, the probability is almost 20 percent higher. It is highly possible that firms
use self-generated electricity to relieve the impact of electricity shortages on their production.
4.2 Robustness Tests
As can be seen from Tab. 1, the small firms have the smallest sample size and the lowest ratio in
spending on R&D and generator. Spending on R&D and generator have a higher request for financing
ability. However, small firms are in an inferior positon. Missing of the control for firms’abilities to
obtaining finance may result in misleading conclusions. In Tab. 4, the insignificant coefficients on power
outages may catch not only the effect of power outages, but also the effect of financial abilities. The firms
reported to what degree “access to finance”is an obstacle to their operation in the WBES. Thus, we add
afinance-specific variable, finaccess, to Model 1. finaccess is equal to zero when access to finance is
reported as “no obstacle”or “minor obstacle”, and equal to one when access to finance is regarded as
“moderate obstacle”,“major obstacle”,or“very severe obstacle”. The results are presented in column (1)
and column (4) of Panel A and Panel B in Tab. 5. After controlling for the financial abilities, there
remain significant positive relationships between electricity shortages and two types of responses, namely,
investment in R&D and self-generation. Our results remain qualitatively unchanged from those in Tab. 4.
Table 5: Robustness tests
Dependent variable: response
Panel A (1) (2) (3) (4) (5) (6)
response =R&D response =licensed_tech
outage 0.321*** (0.107) 0.368*** (0.107) 0.383*** (0.114) 0.081 (0.119) 3.932 (4.299) 0.072 (0.125)
outages 0.505*** (0.194) 0.522** (0.193) 0.457** (0.197) −0.114 (0.201) 13.862 (10.293) −0.115 (0.205)
constant −1.993* (1.066) −1.332 (1.033) −0.822 (1.093) −6.797*** (1.072) −34.702 (27.646) −6.859*** (1.151)
observations 1611 1611 1410 1607 1609 1405
R
2
0.09 0.08 0.07 0.12 0.02 0.11
Panel B response =temp_w response =self-g
outage 3.190 (4.363) 3.588 (4.245) 3.912 (4.855) 1.147*** (0.111) 1.174*** (0.111) 1.209*** (0.119)
outages 14.329 (10.390) 14.378 (10.477) 14.188 (10.923) 1.195*** (0.189) 1.219*** (0.188) 1.241*** (0.192)
constant −30.775 (28.273) −25.658 (27.732) −24.331 (31.737) −2.344* (1.396) −2.355* (1.407) −2.542* (1.457)
observations 1598 1598 1397 1620 1620 1418
R
2
0.02 0.02 0.02 0.17 0.18 0.18
Notes: (1) *, **, ***: Coefficients are statistically significant at the levels of 10%, 5% and 1%, respectively. (2) In the parentheses are robust standard
error. (3) Control variables and fixed effects are also included in estimating, but not reported here in order to save space.
EE, 2021, vol.118, no.4 1003
To investigate whether the above results hold, we use another measure of finaccess. When access to
finance is taken as “major obstacle”or “very severe obstacle”,finaccess is equal to one. Otherwise, it is
equal to zero. The results are reported in column (2) and column (5) of Panel A and Panel B. We get
consistent results, too.
Another source of bias is the sample size of the small firms. The small firms take only 13.6% of the
whole sample. The WBES sample for China is selected using stratified random sampling at three levels:
Industry, establishment size, and region. To some degree, the small sample size of the small firms is due
to their small part in the industry or in the region. For the sake of robustness, Model 1 is estimated
without the small firms. The results are presented in column (3) and column (6) of Panel A and Panel B
in Tab. 5. Again the results are very similar to the previous ones.
4.3 Heterogeneity in Responses to Electricity Shortages
In sector 4.1, we investigate the ways firms respond to electricity shortages using all the firms as a
sample, and the results are proved robust in Sector 4.2. However, how to respond to electricity shortages
is also related to some characteristics of the firms. For example, investment in R&D and licensed
technology incur high costs. Small firms usually have less financial resource to cover the costs. On the
other hand, small firms are more flexible in management and production plan so that it is more possible
for them to offset the impact through coordinating production plans with electricity supply. The WBES
classifies firms into three groups: Small, medium, large according to their number of employees. The
small firms are those with less than 20 employees (excluding 20). The medium-sized firms are those with
20 to 100 employees (excluding 100). The firms with more than 100 employees are large. Tab. 6 reports
the results of the heterogeneous responses to electricity shortages across firms of different sizes.
As can be seen in Tab. 6, the heterogeneous responses exist mainly in the relationship between electricity
shortages and investment in R&D. In particular, column (1) and column (2) in Panel A represent the results
Table 6: Heterogeneous responses to electricity shortages
Dependent variable: response
Independent variables (1) (2) (3) (4) (5) (6)
small firms medium-sized firms large firms small firms medium-sized firms large firms
Panel A
response =R&D response =liecensed_tech
outage 0.233 (0.309) 0.203 (1.151) 0.729*** (0.193) −0.029 (0.417) 0.037 (0.187) 0.127 (0.178)
outages 0.449 (0.802) 0.263 (0.307) 0.606** (0.263) —— 0.017 (0.343) −0.212 (0.259)
constant −4.987* (2.812) −0.340 (1.561) −0.965 (1.601) −4.706 (3.313) −6.736*** (1.760) −7.203*** (1.587)
observations 225 695 701 223 691 700
R
2
0.04 0.04 0.12 0.25 0.14 0.09
Independent variables Panel B
response =temp_w response =self_g
outage 2.516 (1.525) 0.813 (1.723) 10.907 (11.746) 0.792** (0.322) 1.273*** (0.164) 1.266*** (0.183)
outages 2.114 (2.293) 1.287 (3.740) 21.613 (17.514) —— 1.460*** (0.305) 1.129*** (0.246)
constant −17.355 (10.806) −74.277** (32.626) −2.623 (56.730) −0.795 (3.556) −2.146 (2.280) −2.416 (1.932)
observations 225 694 690 216 698 707
R
2
0.05 0.04 0.03 0.07 0.17 0.21
Notes: (1) *, **, ***: Coefficients are statistically significant at the levels of 10%, 5% and 1% respectively. (2) In the parentheses are robust standard error. (3)
Other control variables and fixed effects are also included in estimating, but not reported here in order to save space. (4) “—” means perfect collinearity.
1004 EE, 2021, vol.118, no.4
for the small firms and the medium-sized firms, respectively. The coefficients on outage and outages are
positive, but not significant. It is probably because R&D is a too costly for them. Contrary to the small
and the medium-sized firms, we find a significant positive relationship between electricity shortages and
R&D investment in the large firms as shown in column (3) Panel A.
From column (4) to column (6) in Panel A and from column (1) to column (3) in Panel B are results for
using licensed technology and temporary workers, respectively. No coefficients on the indicators of
electricity shortages are significant. In contrast, all the coefficients on the indicators of electricity
shortages in columns (4), (5), and (6) in Panel B are positive and significant, implying that all firms,
regardless of their sizes, are likely to turn to self-generation. Our finding differs from Fisher-Vanden et al.
[1] in respective of large firms. Their sample focused on the largest electricity users and they found that
large firms did not self-generate electricity often.
So far, we have consistently found that when the firms are faced with electricity shortages, they are likely
to invest in R&D, or to generate electricity themselves. However, we do not find any evidence for adopting
licensed technology or hiring temporary workers as solutions to electricity shortages.
4.4 Effects on Labor Productivity
In Sections 4.1, 4.2, and 4.3, our empirical analyses reveal that investment in R&D and self-generation
are positively related to electricity shortages. We want to find out whether these responses can relieve the
negative impact of electricity shortages on productivity. To this end, we estimate Model 2. Our interest is
on the interaction terms involving the responses and the indicators of electricity shortages. We expect that
the coefficients on the interaction terms will be positive, if a certain response can offset the impact of
electricity shortages on productivity.
Tab. 7 reports the results. From column (1) to column (4) are the results for R&D investment. The result
for all firms in column (1) shows that no interaction terms are significant, while the coefficients on R&D is
positive and significant at the 1% level. Investing in R&D can promote labor productivity, but cannot offset
the impact of electricity shortages. The results from column (2) to column (4) reveal that in medium-sized
firms it is not the case. In particular, the coefficient on the interaction term of outages and R&D in
column (3) is positive and significant at the 5% level, implying that R&D investment in medium-sized
firms can improve labor productivity by offsetting the impact of electricity shortages. We do not find the
same effects in the large firms or in the small firms. Instead, we find that in the large firms the coefficient
on R&D is significant and positive, while the coefficients on the interaction terms are not significant.
Therefore, the positive effect of R&D investment on productivity exists mainly in the large firms, while
the effect of offsetting the negative impact of electricity shortages exists mainly in the medium-sized firms.
From column (5) to column (8) in Tab. 7 are the results for self-generation. To our surprise, the
coefficients on the interaction terms are all negative. Particularly, the coefficients on the interaction terms
of self-generation and the indicator of outages are negative and significant in the full sample and the sub-
sample of large firms. Generating electricity on site not only requires additional capital and diesel
purchases, but also crowds out other investment opportunities and reduces productivity. When power
outages happen occasionally, self-generation will ensure firms from output reduction. However, self-
generation is costly. Frequent self-generation will make it less advisable because it will increase
production cost disproportionally, especially when the production is highly electricity dependent, such as
in large firms who use more machines or more electricity-intensive technologies. Therefore, we reach the
conclusion that firms do not self-generate electricity to reduce the loss of labor productivity due to
electricity shortages.
EE, 2021, vol.118, no.4 1005
5 Conclusions
In this study, we explore the responses of manufacturing firms to electricity shortages and their effects on
labor productivity using the WBES dataset. We find that when facing electricity shortages, firms of all sizes
switch to self-generation. However, self-generation causes productivity loss, which is most obvious in the
large firms to whom generators are most accessible. We also find that firms, especially large firms, are
more probable to spend on R&D. Nevertheless, the medium-sized firms benefit from R&D from the
perspective of reducing productivity loss due to electricity shortages. The large firms can improve their
productivity through R&D, but cannot offset the negative impact of electricity shortages. Besides, we do
not find that using more labors or licensed technologies can play important roles in coping with
electricity shortages. To sum up, our study reveals that self-generation is the most adopted, but also costly
solution to electricity blackouts. Technology developments from R&D can promote productivity, but
cannot offset the negative impact of blackouts on productivity.
The results of this paper underline the importance of infrastructure supply in developing countries. Faced
with infrastructure constrains—in this case, unreliable electricity supply—firms may use costly self-generation
in an attempt to rely less on that infrastructure. However, this damages productivity. Though new technology is
usually supposed to promote productivity, our findings remind us that electricity shortages cannot be offset by
technology innovations in manufacturing sectors, because technology development makes machines more
widely used, and thus production becomes more electricity-intensive. Although this paper addresses
electricity shortages specifically, one can imagine that firms may react in undesirable ways to cope with
other infrastructure constraints. The findings in this study also imply the important role the government
should play in infrastructure supply, as it can hardly be economically self-sufficient at firm level. Although
our study seem to suggest that more power plants be built, this conclusion could not be made at the
expenses of environment. Actually, it is more advisable for policy makers to consider other mechanisms to
trade off the social cost of investing in energy supply with the benefits.
Funding Statement: This work is supported by “Philosophy and Social Sciences Planned Project of Anhui
Province, China”(Grant No. AHSKY2020D37).
Table 7: Estimations for the effects on labor productivity
Dependent variable: laborprod
Independent
Variables
(1) (2) (3) (4) (5) (6) (7) (8)
all firms small medium large all firms small medium large
response =R&D response = self_g
outage•
response
−0.089
(0.146)
0.206
(0.573)
−0.111
(0.173)
0.049
(0.236)
−0.158
(0.176)
−0.841
(0.524)
0.001
(0.221)
−0.014
(0.315)
outages•
response
0.101
(0.333)
−1.913
(2.153)
0.743**
(0.364)
−0.162
(0.279)
−0.607**
(0.293)
—— −0.303
(0.426)
−0.737***
(0.283)
response 0.325***
(0.055)
0.243
(0.154)
0.118
(0.078)
0.554***
(0.087)
0.259**
(0.101)
0.241
(0.205)
0.130
(0.155)
0.348**
(0.160)
constant 11.361***
(0.758)
11.646***
(1.844)
11.966***
(1.013)
10.444***
(1.262)
11.625***
(0.761)
11.690***
(1.787)
12.047***
(1.027)
11.116***
(1.281)
observations 1622 225 696 701 1631 226 698 707
R
2
0.05 0.07 0.07 0.11 0.04 0.06 0.06 0.06
Notes: (1) *, **, ***: Coefficients are statistically significant at the levels of 10%, 5% and 1%, respectively. (2) In the parentheses are robust standard
error. (3) Independent variables outage and outages, other control variables, and fixed effects are also included in estimating, but not reported here in
order to save space. (4) “——” means perfect collinearity.
1006 EE, 2021, vol.118, no.4
Conflicts of Interest: The authors declare that they have no conflicts of interest to report regarding the
present study.
References
1. Fisher-Vanden, K., Mansur, E. T., Wang, Q. (2015). Electricity shortages and firm productivity: Evidence from
China’s industrial firms. Journal of Development Economics, 114, 172–188. DOI 10.1016/j.jdeveco.2015.01.002.
2. Lin, B., Wang, Y. (2019). Inconsistency of economic growth and electricity consumption in China: A panel VAR
approach. Journal of Cleaner Production, 229, 144–156. DOI 10.1016/j.jclepro.2019.04.396.
3. Tsai, C. M. (2011). The reform paradox and regulatory dilemma in China’s electricity industry. Asian Survey,
51(3), 520–539. DOI 10.1525/as.2011.51.3.520.
4. Farhadi, M. (2015). Transport infrastructure and long-run economic growth in OECD countries. Transportation
Research, 74, 73–90. DOI 10.1016/j.tra.2015.02.006.
5. Banerjee, A., Duflo, E., Qian, N. (2012). On the road: Access to transportation infrastructure and economic growth
in China. NBER Working Paper No. 17897.
6. Yu, N., de Jong, M., Storm, S., Mi, J. (2013). Spatial spillover effects of transport infrastructure: Evidence from
Chinese regions. Journal of Transport Geography, 28, 56–66. DOI 10.1016/j.jtrangeo.2012.10.009.
7. Dimitriou, D. J., Mourmouris, J. C., Sartzetaki, M. F. (2015). Economic impact assessment of mega infrastructure
pipeline projects. Applied Economics, 47(40), 4310–4322. DOI 10.1080/00036846.2015.1026591.
8. Dimitriou, D., Sartzetaki, M. (2018). Assessing air transport socio-economic footprint. International Journal of
Transportation Science and Technology, 7(4), 283–290. DOI 10.1016/j.ijtst.2018.07.001.
9. Beguy, O., Dessus, S., Garba, A., Hayman, J., Herderschee, J. (2015). Modeling the impact of large infrastructure
projects: A case study from Niger. MFM Discussion Paper No. 7.
10. Barnes, D. F., Khandker, S. R., Nguyen, M. H., Samad, H. A. (2016). Welfare impacts of rural electrification:
Evidence from Vietnam. Policy Research Working Paper, 61(3), 659–692.
11. Hallward-Driemeier, M., Stewart, D. (2004). How do investment climate conditions vary across countries, regions
and types of firms? Background Paper for the 2005 World Development Report, World Bank, Washington, DC.
12. Dollar, D., Hallward-Driemeier, M., Mengistae, T. (2005). Investment climate and firm performance in developing
Economies. Economic Development and Cultural Change, 54(1), 1–31. DOI 10.1086/431262.
13. Chakravorty, U., Pelli, M., Ural Marchand, B. (2014). Does the quality of electricity matter? Evidence from rural
India. Journal of Economic Behavior & Organization, 107, 228–247. DOI 10.1016/j.jebo.2014.04.011.
14. Lipscomb, M., Mobarak, A. M., Barham, T. (2013). Development effects of electrification: Evidence from the
topographic placement of hydropower plants in Brazil. American Economic Journal: Applied Economics, 5(2),
200–231. DOI 10.1257/app.5.2.200.
15. Cheng, Y. S., Wong, W. K., Woo, C. K. (2013). How much have electricity shortages hampered China’s GDP
growth? Energy Policy, 55, 369–373. DOI 10.1016/j.enpol.2012.12.015.
16. Ou, P., Huang, R., Yao, X. (2016). Economic impacts of power shortage. Sustainability, 8(7), 687. DOI 10.3390/
su8070687.
17. Shahbaz, M. (2015). Measuring economic cost of electricity shortage: Current challenges and future prospects in
Pakistan. Mpra Paper, No. 67164.
18. Siddiqui, R., Jalil, H. H., Nasir, M., Malik, W. S., Khalid, M. (2011). The cost of unserved energy: Evidence from
selected industrial cities of Pakistan. PIDE-Working Papers No. 2011:75.
19. Foster, V., Steinbuks, J. (2009). Paying the price for unreliable power supplies: In-house generation of electricity
by firms in Africa. Policy Research Working Paper Series 4913, World Bank, Washington, DC.
20. Alby, P., Dethier, J. J., Straub, S. (2013). Firms operating under electricity constraints in developing countries.
World Bank Economic Review, 27(1), 109–132. DOI 10.1093/wber/lhs018.
21. Abeberese, A. B. (2017). Electricity cost and firm performance: Evidence from India. Review of Economics and
Statistics, 99(5), 839–852. DOI 10.1162/REST_a_00641.
EE, 2021, vol.118, no.4 1007
22. Rud, J. P. (2012). Electricity provision and industrial development: Evidence from India. Journal of Development
Economics, 97(2), 352–367. DOI 10.1016/j.jdeveco.2011.06.010.
23. Allcott, H., Collardwexler, A., Connell, S. D. O. (2016). How do electricity shortages affect productivity?
Evidence from India. American Economic Review, 106(3), 587–624. DOI 10.1257/aer.20140389.
24. Elliott, R., Sun, P., Zhu, T. (2019). Electricity prices and industry switching: Evidence from Chinese manufacturing
firms. Energy Economics, 78, 567–588. DOI 10.1016/j.eneco.2018.11.029.
25. Montalbano, P., Nenci, S. (2019). Energy efficiency, productivity and exporting: Firm-level evidence in Latin
America. Energy Economics, 79, 97–110. DOI 10.1016/j.eneco.2018.03.033.
1008 EE, 2021, vol.118, no.4
Available via license: CC BY 4.0
Content may be subject to copyright.
