Bio-based and Applied Economics
BAE

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404
Copyright: © 2023 Squarcina, M., Romano, D. 
Open access, article published by Firenze University Press under CC-BY-4.0 License. 
Firenze University Press | www.fupress.com/bae

 

Citation: Squarcina, M., Romano, D. 
(2023). The impact of COVID-19 on house-
hold income and participation in the 
agri-food value chain: Evidence from 
Ethiopia. Bio-based and Applied Eco-
nomics 12(4): 333-366. doi: 10.36253/bae-
13404 

Received: July 21, 2022 
Accepted: October 23, 2023 
Published: December 31, 2023 

Data Availability Statement: All rel-
evant data are within the paper and its 
Supporting Information files. 

Competing Interests: The Author(s) 
declare(s) no conflict of interest. 

Editor: Davide Menozzi 

ORCID 
MS: 0000-0003-1128-1568 
DR: 0000-0001-7120-8050

The impact of COVID-19 on household income 
and participation in the agri-food value chain: 
Evidence from Ethiopia

Margherita Squarcina1,2,*, Donato Romano1

1 Department of Economics and Management, Universita Degli Studi di Firenze, Florence, 
Italy
2 Department of Agricultural Economics and Rural Development, Georg-August-Universi-
ty, Göttingen, Germany
*Corresponding author: margherita.squarcina@uni-goettingen.de

Abstract. The effects of COVID-19 have been highly heterogeneous, crucially depending 
on household livelihoods. In the context of households reliant on agri-food systems, the 
extent of these effects significantly depends on their position within the value chain. An 
assessment of the COVID-19 effects along the agri-food value chain and the identification 
of pivotal factors influencing these outcomes are key for designing appropriate responses 
and targeting the population most in need should a crisis akin to COVID-19 emerge in 
the future. Using a longitudinal dataset from Ethiopia, composed of a pre-COVID base-
line and six follow-up phone-based surveys, this paper estimates the COVID-19-induced 
change in household income and job participation, tracing its evolution throughout 
seven months after the pandemic onset. Applying both longitudinal and cross-sectional 
econometric models, we show that the COVID-19 shock reduced both employment and 
income, with increasingly negative impacts over time. Despite initial resilience in the face 
of restrictive measures, farming eventually emerged as the most affected segment within 
the agri-food value chain over the medium term. Access to formal institutions such as 
insurance and credit services, formal contractual arrangements, and secured land owner-
ship title played a key role in mitigating the likelihood of income loss.

Keywords: COVID-19, food value chain, labor market, income loss, Ethiopia.
JEL Codes: I15, O12, Q12.

1. INTRODUCTION

The COVID-19 pandemic caused unprecedented disruptions in many 
value chains at domestic as well as global levels (Moosavi et al., 2022), 
including the bioeconomy and specifically the agri-food value chains 
(AFVCs) (Devereux et al., 2020), although significant heterogenous effects 
were reported1. Although some segments of AFVC such as farming have 

1 For instance, in the short run the bioeconomy – i.e. the economic activities that depend on the 
use of biological resources, including agriculture and food processing – showed a level of resil-

https://doi.org/10.36253/bae-13404
http://www.fupress.com/bae
https://doi.org/10.36253/bae-13404
https://doi.org/10.36253/bae-13404
https://orcid.org/0000-0003-1128-1568
https://orcid.org/0000-0001-7120-8050
mailto:margherita.squarcina@uni-goettingen.de


334

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

been initially less affected by restriction decisions, 
downstream segments such as food services, restaurants, 
and retail as well as midstream segments such as pro-
cessing, logistics, and transportation, have been impact-
ed since the onset of the crisis2. The general conclusion 
of early studies is that the COVID-19 impact is differ-
entiated across different segments of the AFVC as well 
as within each segment (Diao et al., 2020; Tamru et al., 
2020; Tesfaye et al., 2020). 

The pandemic and the related restrictions imple-
mented by governments raised many challenges to indi-
viduals and households participating in the AFVC. The 
ability to absorb, adapt, and even transform the way a 
livelihood is gained by individuals and households – in 
short, their resilience to the COVID-19 shock – has been 
often limited by many factors such as access to technol-
ogy, financial services, or social safety nets3. Indeed, 
many agents had limited options to cope with the COV-
ID-19 shock, resulting in income reduction or job loss 
and eventually increasing poverty and food insecurity. 
Assessing COVID-19 impacts across AFVC segments 
and identifying the main factors that determined those 
impacts on AFVC participants and their options to 
adapt to the “new normal” is then crucial for designing 
appropriate responses and targeting the groups most in 
need should a shock similar to COVID-19 occur again 
in the future. 

Using Ethiopia as a case study, this study aims at: (i) 
assessing which segments of the AFVC have been most 
affected by the pandemic, in terms of labor participation 
and income loss; and (ii) identifying which factors at the 
household level have mostly influenced the impact of 
COVID-19 on income, and specifically on farm income. 
Ethiopia has been selected for several reasons. Its econ-
omy is mainly based on agriculture, which accounts for 
34% of GDP (World Bank, 2021), 80% of the population 
depends on agriculture (Njeru et al., 2016), and small-
holder farming accounts for 95% of agricultural produc-
tion (Tigre and Heshmati, 2022). However, new com-
mercial and gig economy clusters are emerging in the 
country, as is the case of intensive vegetable cultivation 

ience relatively higher than the overall economy in Europe. However, 
this result was mainly driven by the technology-intensive sectors of the 
bioeconomy, such as biochemistry and bioelectricity, which partially 
offset the negative impact on the more traditional sectors of biomass 
processing, namely agriculture and food processing (Lasarte-López et 
al., 2023).
2 Indeed, it was initially expected that farming experienced less direct 
effects, except where hired labor was important, although interlinkages 
with the other segments of the chain may have caused income losses 
and production disruption (Swinnen, 2020).
3 For instance, Cesaro et al. (2022) found that financial liquidity and 
repairment of equipment and machinery were the difficulties most 
reported by farmers in Italy in the short run.

in the central Rift Valley (Minten et al., 2020). These 
new activities challenge small farmers’ and small enter-
prises’ participation in the AFVC, which are compound-
ing with already existing structural constraints such as 
low access to credit and extension, weak labor market, 
and high transaction costs (Croppenstedt et al., 2003; 
Bryan et al., 2009; Asfaw et al., 2011; Harvest SA, 2012). 
In such a situation, the COVID-19 shock could push 
smallholder farmers and small and medium enterprises 
out of the market. 

The first case of COVID-19 in the country was 
reported on March 13th, 20204. In the same month, the 
federal government implemented a set of containment 
measures, such as school closure, physical distancing, 
and restrictions on gathering and transportation (Baye, 
2020). In April, a five-month state of emergency was 
declared, though economic activities continued to oper-
ate. Although farmers could keep working, they faced 
many challenges. With borders shut, imported inputs 
were more difficult to find and their price increased 
(Hirvonen et al., 2021b, 2021c, and 2021d). Moreover, 
restrictions on movement made it almost impossible 
for farmers to reach the markets. This eventually led to 
a drop in agri-food sales, particularly of some vegeta-
bles such as tomatoes, papaya, and watermelon (Molla, 
2020). The travel restrictions also doubled transport 
costs, with a further domino effect on production, rais-
ing the farmgate and retail prices of some products, such 
as tomatoes (Hirvonen et al. 2021b). Additionally, since 
many farmers could not store their goods – particularly 
perishable produce – they were forced to accept the low 
prices set by buyers (Ababulgu et al., 2022). Hired labor 
was also affected. Many rural workers returned to their 
homes and the reduced labor supply pushed up the costs 
of labor (Agajie, 2020). Effects were driven also by the 
fear of contagion. People associated raw vegetables with 
infection, reducing their purchases (Hirvonen et al., 
2021a; Tamru et al., 2020). This determined a significant 
reduction in local market sales as well as exports (Abab-
ulgu et al., 2022).

Although anecdotal evidence exists on the impacts 
of COVID-19 on AFVC participation and income, rig-
orous empirical studies based on household-level sur-
vey data are few. Josephson et al. (2021) used the World 
Bank phone-based surveys of Ethiopia, Malawi, Nigeria, 
and Uganda to document the socioeconomic impacts 
of the pandemic. They found that 77% of households 
across the four countries experienced an income loss in 
the immediate aftermath of the pandemic. However, the 
authors were not able to measure how much of the loss 

4 For details, see https://www.afro.who.int/news/first-case-covid-19-con-
firmed-ethiopia.

https://doi.org/10.36253/bae-13404
https://www.afro.who.int/news/first-case-covid-19-confirmed-ethiopia
https://www.afro.who.int/news/first-case-covid-19-confirmed-ethiopia


335The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

can be directly determined by the pandemic, given the 
descriptive nature of their analysis. According to this 
study, Ethiopian households are significantly less likely 
to experience an income loss compared to those from 
the other three countries. 

More recently, the same dataset has been used by 
Rudin-Rush et al. (2022) to document trends in food 
security over the twelve months after the onset of 
the COVID-19 pandemic. This study reports a sharp 
increase in food insecurity in the aftermath of the pan-
demic, with a subsequent gradual decline. Furthermore, 
rural households were more negatively affected than 
urban households in terms of food security. 

IFPRI conducted a series of monthly phone-based 
surveys between May and August 2020 (i.e., up to five 
months after the pandemic onset) interviewing near-
ly 600 households in Addis Ababa (Hirvonen et al., 
2021a). More than half of respondents reported a fall in 
income relative to their average pre-pandemic income at 
the same time of the year (Hirvonen et al., 2020), with 
the proportion of affected households increasing from 
May to July (Hirvonen et al., 2021a). Poorer households 
more likely reported income losses, with a significant 
worsening of household food security and nutritional 
status. Income loss and unemployment were identi-
fied as the most common shocks experienced by the 
respondents (Abate et al., 2020; de Brauw et al., 2020; 
Hirvonen et al., 2020). Despite income loss, Zhang et al. 
(2022) found that the population in Addis Ababa was 
not affected on average in terms of food security. How-
ever, the situation in other regions of the country was 
much different, especially in rural areas and among vul-
nerable individuals and households (Abay et al., 2023, 
Zhang et al., 2022).

Hirvonen et al. (2021b) relied on a large value chain 
survey administered by IFPRI in February 2020 and fol-
low-up phone interviews collected in May 2020 to ana-
lyze the disruption of the vegetable value chain from the 
main producing areas in the Central Rift Valley to Addis 
Ababa, including changes in prices and adjustments 
in the marketing activities of the participants – from 
farmers to wholesalers and retailers. They found that 
nearly 60% of the smallholders and more than 60% of 
the investors reported less income than usual. They also 
found that the pandemic in Ethiopia disrupted trade not 
only between neighboring countries but also among sub-
national geographies, thus determining high volatility 
in agricultural prices (Hirvonen et al., 2021b). However, 
they found that the changes in wholesale and retail mar-
keting margins were relatively low, suggesting a resilient 
response of the domestic food value chains during the 
pandemic in Ethiopia. 

Although these studies provided important early 
estimates of the effects of the pandemic on relevant indi-
cators of welfare, they present some limitations. Some 
of them are based on a non-representative sample. The 
study of Hirvonen et al. (2021d) focuses on the vegeta-
ble value chain only, while Hirvonen et al. (2021b) focus 
only on households living in Addis Ababa. Most of the 
existing studies focus only on one or a few points in 
time, failing to capture the evolving impact of COVID-19 
over time. Other studies look at the impact on employ-
ment, such as Khamis et al. (2021), but they do not spe-
cifically disaggregate the analysis across AFVC segments. 
Our study aims to address these limitations contributing 
to estimating the magnitude of AFVC disruption caused 
by the COVID-19 pandemic in Ethiopia over a relatively 
longer time (seven months from the pandemic onset) and 
looking specifically at differentiated impacts on various 
AFVC segments. It also helps to identify the main factors 
that contributed to offset the negative consequences of 
COVID-19 shock and to keep adequate levels of income 
for AFVC participants. Although the data present some 
limitations in terms of representativeness (cf. Section 
2), we think the findings emerging from this study are 
relevant not only because they provide a better under-
standing of the COVID-19 impact in Ethiopia, but also 
because they can contribute to a better management of 
COVID-19-like crises should they emerge in the future.

The paper is organized as follows. The next section 
describes the data used. Section 3 presents some descrip-
tive statistics, with specific reference to employment 
and income. Section 4 describes the empirical strategy 
adopted. Section 5 presents and discusses the results of 
the analysis. Section 6 concludes. 

2. DATA 

The analysis uses a seven-rounds longitudinal data-
set, which includes a baseline pre-pandemic face-to-face 
survey and six follow-up phone surveys. Pre-COVID 
data come from the 2018/19 Ethiopia Socioeconomic 
Survey (ESS), which is part of the World Bank’s Liv-
ing Standards Measurement Study – Integrated Surveys 
on Agriculture (LSMS-ISA). It covers all regions of the 
country and is representative at national, urban/rural, 
and regional levels. The other six rounds of data are part 
of the World Banks’s COVID-19 High-Frequency Phone 
Survey of Households (HFPSH) 2020. This phone-based 
survey is a 15-minute questionnaire administered to 
a subsample of the ESS 2018/19 households from April 
to mid-October (Figure 1). The World Bank team inter-
viewed the same households in each round, leading to a 

https://doi.org/10.36253/bae-13404


336

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

balanced dataset of 2,347 households5. To obtain unbi-
ased estimates, sampling weights at the household level 
have been constructed by the World Bank team follow-
ing Himelein (2014), thus having a sample that is repre-
sentative at the national and urban/rural levels. 

A major problem with the HFPSH surveys is that 
phone penetration in rural Ethiopia is still low, with 
only 40% of rural households having access to a phone. 
Therefore, data are representative only of those house-
holds that have access to phones in urban areas (90% of 
all urban households) and better-off rural households 
that have access to mobile phones (Wieser et al., 2020). 
However, these rural households are systematically dif-
ferent from the majority of rural households (Ambel et 
al., 2020a). Additionally, only one member per house-
hold – typically the household head or the spouse – has 
been interviewed, but household heads could systemati-
cally differ from the rest of the population, undermin-
ing the representativeness of the sample at the individ-
ual level6. 

5 Each COVID-19 HFPSH survey has a slightly different number of 
observations, ranging from 2,704 to 3,249 households. In order to have 
a balanced panel we reduced the sample to 2,347 observations. For 
more information on sampling design please visit https://microdata.
worldbank.org/index.php/catalog/3716.
6 Further discussion about this issue is presented in section 4.3. 

A key methodological concern is that factors other 
than the COVID-19 crisis could drive the evolution of 
outcomes over time. Specifically, month-to-month sea-
sonality could represent an issue. In principle, it can be 
controlled by including month fixed effects. However, 
this could not be done due to the different time reference 
between the baseline and phone surveys, especially for 
the employment variable. While the pre-COVID survey 
considers the employment activities over the preceding 
twelve months, including both planting and harvesting 
seasons, questions on employment in the phone sur-
veys consider only the week before the interview. There 
could be then an underestimation of the farming-related 
employment rate. Luckily, the phone survey covers the 
sowing and the harvesting periods of the two main crop 
seasons (Figure 1)7. Therefore, although it is not possible 
to fully rule problems of seasonality out, it is likely that 
it does not significantly affect our estimates. 

Seasonality can also bias the analysis because of its 
impact on farm income. There are two rainy seasons 
over the year: the small rainy season (belg), which occurs 
between March and May, and the main rainy season 

7 Only sugarcane and taro are neither planted nor harvested in the peri-
od under analysis. Source: http://www.fao.org/agriculture/seed/cropcal-
endar/welcome.do;jsessionid=62FFB1AC3CB6FA74244A91586E5E1758.

13 March

1st case

22 April
14 May

1st wave

4 June

2nd wave

26 June

3rd wave 4th wave

27 July
14 August

14 Oct

1st crop season (Belg - all cereals) 

growingsowing harvesting

5th wave

2k

1.5 k

1 k

500

0

17 Sept

6th wave

Daily COVID-19 cases

2st crop season (Meher – main season) 

sowing growing

harvesting

harvesting

Figure 1. Timeline with daily COVID-19 cases, surveys date, and crop seasons in Ethiopia. Source: data on COVID-19 daily cases retrieved 
from https://covid19.who.int/region/afro/country/et; information on crop seasons retrieved from https://www.prepdata.org/stories/ethiopia-
climate-and-agriculture; date of COVID-19 HFPSH data collection retrieved from https://microdata.worldbank.org/index.php/catalog/3716.

https://doi.org/10.36253/bae-13404
https://microdata.worldbank.org/index.php/catalog/3716
https://microdata.worldbank.org/index.php/catalog/3716
http://www.fao.org/agriculture/seed/cropcalendar/welcome.do;jsessionid=62FFB1AC3CB6FA74244A91586E5E1758
http://www.fao.org/agriculture/seed/cropcalendar/welcome.do;jsessionid=62FFB1AC3CB6FA74244A91586E5E1758
https://covid19.who.int/region/afro/country/et
https://www.prepdata.org/stories/ethiopia-climate-and-agriculture
https://www.prepdata.org/stories/ethiopia-climate-and-agriculture
https://microdata.worldbank.org/index.php/catalog/3716


337The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

(meher), which takes place between June and September8 
(Hirvonen et al., 2016). Around 90% of the total crop 
production is done during the meher season (Taffesse et 
al., 2013). Farmers usually run out of stock between July 
and September, which can result in increasing household 
food insecurity (Dercon and Krishnan, 2000; Hirvonen 
et al., 2016; Gilbert et al., 2017; Sibhatu and Qaim, 2017; 
Roba et al., 2019). However, seasonality-induced food 
shortage is quite homogeneous across farmers, and it 
is captured by a variable that controls for the aggregate 
time trend (cf. Section 4.3). 

Another factor to consider in the analysis is the 
desert locusts’ outbreak, i.e. the most destructive migra-
tory pests in the world (Cressman et al., 2016; Lazar 
et al., 2016), that swarmed from Yemen to the Horn of 
Africa in the summer of 2019. In the fourth round of 
phone surveys9, 45% of farmers self-reported that they 
experienced desert locusts on their farm, and 41% of 
households experienced locusts in their kebele10. Desert 
locusts have negative consequences on income because 
they destroy the crops and the fodder for livestock. 
Additionally, labor time is required to spray the chemi-
cals on the area under cultivation. 

3. DESCRIPTIVE STATISTICS

3.1. Employment

The first round of the phone-based survey asked 
if the individual did any work in the seven days before 
the interview, if the individual was working before the 
COVID-19 outbreak, and if the current work is the same 
as before the pandemic. For the other rounds of data, 
the questions were the same, but using as reference time 
the previous call. As shown in Figure 2, the employment 
rate experienced a significant reduction in the after-
math of the COVID-19 outbreak. Overall employment 
dropped by 11 percentage points. However, labor activi-
ties recovered quickly over the next months, exceeding 
the employment rate before COVID-19 (Ambel et al., 
2020b), driven by own farming activity. 

The dynamics of labor mobility are somehow dif-
ferent within the various AFVC segments11 (Table 1). 

8 This refers to the growing period of the season.
9 Information on desert locusts is available only in rounds 4 and 6. 
However, in round 6 very few respondents answered the questions relat-
ed to locusts, so it is not possible to produce reliable estimates. 
10 The kebele is the smallest administrative unit of Ethiopia, i.e. a neigh-
bourhood or a localized and delimited group of people consisting of at 
least 500 families.
11 The variable of labor participation in AFVC activities has been 
decomposed into three segments, namely: a) upstream (primary pro-

The upstream segment was quite stable, with 83% of 
people remaining in the same segment of employment 
and 12% moving towards non-AFVC activities after 
seven months. In the case of midstream activities, only 
26% remained in the same segment, while 39% moved 
towards non-AFVC activities, 23% moved to upstream 
activities, and the remaining 12% moved to downstream 
activities. Similarly, in the downstream segment, only 
27% on average did not change the segment of employ-
ment, while most of the people who did it, moved to 
midstream activities (49%). Finally, almost two-thirds 
of the ones who were not originally working in AFVC 
activities remained outside the AFVC, while the ones 
who entered the AFVC split mainly between midstream 
(14%) and upstream (18%) activities. 

Employment changes can be in part driven by sea-
sonality. Indeed, seasonal migration in Ethiopia occurs 
both from rural to urban areas, used as a coping strat-
egy during the dry season (Asefawu, 2022), and also 
towards northwest Ethiopia for temporary employment 
on large-scale agricultural farms during the rainy sea-
son (Schicker et al., 2015). However, respondents report-
ed that the main reason for stop working is COVID-19, 
especially in the early phone rounds. Between April and 
May (round 1), more than half of individuals stated that 
they lost their job because of the pandemic (Figure 3). 
In the last rounds instead, being “temporarily absent” is 
the main reason to stop working. This can be indirectly 
associated with the pandemic since many who tempo-
rarily left their job in the city migrated to rural areas12. 

duction, including farming, fishing, forestry and hunting), b) midstream 
(manufacturing of food products, including processing; wholesale and 
retail trade; transport; and distribution), and c) downstream (restau-
rants and bars).
12 Detailed information on the number of individuals that started to 
work again in each round, and the reason for having stopped working 

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

 Pre-Covid  Round 1  Round 2  Round 3  Round 4  Round 5  Round 6

 Total employment  Own farm  On farm wage  Off farm wage  Off farm self-employment

Figure 2. Evolution of employment in Ethiopia, 2018/19 – mid 
October 2020.  Source: Own elaboration from ESS 2018/2019 and 
HFPSH 2020. Note: Sampling weights applied. 

https://doi.org/10.36253/bae-13404


338

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

3.2. Income

Respondents to the phone-based survey were asked 
to assess the income change experienced by the house-
hold compared to the situation before the COVID-19 
outbreak in the first-round interview, and compared to 
the previous call in the subsequent rounds. The possi-
ble answers ranged from “total loss” to “income reduc-
tion”, “no change” and “income increase”. The categori-
cal nature of the question does not allow for comput-

in the previous round is reported Table A.1 in the Appendix.

ing accurate estimates of the magnitude of COVID-19 
impact on income, limiting the analysis to the qualita-
tive incidence of the pandemic (De Weerdt, 2008). In 
the case of farming income, the answer highly depends 
on the harvest time of cultivated crops. In fact, the bulk 
of crop sales occurs between December and Febru-
ary, though April usually records the largest sales (Hir-
vonen et al., 2016). Figure 4 (panel a) shows a general-
ized decreasing trend of the share of households that 
reported a reduction of income or a total loss between 
rounds, not only for farming income but also for other 

Table 1. Labor transition matrix across AFVC segments and non-AFVC, 2018/19 – mid-October 2020.

  N. Obs.

Round 6

Downstream Midstream Upstream Non-AFVC

Pre-Covid

AFVC Downstream 145 27.5 48.5 6.2 17.8
AFVC Midstream 184 12.4 26,0 22.9 38.8
AFVC Upstream 517 0.6 3.6 83.3 12.1

Non-AFVC 834 4.0 14.4 17.8 63.9

Source: Own elaboration from ESS 2018/2019 and HFPSH 2020.
Note: Upstream: agricultural production and agricultural employment, including fisheries, forestry, and hunting; Midstream: manufacturing 
of food products, including processing, trade, and transport; Downstream: restaurants and bars; Non-AFVC: all other employment activi-
ties. Sampling weights applied. 

0%

10%

20%

30%

40%

50%

60%

Covid-19 Seasonal/casual
worker

Contract ended Temporarily
absent

Retired Being ill Need to care for
ill relative

 Round 1  Round 2  Round 3  Round 4  Round 5  Round 6

Figure 3. Reason to stop working, share on round total. Source: Own elaboration from HFPSH 2020. Note: sampling weights applied. 

https://doi.org/10.36253/bae-13404


339The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

sources of income. If we compare the income change to 
the situation before the COVID-19 outbreak13 (Figure 4, 
panel b), the trend is different. The share of households 
that reported a reduction in income compared to the 
pre-COVID situation shrank only for non-farm business 
while it did not change significantly over time for other 
income sources. In the case of total income, the share 
of households experiencing a reduction/total loss even 
increases over time, up to 9 percentage points increase 
in the sixth round compared to the baseline. 

Table 2 reports the descriptive statistics of the out-
come variables per each round. Specifically, the employ-
ment variables show the rate of people employed in each 
sector, while the income variables report the share of 
households that experienced a reduction in income or a 
total loss compared to the baseline.

There are some differences between the overall base-
line sample and the phone-based sub-sample (Table 3). 
Respondents to the HFPS sample are mainly located in 
urban areas, the majority of them are males, and their 
employment rate is higher. They are generally older, more 
educated, and more employed through a formal job con-
tract. The rate of non-farm employment activities is high-

13 The change of income is computed backward up to the baseline. If, for 
instance, in round 2 income did not change compared to the previous 
round, and in round 1 it increased compared to the baseline, in round 
2 it also increased compared to the baseline. The change is assumed to 
occur with the same amount, therefore if a household first reports an 
increase, and then a reduction, the net effect is null. We are aware of the 
arbitrariness of this methodology. For this reason, the analysis has been 
also conducted round by round, finding similar results, as reported in 
the Appendix.

er compared to the baseline population. Vice versa, the 
rate of farm-related activities is similar and the same is 
true for the employment rate in the upstream segment.

Given these differences, the results of the analysis 
could not be generalized to the whole Ethiopian popu-
lation. To check for possible problems of representa-
tiveness, we ran a robustness check using individually-
adjusted weights (cf. Section 4.3). 

4. EMPIRICAL STRATEGY

4.1. Outcome variables

To assess the impact14 of COVID-19 on income 
and employment we estimated a household fixed effects 
model with a continuous treatment variable, adapting 
the approach implemented by Amare et al. (2020). We 
use two dependent variables, namely: participating in 
labor activities, considering any type of activity as well 
as specific sectors; and income change, looking at both 
the total income and the different income sources. 

Labor activities are grouped into own-farm, on-
farm wage employment, off-farm self-employment, and 
off-farm wage employment. We also consider employ-
ment according to segments of AFVC (i.e., downstream, 
midstream, and upstream activities as defined above) 
given the expected differentiated impact of COVID-19 

14 We use the terms “impact” and “effect” throughout the paper, but we 
acknowledge that we are not able to fully identify a causal mechanism 
with our estimation strategy due to the limitations described in Section 2.

a) Share of HHs experiencing income reduction or total 
loss, round-to-round. 

b) Share of HHs experiencing income reduction or 
total loss, comparison with the pre-COVID situation. 

  

 

 

10%

20%

30%

40%

50%

60%

70%

80%

90%

Round 1 Round 2 Round 3 Round 4 Round 5 Round 6

 Total income  Farming  Wage employment  Non-farm business

10%

20%

30%

40%

50%

60%

70%

80%

90%

Round 1 Round 2 Round 3 Round 4 Round 5 Round 6

 Total income  Farming  Wage employment  Non-farm business10%

20%

30%

40%

50%

60%

70%

80%

90%

Round 1 Round 2 Round 3 Round 4 Round 5 Round 6

 Total income  Farming  Wage employment  Non-farm business

Figure 4. Share of HHs experiencing income reduction or total loss per income source category, share by income source. Source: Own 
elaboration from ESS 2018/2019 and HFPSH 2020. Note: Sampling weights applied. 

https://doi.org/10.36253/bae-13404


340

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

related restrictions on different stages of the value chain 
(Reardon et al., 2020b, Swinnen and McDermott, 2020). 
For each labor activity, we computed a dummy equal 
to 1 if the individual operated in that activity, and zero 
otherwise. 

We consider total income and specific income-gen-
erating activities, namely family farming, non-farm fam-
ily business, wage employment of household members, 
and other sources of income (pension, remittances, etc.). 
The variables take the values -2 (total loss), -1 (income 
reduction), 0 (no change), and 1 (income increase).

4.2. Treatment variable

The main variable of interest is the number of con-
firmed cases of COVID-19 over the number of inhabit-
ants in each region. This information has been retrieved 
from the Ethiopia COVID-19 Monitoring Platform15 
and weekly governmental bulletins16. This variable cap-
tures the evolution and the spread of the virus across the 
country. The variable has been transformed using the 
inverse hyperbolic sine (IHS) transformation, to account 
for zero cases in the first post-COVID survey. Regres-

15 Available at this link: https://www.covid19.et/covid-19/.
16 See https://www.ephi.gov.et/.

sion results can be interpreted as the log transformation 
(Johnson, 1949; Burbidge et al., 1988). 

This variable presents some limitations. Firstly, the 
ratio of confirmed cases over the number of tests would 
do a better job than using the number of the total popu-
lation in each region, but unfortunately, data on testing 
disaggregated at the regional level are not available. Sec-
ondly, the number of confirmed cases probably underes-
timates the real infection level due to the limited testing 
capacity of the country17. Although the testing capacity 
is presumably unequal across regions, as access to basic 
health care in Ethiopia is highly unequal (Woldemichael 
et al., 2019; Alene et al., 2021), the use of fixed effects 
estimator (cf. Section 4.3) should partially mitigate the 
issue, controlling for differences across regions that do 
not vary over time. 

Thirdly, the number of confirmed cases does not 
adequately proxy the treatment variable, i.e. the vari-
ation in terms of access to the market and restrictions 
imposed by the government, which in turn affect labor 
participation and income. However, we can assume 
that as the number of confirmed cases increases in a 
region, both the restrictions imposed by the government 
and the individually self-imposed restrictions would 
increase. Indeed, data confirm that the economic and 
health effects of the pandemic covary in Ethiopia. When 
using daily data retrieved from the Oxford COVID-19 
Government Response Tracker (OxCGRT), the correla-
tion between the COVID-19 cases and the stringency 
index is positive and significant18. Although there could 
be a time lag between the implementation of the restric-
tions and the effect in terms of COVID-19 cases, this 
lag is shorter (7 to 14 days depending on the restric-
tion type and stringency as well as on the rate of infec-
tion of the specific COVID-19 variant) than the period 
analyzed in each round (i.e., one month). Therefore, the 
average effect of the restrictions over a month should be 
captured by the number of confirmed cases. It is also 
important to consider the heterogeneity of the response 
across the regions. Indeed, although measures were 
coordinated at the national level, each regional state in 
Ethiopia tailored policy implementation to the local situ-
ation through its own Public Health Emergency Opera-

17 The virus spread unheavenly across regions. In particular, the Addis 
Ababa region reported the highest proportion of cases per million pop-
ulation, followed by Harar and Dir Dawa. Factors that can explain this 
heterogeneity are a different testing capacity, driven by better infrastruc-
ture, especially in the capital and in other urban areas, population den-
sity, and degree of internal and international connectivity.
18 Similar results were found in other countries. For instance, Amare et 
al. (2021) in Nigeria found that the variables of COVID-19 cases and 
government restrictions produced the same results, confirming that the 
two variables can proxy each other.

Table 2. Descriptive statistics of employment and income outcomes.

Round

Baseline 1 2 3 4 5 6

Employment: % of individuals
Total employment 75% 64% 85% 86% 87% 88% 88%
Downstream 4% 1% 2% 2% 2% 2% 2%
Upstream 40% 37% 55% 55% 56% 58% 57%
Midstream 5% 7% 8% 8% 8% 8% 9%
Out of FVC 25% 18% 20% 21% 21% 20% 20%
Own farm 39% 36% 53% 53% 54% 56% 56%
On-farm wage 1% 1% 2% 2% 2% 2% 2%
Off-farm self-employment 28% 18% 20% 20% 21% 20% 20%
Off-farm wage employment 19% 9% 10% 10% 11% 10% 10%

Income: % of households
Total income 56% 67% 70% 72% 72% 71%
Farming 42% 50% 51% 47% 45% 41%
Wage employment 36% 36% 34% 35% 36% 33%
Non-farm business 86% 82% 82% 76% 68% 65%

Source: Own elaboration from ESS 2018/2019 and HFPSH 2020.
Note: Employment variables report the share of people employed 
in each round. Income figures show the share of households that 
reported income reduction or total loss compared to the baseline. 
Sampling weights applied.

https://doi.org/10.36253/bae-13404
https://www.covid19.et/covid-19/
https://www.ephi.gov.et/


341The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

tions Centre (PHEOC)19. This calls for using regionally 
disaggregated variables. 

Fourthly, the number of confirmed cases does not 
capture spillover effects that may occur across regions. 

19 Source: https://www.acceleratehss.org/wp-content/uploads/2022/03/
Covid-Collaborative-Ethiopia-Case-Study.pdf 

Indeed, each region is treated as an independent entity 
assuming that each of them does not have interactions 
with the rest of the country and no aggregate impacts 
occurred. This assumption does not hold when two or 
more regions have strong economic relationships. For 
instance, this may happen when a food value chain 
crosses over regional boundaries – e.g. a food item is 
produced in a region and consumed in another – or 
workers commute between different regions. In these 
cases, should one region be affected differently than 
others, this effect would affect not only that specific 
region, but also other geographically closer or economi-
cally linked regions. However, as regions in Ethiopia are 
quite large and people are mostly working in the local 
economy (e.g. high share of family farmers), the spillo-
ver effect should be limited. Additionally, the Ethiopian 
political system based on ethnic federalism, where the 
regions have been identified based on “settlement pat-
terns, identity, languages” (Article 46.2 of the Ethiopian 
Constitution), makes it easier to conceptualize regions 
as separate economies. Evidence indeed shows that labor 
mobility and internal migration in Ethiopia are limited 
(Bundervoet, 2018) because migration across regional 
boundaries often creates social tensions and violence 
(Breines, 2020; Fessha and Dessalegn, 2020). 

4.3. Model specification

The base model is the following: 

yhrt = αhr + β0Timet + β1(Casesr * Timet) + εhrt (1)

where yhrt is the outcome variable – either labor or 
income – defined for each household h in region r and 
round t; αhr captures household fixed effects, allowing 
controlling for unobserved time-invariant heterogene-
ity among households; Casesr is the number of con-
firmed COVID-19 cases per million population in each 
region; Timet is a dummy variable representing the time 
of observation, equal to 1 for the post-COVID round 
and 0 for the pre-COVID round, whose coefficient cap-
tures the aggregate time trend in the labor market and 
income composition; the interaction term between time 
and the number of cases captures the differential impact 
of COVID-19 on labor participation and income change 
across regions due to different exposure to the virus; εhrt 
is the error term. 

Considering that the virus spread unevenly across 
regions over time, we need to control for this. Regions 
that experienced the virus earlier are indeed more likely 
to report more cases than the other regions. A first spec-
ification of the base model introduces the variable Day1r, 

Table 3. Comparison of individual characteristics between the base-
line sample and phone-based subsample.

Variable
Baseline 

population
Phone-based 
sub-sample

Student’s t 
significance

Rural 0.72 0.64 ***
(0.45) (0.48)

Sex: 1=female 0.51 0.27 ***
(0.50) (0.45)

Employed in any activity 0.75 0.85 ***
(0.43) (0.35)

Age 30.69 38.33 ***
(16.38) (13.76)

Not engaged in Education, 
Employment or Training 0.10 0.11

(0.30) (0.31)
Literacy rate 0.55 0.63 ***

(0.50) (0.48)
Formal job contract 0.04 0.10 ***

(0.19) (0.30)
Years of education 3.70 4.75 ***

(4.32) (5.12)
Agricultural wage work 0.01 0.01

(0.09) (0.09)
Non-farm self-employment 0.10 0.15

(0.29) (0.36)
Non-farm wage work 0.12 0.22 ***

(0.32) (0.42)
Own farm work 0.63 0.63 ***

(0.48) (0.48)
Upstream of AFVC 0.63 0.64

(0.48) (0.48)
Midstream of AFVC 0.03 0.04 ***

(0.16) (0.20)
Downstream of AFVC 0.01 0.01 **

(0.10) (0.12)
N. of observations 19,910 2,347

Note: the first column includes all individuals at the baseline. The 
second column includes only individuals from the baseline who 
were tracked in the phone-based surveys. Sampling weights applied. 
Standard deviation in parenthesis. Children below 11 years old 
dropped from the sample. Mean difference is computed through a 
linear regression, where the independent variable is a dummy equal 
to one if the individual belongs to the phone subsample. *** p<0.01, 
** p<0.05, * p<0.1.

https://doi.org/10.36253/bae-13404
https://www.acceleratehss.org/wp-content/uploads/2022/03/Covid-Collaborative-Ethiopia-Case-Study.pdf
https://www.acceleratehss.org/wp-content/uploads/2022/03/Covid-Collaborative-Ethiopia-Case-Study.pdf


342

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

which reports the number of days that occurred from 
the first COVID-19 case at the national level to the first 
COVID-19 case registered in the region:

yhrt = αhr + β0Timet + β1(Casesr * Timet) +
β2(Day1r * Timet) + εhrt

 (2)

To differentiate the impact of the isolated interac-
tions and the impact of the combined spatial and tem-
poral variabilities, we consider also a specification that 
includes the triple interaction between the time dummy, 
the number of confirmed cases per million inhabitants, 
and the variable as follows:

yhrt = αhr + β0Timet + β1(Casesr * Timet) +
β2(Day1r * Timet) + β3(Casesr * Day1r * Timet) + εhrt

 (3)

As an additional specification, we include in (3) 
some control variables available in the phone-based 
post-COVID surveys, which are not captured by the 
fixed effects. These variables are the presence of another 
member in the household who lost a job in the after-
math of the pandemic, and if the household received any 
assistance since the outbreak of the pandemic.

The analysis has been conducted for each post-COV-
ID round, comparing it with the baseline. In this way, 
it is possible to observe the evolution of the response to 
the crisis over time. We expect that regions more affect-
ed by the pandemic will report a higher reduction in 
labor participation and income and that the effect will 
increase the pandemic deepening over time20. 

We used a linear probability model with household 
fixed effects. The advantage of this model compared to 
a logit or conditional logit model with fixed effects is 
the inclusion of all observations. In fact, the logit mod-
el with fixed effects drops the units that show no vari-
ability in the dependent variable (Beck, 2018), drastically 
reducing the number of observations in case of small 
variability. In our data, this would result in an 80% 
reduction of the sample size. 

All analyses have been carried out using the bal-
anced sample. However, given the existence of signifi-
cant attrition rates, we replicated the analysis also using 
the unbalanced sample, finding consistent results (cf. 
Figures A.2 and A.3 in the Appendix). 

An important issue that could have affected our esti-
mates is the desert locust outbreak experienced by some 
regions of the country in the period of analysis, which 

20 We also estimated the impact of COVID-19 from wave to wave, com-
paring the outcome with the previous interview and results still hold 
(cf. Figure A.1 in the Appendix).

might have harsh consequences on production21. For this 
reason, it is important to consider this shock on farm-
ing employment and income. The HFPH surveys report 
information on desert locust outbreak only in the 4th 
wave. We retrieved GIS data on desert locusts from the 
FAO Locusts Hub22 and merged it with the households’ 
location. Given that the household coordinates refer to 
the dwelling, and not to the parcel, and they are slight-
ly modified for privacy reasons, we created a buffer of 3 
km around the household centroid to account for these 
factors (on average the parcel is 1.7 km distant from the 
dwelling). Regarding the location of locusts, we consid-
ered the area surveyed, which is 580 hectares on aver-
age. Figure 5 reports the location of households (in pur-
ple) and where the desert locusts have been observed (in 
green) over the year 2020. 

Although GIS data are quite accurate and reliable, 
many data gaps undermine the quality of the informa-
tion and might represent a limitation of our analysis. 
Firstly, household coordinates have been slightly modi-
fied for privacy reasons and this might determine some 
measurement bias. Secondly, only the distance of the 
parcel is available, not the direction: it is not possible to 
know exactly where it is located. Thirdly, the informa-
tion provided for locusts does not account for the locust 
swarm movements over time, excluding areas outside 
sampled locations. For these reasons, self-reported infor-
mation could be more reliable to measure the effect of 
these pests on farm crops. Therefore, we report estimates 
of the impact of locust outbreak using GIS data as well 
as self-reported data (cf. Section 5.1).

The second part of the analysis aims at identifying 
the main determinants influencing changes in income in 
the presence of COVID-19. In doing this, we use a prob-
ability model with regressors in time t (pre-COVID) and 
the dependent variable in time t + 1 (post-COVID). In 
this way, we can estimate which attributes that were in 
place in normal conditions are more likely to affect the 
outcome during the pandemic. The probability that the 
outcome variable takes a certain value is given by

Prob(yh,t+1 = j) = xT
h,tβc+ uh,t+1 (4)

where h is the household, x is a column vector of observ-
able variables, namely the attributes and factors in time 
t, uh,t+1 is the error term, and j takes the value 1 if the 

21 The desert locust outbreak appeared in Ethiopia in second half of 
2019. By January 2020, the outbreak was already significantly affecting 
the country, peaking around mid 2020 around the harvesting time of 
the first crop season (belg) when most cereals were ready to be harvest-
ed (see Figure 1). According to FAO, the outbreak was the worst in 25 
years in the country.
22 https://locust-hub-hqfao.hub.arcgis.com/. 

https://doi.org/10.36253/bae-13404
https://locust-hub-hqfao.hub.arcgis.com/


343The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

outcome is dichotomous, or multiple values if it is cate-
gorical. The regressors include household characteristics, 
level of infrastructure and variables at the community 
level, economic-related variables, and agricultural-relat-
ed variables when considering farm income. 

The dependent variable is the change in income 
at the household level. We decided to not consider the 
employment status because there could be problems of 
endogeneity because of omitted variable bias. This could 
occur mainly by external factors, for which information 
is not provided in the survey. An example could be the 
loss of job due to the employing company shut-down. 
In addition to econometric issues, as the job loss mainly 
depends on factors beyond household or individual con-
trol, investigating the household-related determinants 
of the loss of employment due to the COVID-19 crisis 
would make little sense. 

The estimation has been conducted using a maxi-
mum likelihood estimator. We used the ordered pro-
bit model to account for the categorical nature of the 

dependent variable. However, given that the response 
rate for total loss and income increase was very low, 
we also created a dummy equal to 1 if income did not 
change or increase, and 0 otherwise. In this case, we 
used a probit model. 

5. RESULTS

5.1. Impact of COVID-19 cases

The impact on employment

Table 4 reports the impact on employment at round 
1 as resulting from the different model specifications23, 

23 Table 4 reports the estimation for round 1 as an example. Then we 
provide a visual estimation of our model results (e.g. Figure 5) that 
makes easier understanding the evolution of the relevant outcomes over 
the analyzed period. The estimates of each model are available upon 
request to the authors.

Figure 5. Map of households’ location (purple circles) and locust swarm sites (green circles), 2020. Source: own elaboration using data from 
FAO Locusts Hub and ESS 2018/2019.

https://doi.org/10.36253/bae-13404


344

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

starting from model (1), which is a simple OLS over the 
pooled sample, to model (5), which includes all the vari-
ables and their interaction terms, the individual/house-
hold fixed effects, and the controls. We consider the 
last model as the best suited model for the analysis. In 
fact, from the theoretical viewpoint, the within estima-
tor of the fixed effects model is robust to many types of 
omitted variable bias24. Furthermore, the inclusion of all 
regressors in model 5 allows controlling for more vari-
ables and provides insights on the role of such controls 
in determining the outcome variables. This also leads to 
higher adjusted R-square statistics as shown in Table 4.

As expected, our variable of interest, i.e. the interac-
tion term Cases*Time, has a negative sign and is statis-
tically significant, meaning that COVID-19 negatively 
impacted employment, while the other interaction terms 
are not statistically significant.

Figure 6 reports the coefficient of the interaction 
term between the time trend and the COVID-19 cases 
for each round, firstly considering any labor activities 
and then looking at specific sectors or segments of the 
AFVC. These results show how COVID-19 negatively 
impacted employment activities in Ethiopia. They also 
show that the severity of the impact increased over time. 

24 However, it is more inefficient than an OLS estimator, because it 
reduces the variation of the independent and dependent variables used 
for estimation. Indeed, it is more affected by measurement errors and by 
omitted variables that are not constant within the household.

Decomposing the impact along the AFVC, we can see 
that upstream activities are the most affected. Although 
this segment had initially been relatively less affected, 
it shows increasingly negative impacts in subsequent 
rounds. Downstream and midstream segments have 

Table 4. Regression results over different models, employment – round 1.

Dependent variable: individual employed in any activity  

Variable (1) (2) (3) (4) (5)

Time -0.0684*** -0.0758*** -0.0657*** -0.0658*** -0.0709***
  (0.0137) (0.0127) (0.0185) (0.0193) (0.0196)
Cases*Time -0.0438*** -0.0353*** -0.0362*** -0.0361*** -0.0360***
  (0.00866) (0.00577) (0.00607) (0.00651) (0.00654)
Days*Time -0.000395 -0.000386 -0.000364
  (0.000505) (0.000644) (0.000640)
Cases*Days*Time -9.72e-06 -1.53e-06

(0.000383) (0.000383)
Constant 0.746*** 0.746*** 0.746*** 0.746*** 0.746***
  (0.0163) (0.00507) (0.00507) (0.00507) (0.00507)
Controls No No No No Yes
Fixed effects No Yes Yes Yes Yes

Observations 4,694 4,694 4,694 4,694 4,694
R-squared 0.042 0.071 0.082 0.107 0.116
Number of pid   2,347 2,347 2,347 2,347

Note: Estimates are computed using a linear probability model. Sampling weights applied. Standard errors clustered at the household level. 
*** p<0.01, ** p<0.05, * p<0.1. Data refers to the 1st wave.

Figure 6. Impact of COVID-19 cases on employment over time. 
Source: Own calculation from ESS 2018/2019 and HFPSH 2020. 
Note: Dependent variable = dummy equal to 1 if the individual is 
employed. Dots are coefficients estimated from a linear probabil-
ity model with household fixed effects. Each post-COVID round 
is compared with the baseline. Bars are 95% confidence intervals. 
Sampling weights applied. Standard errors clustered at the house-
hold level. *** p<0.01, ** p<0.05, * p<0.1. 

https://doi.org/10.36253/bae-13404


345The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

also been negatively affected, but in this case, the impact 
did not significantly change over time. In the case of 
non-AFVC, after an initial negative impact, the coeffi-
cients became no longer significant from the third round 
onwards. This could mean that the COVID-19 cases no 
longer had an impact or that employment effects within 
this category offset each other. For instance, among the 
off-farm self-employment, construction and manufac-
turing reported a positive effect, while trade and restau-
rants, hotels, and bars showed negative coefficients. 

The impact on income

Table 5 shows the results of the various models esti-
mating the impact of COVID-19 on income change. 
Again, the interaction term Cases*Time is negative and 
most of the time statistically significant25. 

The impact of COVID-19 on income (Figure 7), 
takes more time to occur. Households indeed can rely on 
savings or other coping strategies in the short run. How-
ever, from the third round onwards total income has 
been negatively affected by COVID-19 cases, and, simi-
larly to employment, the effect increases over time. Wage 
income and off-farm business income do not seem to 

25 The coefficient loses significance when the triple interaction term is 
added. However, from the third round onwards it is statistically signifi-
cant. 

have been significantly affected, while it is interesting to 
see the impact on farm family farming. After an initial 
positive effect, in the last three rounds COVID-19 cases 
have significantly and negatively impacted farm income. 
This can be explained because initially, the virus spread 

Table 5. Regression results over different models, income – round 1.

Dependent variable: change in total HH income    

Variable (1) (2) (3) (4) (5)

Time -0.567*** -0.567*** -0.544*** -0.558*** -0.549***
  (0.0274) (0.0274) (0.0374) (0.0404) (0.0412)
Cases*Time -0.0246** -0.0246** -0.0266** -0.0157 -0.0148
  (0.0112) (0.0112) (0.0114) (0.0118) (0.0119)
Days*Time -0.000879 5.95e-05 1.58e-05
  (0.00110) (0.00162) (0.00161)
Cases*Days*Time -0.000967 -0.000970

(0.000864) (0.000864)
Constant 0 -0 -0 -0 0
  (3.08e-10) (0.0106) (0.0106) (0.0106) (3.08e-10)
Controls No No No No Yes
Fixed effects No Yes Yes Yes Yes

Observations 4,691 4,691 4,691 4,691 4,691
R-squared 0.336 0.503 0.503 0.504 0.505
Number of pid   2,347 2,347 2,347 2,347

Note: Estimates are computed using a linear probability model. Sampling weights applied. Standard errors clustered at the household lev-
el.*** p<0.01, ** p<0.05, * p<0.1. Data refers to the 1st wave.

Figure 7. Impact of COVID-19 cases on income over time. Source: 
Own calculation from ESS 2018/2019 and HFPSH 2020. Note: 
Dependent variable = categorical variable of income change, ranging 
from -2 (total loss) to 1 (increase). Dots are coefficients estimated 
from a linear probability model with household fixed effects. Each 
post-COVID round is compared with the baseline. Bars are 95% 
confidence intervals. Sampling weights applied. Standard errors clus-
tered at the household level. *** p<0.01, ** p<0.05, * p<0.1. 

https://doi.org/10.36253/bae-13404


346

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

in the cities, marginally hitting farmers livelihood in 
rural areas. Then the virus spread across the whole 
country, affecting also people located in more remote 
areas. Additionally, if initially smallholders and subsist-
ence farm households were more advantaged against the 
measures implemented by the government because they 
relied less on external inputs and markets, this advan-
tage disappeared over time, due to the limited coping 
mechanisms they had available. 

The impact of locust outbreak on farm employment and 
income

The inclusion of the dummy for respondents who 
self-reported to have experienced the desert locust shock 
on the farm has a significant impact on changing the 
coefficients associated with the number of COVID-19 
cases. Results are reported in Table 6. For employment, 
the coefficient of the COVID-19 cases loses significance, 
while having locusts on the farm is positively and sig-
nificantly associated with labor activities. This confirms 
the additional labor time required to spray the chemicals 
all over the land. Regarding income, compared to previ-
ous results, where the coefficient of COVID-19 cases was 

-0.621, the inclusion of desert locusts increases the mag-
nitude of the coefficient to -1.103, strengthening the neg-
ative impact of COVID-19 cases on farm income. These 
results show that it is important to consider multiple 
shocks experienced by individuals and households when 
assessing the impact of a certain event. 

When using the georeferenced data (Table 7), the 
locust variable loses significance for own farm labor 
activities. Instead, the impact of locust outbreak is sig-
nificant and negative in the case of farm income. The 
effect of the locust dummy is larger in the 4th wave 
(-0.377), which corresponds to the most damaging peri-
od for crops, given the locust life cycle as well as the 
timing of the crop season (peak harvesting in the first 
crop season, cf. Figure 1). The inclusion of the locusts’ 
data over all six waves does not significantly affect the 
impact of COVID-19 cases on farm income, showing 
only slight changes from the model not including the 
locust dummy estimates.

5.2. Determinants of income change

In this section, the results of the regressions aimed 
to identify the main determinants of income change are 
presented. Regressors have been grouped into three cat-
egories: household characteristics, infrastructures, and 
economic-related variables. As dependent variables, we 
considered the change in total and farm incomes. For 
illustrative reasons, this section reports only the results 
of the models using a dichotomous dependent variable. 
The estimates of the ordered probit model are reported 
in the Appendix (Tables A.4 and A.5).

Total income change

Figure 8 reports the estimated coefficients of house-
hold characteristics over the six rounds. The only signifi-
cant variable here is the level of education of the house-
hold head. A higher level of education is positively asso-
ciated with a higher probability of not experiencing an 
income reduction/total loss. Living in rural areas shows 
a positive and significant coefficient only in the first 
round, consistent with previous analyses that show that 
rural areas were initially less affected. 

Economic-related variables (Figure 9) show some 
interesting patterns. Having a formal job contract is asso-
ciated with a higher probability of income increase or 
unchanged income level. A similar relationship can be 
found with having a bank account and formal insurance, 
although the magnitude and the level of significance are 
lower than in the case of a formal contract. These results 

Table 6. Simultaneous impact of locusts (self-reported data) and 
COVID-19 on own farm employment activities and farm income 
change, 4th round.

Employed in own 
farm activities Farm income change

Time 0.0489 5.242***
 (0.504) (1.893)
Cases*Time 0.0216 -1.103***
 (0.0938) (0.372)
Days*Time -0.0237** -0.0998***
 (0.0115) (0.0350)
Days*Time*Cases 0.00333* 0.0194***
 (0.00200) (0.00671)
Locusts on the farm 0.134* -0.0244
 (0.0685) (0.110)
Constant 0.542*** -0
 (0.00927) (0.0111)
Controls yes yes
FE yes yes
Observations 2,961 2,639
R-squared 0.088 0.309
Number of pid1 2,347 2,347

Note: Estimates are computed using a linear probability model. Sam-
pling weights applied. Standard errors clustered at the household 
level.*** p<0.01, ** p<0.05, * p<0.1. Data refers to the 4th wave.

https://doi.org/10.36253/bae-13404


347The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

show that access to formal institutions is a winning strat-
egy to contrast the negative consequences caused by the 
crisis. Per capita household income reports a positive 
relationship, meaning that as per capita income increases 
the probability of not experiencing an income reduction 
increases. Richer households are then expected to suffer 
less from the crisis. However, the magnitude of the coef-
ficient is quite small, suggesting that the differential effect 
between poorer and richer households is limited.

Regarding infrastructure variables, none of them 
has a substantial effect on total income (Figure 10). 
Being distant to the urban center, to the main road, or 
to the markets seems to be slightly positively associated, 
sometimes in a significant way, with the probability of 
income increase or unchanged. However, the coefficient 
is lower than 1%. 

Farm income change

The same variables considered in the previous sec-
tion show partly different patterns when considering 
farm income. Looking at the household characteris-
tics (Figure 11), the education of the household head no 
longer seems to play a relevant role, while the household 
size and the age of the household head are associated 
with a higher probability of income reduction, although 
the effect is statistically significant only in a few rounds. 

Table 7. Simultaneous impact of locusts (GIS data) and COVID-19 on farm income change.

  wave 1 wave 2 wave 3 wave 4 wave 5 wave 6

Time -0.377*** -0.981*** -1.163*** 2.829*** 3.007*** 5.228***
  (0.0627) (0.269) (0.363) (0.895) (1.141) (1.273)
Cases*Time -0.0217 0.277** 0.254** -0.620*** -0.519*** -0.815***
  (0.0564) (0.134) (0.129) (0.174) (0.182) (0.196)
Days*Time 0.00110 0.0131** 0.0189** -0.0531*** -0.0581** -0.103***
  (0.00267) (0.00638) (0.00858) (0.0169) (0.0250) (0.0273)
Cases*Days*Time -0.00175 -0.00621** -0.00654** 0.0104*** 0.00923** 0.0153***
  (0.00185) (0.00281) (0.00272) (0.00310) (0.00365) (0.00393)
Locust dummy -0.307*** -0.350*** -0.0973 -0.377*** 0.0327 -0.00324
  (0.104) (0.129) (0.156) (0.144) (0.245) (0.213)
Constant 0.00328 0.00398 0.00131 0.00455 -0.000442 4.29e-05
  (0.0114) (0.0126) (0.0111) (0.0131) (0.0163) (0.0139)
Controls yes yes yes yes yes yes
FE yes yes yes yes yes yes
Observations 3,025 2,882 2,850 2,853 2,844 2,843
R-squared 0.386 0.415 0.384 0.225 0.102 0.099
Number of pid1 2,347 2,347 2,347 2,347 2,347 2,347

Note: Dependent variable: categorical variable of income change, ranging from -2 (total loss) to 1 (increase). Estimates are computed using 
a linear probability model with household fixed effects. Sampling weights applied. Standard errors clustered at the household level.*** 
p<0.01, ** p<0.05, * p<0.1.

Figure 8. Effects of households’ characteristics on total income 
change over time. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
total income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

https://doi.org/10.36253/bae-13404


348

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Even in the case of farm income (Figure 12), dis-
tance does not show significant effects, except for dis-
tance to a large market, where it seems that the more 
distant the household the higher the probability of farm 
income unchanged or increased. This result may look 
counterintuitive. A possible explanation could be that 
more (economically) isolated households had already put 
in place some strategies to account for the distance from 
large markets, so they were more advantaged relatively 
to those farmers who were used to relying on markets. 
Additionally, given the travel restrictions, domestic food 
value chains could have reshaped to adapt to the new 
situation, shortening their lengths. In this way, people in 
remote areas relied more on locally produced agricultur-
al products instead of going to the main urban market. 

The role of microfinance institutions in the commu-
nity is interesting. Indeed, differently from total income, 
here it shows a positive coefficient, and in the last rounds 
the effect is also statistically significant. This means that 
this type of institution matters in times of crisis. 

As in the total income case, having a bank account 
and formal insurance rise the probability of farm 
income increase (Figure 13). Having a formal job con-
tract does not show a statistically significant effect on 
agricultural income. This makes sense given that most 
households in Ethiopia run family farms and do not 
participate in the formal labor market. 

Regarding the agricultural-related variables (Figure 
14), results seem to suggest that farmers with larger are-
as of land have a higher probability of success compared 
to smallholders, as shown also by the marginal effects 
of land size on the probability that farm income did not 
decrease (Figure A.4 in the Appendix). This result is in 
contrast to findings from other studies conducted in dif-
ferent contexts. Cesaro et al. (2022), for example, found 
that medium-large farms in Italy expressed greater con-
cern about the negative consequences of COVID-19 in 
the short term than small farms. Having a land owner-
ship title or holding the right to use it played an impor-
tant role in cushioning the negative COVID-19 impact. 
Households that use fertilizers and those that have agri-
cultural machinery, although they initially experienced a 
positive or insignificant effect, were subsequently nega-

Figure 9. Effects of economic-related variables on total income 
change over time. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
total income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

Figure 10. Effects of infrastructure variables on total income 
change over time. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
farm income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

https://doi.org/10.36253/bae-13404


349The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

tively affected. This result can be the consequence of the 
mobility and trade restrictions, which decreased inputs 
availability and increased their prices. A less diversified 
crop mix was detrimental to farm income increase in 
early rounds, as shown by the coefficient of the Herfind-
ahl index of crop26. 

5.3. Robustness Checks

Placebo test

To test the validity of the treatment variable used in 
the analysis, we ran a placebo test, imputing the COV-
ID-19 shock in the prior wave of the ESS, collected in 

26 The Herfindahl index is a measure of crop concentration. It is com-
puted as the sum of square of the proportion of individual crop groups 
in a portfolio. The index decreases with an increase in diversification. It 
ranges from 0 (complete diversification) to 1 (complete specialization) 
(Singh et al., 2018).

2015/2016, and considering as baseline the 2012/2014 
ESS survey. If the variable of the number of COVID-19 
cases correctly captures the impact of the COVID-19 
shock, we should not find any significant effect, given 
that at that time the shock did not occur. 

Table 8 reports the results of the test, applied for the 
change of total income at the household level and total 
employment at the individual level. The variable is valid 
when applied to the model of household income, where 
none of the coefficients related to COVID-19 is signifi-
cant. Instead, when running the same model on total 
employment, the coefficient of the interaction between 
time and COVID-19 cases is significant (column 1). 
However, the sign is positive, in contrast to the predicted 
effect that the shock should have. A possible explana-
tion is that the variable of COVID-19 cases is in a way 
correlated with regional characteristics. For instance, 
we know that COVID-19 has affected some economic 
sectors more than others and, if a region is specialized 
in one sector, this correlation will be significant. If the 
employment rate was expanding between 2014 and 2016 
in that specific sector, the correlation would be positive. 

Figure 11. Effects of households’ characteristics on farm income 
change over time. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
total income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

Figure 12. Effects of infrastructure variables on farm income 
change over time. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
farm income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

https://doi.org/10.36253/bae-13404


350

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Introducing regional income (column 2)27 indeed makes 
the interaction term not significant. 

Inverse probability weights 

To address the problem of representativeness of the 
individual sample, we created individual-level adjust-
ed weights using the inverse probability based on the 
ESS 2018/2019, and we compared the outcomes using 
these weights following Khamis et al. (2021)28. We ran 

27 Regional income can capture the level of economic development of 
the region, which is in turn correlated with other factors, including the 
economic sector.
28 Khamis et al. (2021) relied on the World Bank’s Global Monitoring 
Database. Although they found similar results when applying the cor-
rected weights compared to the original ones, they had a limited set of 
variables available to use for reweighting the estimates, undermining the 
effectiveness of the weights created. In our case, instead, we can con-
sider more variables, increasing the ability to effectively adjust for the 

a logit regression to estimate the probability of being in 
the HFPS subsample over a set of variables at the indi-
vidual level, weighted by the household weights of ESS 
2019. Variables include age, gender, years of completed 
education, living in rural areas, income quintile, being 
employed, working in own farm activities, and being not 
engaged in education, employment or training (NEET). 
Children below 12 years old have been excluded. The 
inverse of the estimated probability is the adjusted 
weight. This procedure gives greater weight to obser-
vations that appeared in the HFPS sample. Figure 15 
reports the coefficients estimated with original weights 
vis-à-vis the adjusted ones. The correlation of the esti-
mates using the two methods is very high, i.e. 98%. This 
result suggests that the labor market outcomes of the 
subsample of individuals are generally consistent with 
the outcomes of the whole working population. 

6. CONCLUSIONS

The analysis showed that COVID-19 negatively 
impacted both household employment and income, the 
more so the longer the time length from the pandemic 

differences between the individuals in the subsample and the rest of the 
population. 

Figure 13. Effects of economic-related variables on farm income 
change over time. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
farm income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

Figure 14. Effects of agricultural-related variables on farm income 
change over time.cSource: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 if 
farm income did not decrease compared to pre-Covid round. Dots 
are average marginal effects from a probit regression. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Robust Standard errors. 
*** p<0.01, ** p<0.05, * p<0.1. 

https://doi.org/10.36253/bae-13404


351The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

onset. Upstream activities, and specifically own farm-
ing, are the most affected segment of the AFVC. Indeed, 
despite an initial positive effect, the impact then became 
negative and increased in magnitude over time. This 
finding is partly in line with previous studies published 
in the immediate aftermath of the pandemic (Bunder-
voet and Finn, 2020; Reardon et al., 2020a) that show 
that farming was the less affected sector. However, track-
ing the impact over time allowed gaining a more com-
plete understanding of the evolution of the effect, with 
farming increasingly severely affected by the disruption 
of the food value chain. The initial resilience capacity 
of the Ethiopian food marketing systems, as reported 
by Hirvonen et al. (2021b) for the vegetable value chain 
does not seem to persist over time. This highlights the 
importance of monitoring the evolution of the impact of 
the shock over time. Indeed, considering only the initial 
effect could give an incomplete and misleading under-
standing of the actual situation. 

We also showed that the most vulnerable farmers 
have been hit hardest. Small farming households are 
more exposed to the negative consequences of the crisis. 
There is the need then to target specifically this group 
of AFVC actors, especially in situations of crisis. To do 

this, AFVC participants need to have access to specific 
tools that allow them to cope with the shock. Access 
to formal institutions, such as formal insurance, bank 
account, formal contract, and land title are all positively 
associated with a higher probability of income increase. 
The national government should then increase its effort 
in providing improved opportunities to access financial 
services as well as formal institutions. 

Last but not least, multiple shocks dramatically 
worsen the picture. This is the case of the desert locust 
outbreak, that compounded with an already difficult 
situation created by COVID-19. Therefore, policymakers 
should consider the effects of simultaneous shocks when 
designing policy responses to the crisis. 

From our data, it is not possible to identify how the 
above impacts may affect other important dimensions 
of well-being such as food security. Abay et al. (2023) 
found that household food insecurity increased by 11.7 
percentage points. The authors did not assess the rela-
tionship between a reduction in employment/income 
and food security. However, there is evidence from other 
studies that a reduction in employment and income may 
or may not affect food security. Especially when subjec-
tive estimates of income change are used, the relation-
ship is not straightforward. In Hirvonen et al. (2021a), 
for example, self-reported income shocks did not appear 
to be associated with changes in the Household Dietary 
Diversity Score (HDDS). Furthermore, other mecha-
nisms may be in place that can influence food security, 
depending on the type of household considered, its inte-
gration into the food value chain, and the participa-
tion in safety net programs29. Additional analysis of the 
mechanisms and close monitoring of the effects of the 
crisis are then required to respond with appropriate pol-
icies as other crises arise. 

The main limitations of this work are related to the 
type of data available, which reduces the internal and 
external validities of the findings (Abate et al., 2023). 
Indeed, the fact that data are collected through phone 
interviews limits the representativeness of the sample, 
especially considering the low phone penetration in the 
rural areas of the country. The COVID-19 cases variable 
is not fully able to capture the infection rate and the eco-
nomic downturn caused by the policy interventions in the 
country. Additionally, measurement error could be wide-
spread in self-reported data. This is particularly relevant 

29 Abay et al. (2023), in the same study cited above, showed that par-
ticipation in the Productive Safety Net Program (SNP) offsets virtually 
almost all of the COVID-19 induced food insecurity increase (11.7 per-
centage points): the likelihood of becoming food insecure increased by 
only 2.4 percentage points for PSNP households. Qualitatively similar 
results are reported by Maffioli et al (2023) for Myanmar.

Table 8. Placebo test on ESS 2012/2014 and ESS 2015/2016.

Variables Total income 
change

Total employment

(1) (2)

Time 0.0852 -0.294*** -0.363**
(0.154) (0.0850) (0.169)

Time*cases 0.0136 0.0258** 0.0419
(0.0204) (0.0113) (0.0365)

Time*days 0.00274 0.00153 0.00192
(0.00538) (0.00295) (0.00311)

Time*days*cases -0.000364 -0.000233 -0.000312
(0.000701) (0.000386) (0.000431)

Cases*regional income -4.31e-07
(9.34e-07)

Constant -0.00491 0.601*** 0.601***
(0.0109) (0.00583) (0.00584)

Controls Yes Yes Yes
FE Yes Yes Yes
Observations 9,760 21,289 21,289
Number of pid 4,887 11,368 11,368
R-squared 0.023 0.050 0.050

Note: Dependent variables: categorical variable of income change, 
ranging from -2 (total loss) to 1 (increase) (1st column), and dum-
my equal to 1 if the individual is employed (2nd column). Income 
change is computed by comparing the amount of household income 
earned in each round. Robust standard errors in parentheses. *** 
p<0.01, ** p<0.05, * p<0.1

https://doi.org/10.36253/bae-13404


352

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

for the variable of income change, which is highly sub-
jective to respondents’ perception. Income data collected 
through more reliable measures are then needed to avoid 
major measurement errors. Finally, data used in this study 
were not intended to specifically track AFVC partici-
pants. Household surveys based on random sampling of 
the whole economy are typically unable to capture a rep-
resentative picture of the actors across the different seg-
ments of the value chain30. Vice versa, information/data 
retrieved through survey based on representative samples 
of the main AFVCs31 coupled with cascading survey of 
the various AFVC segments would have been better suited 
to grasp a better understanding of the overall effect of the 
COVID-19 crisis on the Ethiopian food system.

30 For instance, less than 100 individuals employed in the downstream 
segment are surveyed in each post-COVID round.
31 Studies on specific value chains in Ethiopia have been conducted only 
for the dairy value chain (see Hirvonen et al., 2021c), and the vegetable 
value chain (Hirvonen et al., 2021d).

REFERENCES

Ababulgu, N., Abajobir, N., and Wana, H. (2022). The 
Embarking of COVID-19 and the Perishable Prod-
ucts’ Value Chain in Ethiopia. Journal of Innovation 
and Entrepreneurship, 11(34). https://doi.org/10.1186/
s13731-022-00224-5. 

Abate, G. T., de Brauw, A., and Hirvonen, K. (2020). 
Food and Nutrition Security in Addis Ababa, Ethio-
pia During COVID-19 Pandemic: June 2020 Report. 
ESSP Working Paper 145. Washington, DC: Interna-
tional Food Policy Research Institute (IFPRI). https://
doi.org/10.2499/p15738coll2.133766

Abate, G. T., de Brauw, A., Hirvonen, K., and Wolle, A. 
(2023). Measuring Consumption Over the Phone: 
Evidence from a Survey Experiment in Urban Ethi-
opia. Journal of Development Economics 161 (2023) 
103026. https://doi.org/10.1016/j.jdeveco.2022.103026 

Abay, K. A., Berhane, G., Hoddinott, J., and Tafere, K. 
(2023). COVID-19 and Food Security in Ethiopia: 

Figure 15. Comparison of weighting methods. Source: own elaboration from ESS 2018/2019 and HFPSH 2020.

https://doi.org/10.36253/bae-13404
https://doi.org/10.1186/s13731-022-00224-5
https://doi.org/10.1186/s13731-022-00224-5
https://doi.org/10.2499/p15738coll2.133766
https://doi.org/10.2499/p15738coll2.133766
https://doi.org/10.1016/j.jdeveco.2022.103026


353The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Do Social Protection Programs Protect? Economic 
Development and Cultural Change 71(2). https://doi.
org/10.1086/715831

Alene, K. A., Gelaw, Y. A., Fetene, D. M., Koye, D. N., 
Melaku, Y. A., Gesesew, H., Birhanu, M. M., Adane, 
A. A., Muluneh, M. D., Dachew, B. A., Abrha, S., 
Aregay, A., Ayele, A. A., Bezabhe, W. M., Gebre-
mariam, K. T., Gebremedhin, T., Gebremedhin, A. 
T., Gebremichael, L., Geleto, A. B., Kassahun, H. T., 
… Kinfu, Y. (2021). COVID-19 in Ethiopia: A Geo-
spatial Analysis of Vulnerability to Infection, Case 
Severity, and Death. BMJ open, 11(2), e044606. htt-
ps://doi.org/10.1136/bmjopen-2020-044606 

Amare, M., Abay, K. A., Tiberti, L., Chamberlin, J. (2021). 
Impacts of COVID-19 on Food Security: Panel Data 
Evidence From Nigeria. Food Policy Volume 101, 
May 2021, 102099. https://doi.org/10.1016/j.food-
pol.2021.102099 

Ambel A. A., Bundervoet T., Asmelash H., and Wieser C. 
(2020a). Monitoring COVID-19 Impacts on House-
holds in Ethiopia. Survey Methodology Document. 
World Bank group. Washington, DC: The World 
Bank. http://documents.worldbank.org/curated/
en/107141590729601148/Survey-Methodology-Docu-
ment 

Ambel, A. A., Cardona Sosa, L. M., Tsegay, A. H., and 
Wieser, C. (2020b). Monitoring COVID-19 Impacts 
on Households in Ethiopia, Report No. 7: Results from 
Six Rounds of High-Frequency Household Phone Sur-
veys. Washington, DC: The World Bank. https://open-
knowledge.worldbank.org/handle/10986/34967. 

Asefawu, G. S. (2022). Seasonal Migration and Household 
Food Security Status in The Drought-Prone Areas 
of Northeast Ethiopia. Environmental Challenges, 
8:100566. https://doi.org/10.1016/j.envc.2022.100566. 

Asfaw, S., Shiferaw, B., Simtowe, F., and Hagos, M. (2011). 
Agricultural Technology, Seed Access Constraints 
and Commercialization in Ethiopia. Journal of Devel-
opment and Agricultural Economics, 3(9): 436–447. 
http://www.academicjournals.org/JDAE 

Baye, K. (2020). COVID-19 Prevention Measures In 
Ethiopia: Current Realities and Prospects. ESSP 
Working Paper 141. Washington, DC; Addis Ababa, 
Ethiopia: International Food Policy Research Insti-
tute (IFPRI); Federal Democratic Republic of Ethio-
pia, Policy Studies Institute. https://doi.org/10.2499/
p15738coll2.133729 

Bryan, E., Deressa, T. T., Gbetibouo, G. A., and Ringler, 
C. (2009). Adaptation to Climate Change In Ethiopia 
and South Africa: Options and Constraints. Environ-
mental Science and Policy, 12(4): 413–426. https://doi.
org/10.1016/j.envsci.2008.11.002 

Bundervoet, T. (2018). Internal Migration in Ethio-
pia: Evidence from a Quantitative and Qualita-
tive Research Study. Washington, DC: The World 
Bank. https://documents1.worldbank.org/curated/
en/428111562239161418/pdf/Internal-Migration-in-
Ethiopia-Evidence-from-a-Quantitative-and-Qualita-
tive-Research-Study.pdf 

Bundervoet, T., and Finn, A. (2020). “Ethiopia Poverty 
Assessment: What Can It Tell Us About Likely Effects 
of The Coronavirus?” Accessed 11 May 2020. Washing-
ton, DC: The World Bank. Retrieved from https://blogs.
worldbank.org/africacan/ethiopia-poverty-assessment-
what-can-it-tell-us-about-likely-effects-coronavirus 

Burbidge, J. B., L. Magee and A. L. Robb (1988). Alterna-
tive Transformations To Handle Extreme Values of The 
Dependent Variable. Journal of the American Statistical 
Association 83: 123–7. https://doi.org/10.2307/2288929

Cesaro, L., Giampaolo, A., Giarè, F., Sardone, R., Scard-
era, A., & Viganò, L. (2022). Italian farms during the 
COVID-19 pandemic: main problems and future 
perspectives. A direct analysis through the Ital-
ian FADN. Bio-Based and Applied Economics, 11(1), 
21–36. https://doi.org/10.36253/bae-9552 

Croppenstedt, A., Demeke, M., and Meschi, M. M. 
(2003). Technology Adoption in The Presence of 
Constraints: The Case of Fertilizer Demand in Ethio-
pia. Review of Development Economics, 7(1): 58–70. 
https://doi.org/10.1111/1467-9361.00175 

de Brauw, A., K. Hirvonen, and G. T. Abate. (2020). Food 
and nutrition security in Addis Ababa, Ethiopia dur-
ing COVID-19 Pandemic: July 2020 Report. IFPRI-
ESSP working paper 148. Washington D.C.: Interna-
tional Food Policy Research Institute (IFPRI). https://
doi.org/10.2499/p15738coll2.133851

De Weerdt J. (2008). Field Notes on Administering Shock 
Modules. Journal of International Development, 20(3), 
398-402. https://doi.org/10.1002/jid.1435 

Dercon, S., and Krishnan, P. (2000). Vulnerabil-
ity, Seasonality and Poverty in Ethiopia. Journal 
of Development Studies 36 (6): 25-53. https://doi.
org/10.1080/00220380008422653

Devereux, S., Bené, C., and Hoddinott, J. (2020). Con-
ceptualising Covid-19’s Impacts On Household Food 
Security. Food Security, 12(4):769–772. https://doi.
org/10.1007/s12571-020-01085-0

Fessha, Y.T., Dessalegn, B. (2020). “Internal Migration, 
Ethnic Federalism and Differentiated Citizenship 
in an African Federation: The Case of Ethiopia”. In: 
Gagnon, AG., Tremblay, A. (eds), Federalism and 
National Diversity in the 21st Century. Federalism 
and Internal Conflicts series. Cham: Palgrave Mac-
millan. https://doi.org/10.1007/978-3-030-38419-7_11 

https://doi.org/10.36253/bae-13404
https://doi.org/10.1086/715831
https://doi.org/10.1086/715831
https://doi.org/10.1136/bmjopen-2020-044606
https://doi.org/10.1136/bmjopen-2020-044606
https://doi.org/10.1016/j.foodpol.2021.102099
https://doi.org/10.1016/j.foodpol.2021.102099
http://documents.worldbank.org/curated/en/107141590729601148/Survey-Methodology-Document
http://documents.worldbank.org/curated/en/107141590729601148/Survey-Methodology-Document
http://documents.worldbank.org/curated/en/107141590729601148/Survey-Methodology-Document
https://openknowledge.worldbank.org/handle/10986/34967
https://openknowledge.worldbank.org/handle/10986/34967
https://doi.org/10.1016/j.envc.2022.100566
http://www.academicjournals.org/JDAE
https://doi.org/10.2499/p15738coll2.133729
https://doi.org/10.2499/p15738coll2.133729
https://doi.org/10.1016/j.envsci.2008.11.002
https://doi.org/10.1016/j.envsci.2008.11.002
https://documents1.worldbank.org/curated/en/428111562239161418/pdf/Internal-Migration-in-Ethiopia-Evidence-from-a-Quantitative-and-Qualitative-Research-Study.pdf
https://documents1.worldbank.org/curated/en/428111562239161418/pdf/Internal-Migration-in-Ethiopia-Evidence-from-a-Quantitative-and-Qualitative-Research-Study.pdf
https://documents1.worldbank.org/curated/en/428111562239161418/pdf/Internal-Migration-in-Ethiopia-Evidence-from-a-Quantitative-and-Qualitative-Research-Study.pdf
https://documents1.worldbank.org/curated/en/428111562239161418/pdf/Internal-Migration-in-Ethiopia-Evidence-from-a-Quantitative-and-Qualitative-Research-Study.pdf
https://blogs.worldbank.org/africacan/ethiopia-poverty-assessment-what-can-it-tell-us-about-likely-effects-coronavirus
https://blogs.worldbank.org/africacan/ethiopia-poverty-assessment-what-can-it-tell-us-about-likely-effects-coronavirus
https://blogs.worldbank.org/africacan/ethiopia-poverty-assessment-what-can-it-tell-us-about-likely-effects-coronavirus
https://doi.org/10.2307/2288929
https://doi.org/10.36253/bae-9552
https://doi.org/10.1111/1467-9361.00175
https://doi.org/10.2499/p15738coll2.133851
https://doi.org/10.2499/p15738coll2.133851
https://doi.org/10.1002/jid.1435
https://doi.org/10.1080/00220380008422653
https://doi.org/10.1080/00220380008422653
https://doi.org/10.1007/s12571-020-01085-0
https://doi.org/10.1007/s12571-020-01085-0
https://doi.org/10.1007/978-3-030-38419-7_11


354

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Gilbert, C.L., Christiaensen, L., and Kaminski, J. (2017). 
Food Price Seasonality in Africa: Measurement 
and Extent. Food Policy 67: 119-132. https://doi.
org/10.1016/j.foodpol.2016.09.016. 

Harvest SA (2012). Challenges and Constraints For 
Small-Scale Farmer. Agricultural Research Coun-
cil, Pretoria. https://www.arc.agric.za/arc-iscw/News 
Articles Library/Challenges and constraints for 
small-scale farmers.pdf. 

Himelein, K. (2014). Weight Calculations for Panel Sur-
veys with Subsampling and Split-off Tracking, Statis-
tics and Public Policy, 1(1): 40-45. http://dx.doi.org/1
0.1080/2330443X.2013.856170.

Hirvonen, K., Taffesse, A. S., and Worku. I. (2016). Sea-
sonality and Household Diets in Ethiopia. Public 
Health Nutrition 19(10): 1723-1730. DOI:  10.1017/
S1368980015003237

Hirvonen, K., Abate, G. T., and de Brauw, A. (2020). 
Food and Nutrition Security in Addis Ababa, Ethio-
pia during COVID-19 Pandemic: May 2020 Report. 
Working Paper 143. Washington, DC: International 
Food Policy Research Institute (IFPRI). https://doi.
org/10.2499/p15738coll2.133731 

Hirvonen, K., de Brauw, A., and Abate., G.T. (2021a). 
Food Consumption and Food Security During the 
COVID-19 Pandemic in Addis Ababa. American 
Journal of Agricultural Economics 103 (3): 772-789. 
https://doi.org/10.1111/ajae.12206

Hirvonen, K., Mohammed, B., Minten, B., and Tamru, 
S. (2021b). Food Prices, Marketing Margins, and 
Shocks: Evidence from Vegetables and the COV-
ID-19 Pandemic in Ethiopia. Agricultural Economics 
52(3): 407-421. https://doi.org/10.1111/agec.12626 

Hirvonen, K., Habte, Y., Mohammed, B., Tamru, S., 
Abate, G. T., and Minten, B. (2021c). Dairy Value 
Chains during the COVID-19 Pandemic in Ethio-
pia: Evidence from Cascading Value Chain Sur-
veys Before and during the Pandemic. IFPRI-ESSP 
Working Paper 160. Washington, DC: International 
Food Policy Research Institute (IFPRI). https://doi.
org/10.2499/p15738coll2.134764 

Hirvonen, K., Mohammed, B., Tamru, S., Abate, G. T., 
and Minten. B. (2021d). Vegetable Value Chains Dur-
ing the COVID-19 Pandemic in Ethiopia: Evidence 
from Cascading Value Chain Surveys Before and 
During the Pandemic. IFPRI-ESSP Working Paper 
159. Washington, DC: International Food Policy 
Research Institute (IFPRI). https://doi.org/10.2499/
p15738coll2.134768 

Johnson, N. (1949). Systems of Frequency Curves Gener-
ated by Methods of Translation. Biometrika: 36(1/2): 
149-176. https://doi.org/10.2307/2332539 

Josephson, A., Kilic, T.,and Michler, J. D. (2021). Socio-
economic Impacts of COVID-19 in Low-Income 
Countries. Nature Human Behaviour: 5 (5):557-565. 
https://doi.org/10.1038/s41562-021-01096-7 

Khamis, M., Prinz, D., Newhouse, D., Palacios-Lopez, 
A., Pape, U., and Weber, M. (2021). The Early Labor 
Market Impacts of COVID-19 in Developing Coun-
tries: Evidence from High-Frequency Phone Surveys. 
Policy Research Working Paper; No. 9510. Washing-
ton, DC: The World Bank. https://openknowledge.
worldbank.org/handle/10986/35025.

Laborde Debucquet, D., Martin, W., and Vos, R. 
(2020). Impacts of COVID-19 on Global Poverty, 
Food Security, and Diets. IFPRI Discussion Paper 
1993. Washington, DC: International Food Policy 
Research Institute (IFPRI). https://doi.org/10.2499/
p15738coll2.134229 

Lasarte-López, J., Grassano, N., M’barek, R., and Ronzon, 
T. (2023). Bioeconomy and resilience to economic 
shocks: insights from the COVID-19 pandemic in 
2020. Bio-Based and Applied Economics. Retrieved 
from https://oaj.fupress.net/index.php/bae/article/
view/14827 

Lazar, M., Piou, C., Doumandji‐Mitiche, B., and Lecoq, 
M. (2016). Importance of Solitarious Desert Locust 
Population Dynamics: Lessons from Historical Sur-
vey Data in Algeria. Entomologia Experimentalis et 
Applicata 161(3): 168-180. https://doi.org/10.1111/
eea.12505

Maffioli, E.M., Headey, D., Lambrecht, I., Zaw Oo, T., 
Tint Zaw, N. (2023). A Prepandemic Nutrition-Sen-
sitive Social Protection Program Has Sustained Ben-
efits for Food Security and Diet Diversity in Myan-
mar During a Severe Economic Crisis. The Journal of 
Nutrition 369 (Feb 2023). https://doi.org/10.1016/j.
tjnut.2023.02.009

Minten, B., Mohammed, B., and Tamru, S. (2020). 
Emerging Medium-Scale Tenant Farming, Gig Econ-
omies, and the COVID-19 Disruption. ESSP Work-
ing Paper 149, August 2020. Washington, DC: Inter-
national Food Policy Research Institute (IFPRI). htt-
ps://doi.org/10.2499/p15738coll2.133909 

Molla, A. (2020). How Can Farmers in Ethiopia Work Safe 
and Smart During COVID-19 Lockdown? Internation-
al Institute for Environment and Development. Blog 
retrieved at https://www.iied.org/how-can-farmers-
ethiopia-work-safe-smart-during-covid-19-lockdown 

Moosavi, J., Fathollahi-Fard, A. M., and Dulebenets, 
M. A. (2022). Supply Chain Disruption During The 
Covid-19 Pandemic: Recognizing Potential Disrup-
tion Management Strategies. International Journal 
of Disaster Risk Reduction, 75:102983. https://doi.

https://doi.org/10.36253/bae-13404
https://doi.org/10.1016/j.foodpol.2016.09.016
https://doi.org/10.1016/j.foodpol.2016.09.016
https://www.arc.agric.za/arc-iscw/News
http://dx.doi.org/10.1080/2330443X.2013.856170
http://dx.doi.org/10.1080/2330443X.2013.856170
https://doi.org/10.2499/p15738coll2.133731
https://doi.org/10.2499/p15738coll2.133731
https://doi.org/10.1111/ajae.12206
https://doi.org/10.1111/agec.12626
https://doi.org/10.2499/p15738coll2.134764
https://doi.org/10.2499/p15738coll2.134764
https://doi.org/10.2499/p15738coll2.134768
https://doi.org/10.2499/p15738coll2.134768
https://doi.org/10.2307/2332539
https://doi.org/10.1038/s41562-021-01096-7
https://openknowledge.worldbank.org/handle/10986/35025
https://openknowledge.worldbank.org/handle/10986/35025
https://doi.org/10.2499/p15738coll2.134229
https://doi.org/10.2499/p15738coll2.134229
https://oaj.fupress.net/index.php/bae/article/view/14827
https://oaj.fupress.net/index.php/bae/article/view/14827
https://doi.org/10.1111/eea.12505
https://doi.org/10.1111/eea.12505
https://doi.org/10.1016/j.tjnut.2023.02.009
https://doi.org/10.1016/j.tjnut.2023.02.009
https://doi.org/10.2499/p15738coll2.133909
https://doi.org/10.2499/p15738coll2.133909
https://www.iied.org/how-can-farmers-ethiopia-work-safe-smart-during-covid-19-lockdown
https://www.iied.org/how-can-farmers-ethiopia-work-safe-smart-during-covid-19-lockdown
https://doi.org/10.1016/j.ijdrr.2022.102983


355The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

org/10.1016/j.ijdrr.2022.102983. 
Njeru, E., Grey, S. and Kilawe, E. (2016). Eastern Africa 

Climate-Smart Agriculture Scoping Study: Ethiopia, 
Kenya, and Uganda. Addis Ababa, Ethiopia: FAO. 
https://www.fao.org/3/i5485e/i5485e.pdf 

Reardon, T., Bellemare, M. F., and Zilberman, D. (2020a). 
“How COVID-19 May Disrupt Food Supply Chains 
in Developing Countries”. In Swinnen, J.M.A, and 
McDermott, J., (eds.). COVID-19 and Global Food 
Security, Pp. 78-80. Washington, DC: International 
Food Policy Research Institute (IFPRI). https://doi.
org/10.2499/p15738coll2.133762_17 

Reardon, T., Mishra, A., Nuthalapati, C. S., Bellemare, 
M. F., and Zilberman, D. (2020b). COVID-19’s Dis-
ruption of India’s Transformed Food Supply Chains. 
Economic and Political Weekly, 55(18), 18-22. https://
www.epw.in/journal/2020/18/commentary/covid-
19s-disruption-indias-transformed-food.html 

Roba, K. T., O’Connor, T. P., O’Brien, N. M., Aweke, C. 
S., Kahsay, Z. A., Chisholm, N., and Lahiff, E. (2019). 
Seasonal Variations in Household Food Insecu-
rity and Dietary Diversity and Their Association 
with Maternal and Child Nutritional Status in Rural 
Ethiopia. Food Security 11(3): 651-664. https://doi.
org/10.1007/s12571-019-00920-3 

Rudin-Rush, L., Michler, J. D., Josephson, A., and Bloem, 
J. R. (2022). Food Insecurity during the First Year of 
the COVID-19 Pandemic in Four African Countries. 
Food Policy 111: 102306. https://doi.org/10.1016/j.
foodpol.2022.102306. 

Schicker, R. S., Hiruy, N., Melak, B., Gelaye, W., Beza-
bih, B., Stephenson, R., Patterson, A. E., Tadesse, Z., 
Emerson, P. M., Richards, Jr., F. O., and Noland, G. S. 
(2015). A Venue-Based Survey of Malaria, Anemia, 
And Mobility Patterns Among Migrant Farm Workers 
in Amhara Region, Ethiopia. PLOS ONE, 10(11):1–22. 
https://doi.org/10.1371/journal.pone.0143829 

Schimdhuber, J., and Qiao, B. 2020. Comparing Crises: 
Great Lockdown versus Great Recession. Rome: FAO. 
https://doi.org/10.4060/ca8833en 

Sibhatu, K. T., and Qaim. M. (2017). Rural Food Security, 
Subsistence Agriculture, and Seasonality. PloS one 
12 (10): e0186406. https://doi.org/10.1371/journal.
pone.0186406 

Singh, K.M., Ahmad, N., Sinha, D.K., Singh, R.K.P., and 
Mishra, R.R. (2018). Diversification and its Determi-
nants: A Search for an Alternative Income and Agricul-
tural Development in Eastern India. International Jour-
nal of Current Microbiology and Applied Sciences. 7(2): 
695-702. https://doi.org/10.20546/ijcmas.2018.702.087

Swinnen, J.M.A. (2020). Will COVID-19 Cause Another 
Food Crisis? An Early Review. IFPRI Blog. Washing-

ton, DC: International Food Policy Research Institute. 
https://www.ifpri.org/blog/will-covid-19-cause-anoth-
er-food-crisis-early-review (accessed 12 July 2020). 

Swinnen, J. and McDermott, J. (2020), Covid-19 and 
Global Food Security. EuroChoices, 19: 26-33. https://
doi.org/10.1111/1746-692X.12288

Tamru, S., Hirvonen, K., and Minten, B. (2020). “Impacts 
of the COVID-19 Crisis on Vegetable Value Chains 
in Ethiopia”. In Swinnen, J.M.A, and McDermott, 
J., (eds.). COVID-19 and Global Food Security, Pp. 
81-83. Washington, DC: International Food Policy 
Research Institute (IFPRI). https://doi.org/10.2499/
p15738coll2.133762_18

Tesfaye, A., Habte, Y., and Minten, B. (2020). COVID-19 
Is Shifting Consumption and Disrupting Dairy Value 
Chains in Ethiopia. IFPRI Blog: Research Post. June 
1, 2020. https://www.ifpri.org/blog/covid-19-shifting-
consumption-and-disrupting-dairy-value-chains-
ethiopia 

Tigre, G. and Heshmati, A. (2022). Smallholder Farm-
ers’ Crop Production and Input Risk Analysis 
In Rural Ethiopia. Applied Economics, 0(0):1–19. 
10.1080/00036846.2022.2094327 

USDA 2008. Ethiopia 2008 Crop Assessment Travel 
Report. Available at https://ipad.fas.usda.gov/high-
lights/2008/11/eth_25nov2008/#:~:text=To%20
avoid%20confusion%20between%20these,harvested%20
between%20September%20and%20February.

Wieser, C., Ambel, A. A., Bundervoet, T., and Tse-
gay, A. H. (2020). Monitoring COVID-19 
Impacts on Households in Ethiopia: Results 
from a High-Frequency Phone Survey of House-
holds. Report #1. Washington D.C.: The World 
Bank. http://documents.worldbank.org/curated/
en/428541643263668314/Monitoring-COVID-19-Im-
pacts-on-Households-in-Ethiopia-Results-from-
11-Rounds-of-High-Frequency-Phone-Surveys-of-
Households-from-April-2020-through-May-2021 

Woldemichael, A., Takian, A., Akbari Sari, A., and Oly-
aeemanesh, A. (2019). Availability and Inequality in 
Accessibility of Health Centre-Based Primary Health-
care in Ethiopia. PloS one, 14 (3): e0213896. https://
doi.org/10.1371/journal.pone.0213896 

World Bank (2021). World Development Indicators. 
Washington D.C.: The World Bank. Available at htt-
ps://data.worldbank.org/indicator/NV.AGR.TOTL.
ZS?locations=ET (accessed on 29/08/2021).

Zhang, W., Persoz, L., Hakiza, S., Biru, L., and Girmat-
sion, L. (2022). Impact of COVID-19 on Food Secu-
rity in Ethiopia. Epidemiologia, 3(2): 161–178. https://
doi.org/10.3390/epidemiologia3020013 

https://doi.org/10.36253/bae-13404
https://doi.org/10.1016/j.ijdrr.2022.102983
https://www.fao.org/3/i5485e/i5485e.pdf
https://doi.org/10.2499/p15738coll2.133762_17
https://doi.org/10.2499/p15738coll2.133762_17
https://www.epw.in/journal/2020/18/commentary/covid-19s-disruption-indias-transformed-food.html
https://www.epw.in/journal/2020/18/commentary/covid-19s-disruption-indias-transformed-food.html
https://www.epw.in/journal/2020/18/commentary/covid-19s-disruption-indias-transformed-food.html
https://doi.org/10.1007/s12571-019-00920-3
https://doi.org/10.1007/s12571-019-00920-3
https://doi.org/10.1016/j.foodpol.2022.102306
https://doi.org/10.1016/j.foodpol.2022.102306
https://doi.org/10.1371/journal.pone.0143829
https://doi.org/10.4060/ca8833en
https://doi.org/10.1371/journal.pone.0186406
https://doi.org/10.1371/journal.pone.0186406
https://doi.org/10.20546/ijcmas.2018.702.087
https://www.ifpri.org/blog/will-covid-19-cause-another-food-crisis-early-review
https://www.ifpri.org/blog/will-covid-19-cause-another-food-crisis-early-review
https://doi.org/10.1111/1746-692X.12288
https://doi.org/10.1111/1746-692X.12288
https://doi.org/10.2499/p15738coll2.133762_18
https://doi.org/10.2499/p15738coll2.133762_18
https://www.ifpri.org/blog/covid-19-shifting-consumption-and-disrupting-dairy-value-chains-ethiopia
https://www.ifpri.org/blog/covid-19-shifting-consumption-and-disrupting-dairy-value-chains-ethiopia
https://www.ifpri.org/blog/covid-19-shifting-consumption-and-disrupting-dairy-value-chains-ethiopia
https://ipad.fas.usda.gov/highlights/2008/11/eth_25nov2008/#
https://ipad.fas.usda.gov/highlights/2008/11/eth_25nov2008/#
http://documents.worldbank.org/curated/en/428541643263668314/Monitoring-COVID-19-Impacts-on-Households-in-Ethiopia-Results-from-11-Rounds-of-High-Frequency-Phone-Surveys-of-Households-from-April-2020-through-May-2021
http://documents.worldbank.org/curated/en/428541643263668314/Monitoring-COVID-19-Impacts-on-Households-in-Ethiopia-Results-from-11-Rounds-of-High-Frequency-Phone-Surveys-of-Households-from-April-2020-through-May-2021
http://documents.worldbank.org/curated/en/428541643263668314/Monitoring-COVID-19-Impacts-on-Households-in-Ethiopia-Results-from-11-Rounds-of-High-Frequency-Phone-Surveys-of-Households-from-April-2020-through-May-2021
http://documents.worldbank.org/curated/en/428541643263668314/Monitoring-COVID-19-Impacts-on-Households-in-Ethiopia-Results-from-11-Rounds-of-High-Frequency-Phone-Surveys-of-Households-from-April-2020-through-May-2021
http://documents.worldbank.org/curated/en/428541643263668314/Monitoring-COVID-19-Impacts-on-Households-in-Ethiopia-Results-from-11-Rounds-of-High-Frequency-Phone-Surveys-of-Households-from-April-2020-through-May-2021
https://doi.org/10.1371/journal.pone.0213896
https://doi.org/10.1371/journal.pone.0213896
https://data.worldbank.org/indicator/NV.AGR.TOTL.ZS?locations=ET
https://data.worldbank.org/indicator/NV.AGR.TOTL.ZS?locations=ET
https://data.worldbank.org/indicator/NV.AGR.TOTL.ZS?locations=ET
https://doi.org/10.3390/epidemiologia3020013
https://doi.org/10.3390/epidemiologia3020013


356

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

APPENDIX 

Figure A.1. Impact of COVID-19 cases on income change, wave 
by wave. Source: Own calculation from ESS 2018/2019 and HFPSH 
2020. Note: Dependent variable = categorical variable of income 
change, ranging from -2 (total loss) to 1 (increase). Dots are coef-
ficients estimated from a linear probability model with household 
fixed effects. Each post-COVID round is compared with the base-
line. Bars are 95% confidence intervals. Sampling weights applied. 
Previous call is considered the baseline. Standard errors clustered at 
the household level. *** p<0.01, ** p<0.05, * p<0.1. 

Figure A.2. Impact of COVID-19 cases on employment over time, 
unbalanced sample. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = dummy equal to 1 
if the individual is employed. Dots are coefficients estimated from 
a linear probability model with household fixed effects. Each post-
COVID round is compared with the baseline. Bars are 95% confi-
dence intervals. Sampling weights applied. Standard errors clustered 
at the household level. *** p<0.01, ** p<0.05, * p<0.1.

Figure A.3. Impact of COVID-19 cases on total income over time, 
unbalanced sample. Source: Own calculation from ESS 2018/2019 
and HFPSH 2020. Note: Dependent variable = categorical vari-
able of income change, ranging from -2 (total loss) to 1 (increase). 
Dots are coefficients estimated from a linear probability model with 
household fixed effects. Each post-COVID round is compared with 
the baseline. Bars are 95% confidence intervals. Sampling weights 
applied. Standard errors clustered at the household level. *** 
p<0.01, ** p<0.05, * p<0.1.

Figure A.4. Marginal effects of land size on the probability that 
farm income change has not decreased. Source: Own calculation 
from ESS 2018/2019 and HFPSH 2020.

https://doi.org/10.36253/bae-13404


357The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Table A.1. Number of individuals that started to work again in each round, by reason for stop working in the previous round.

Reason for stop working
N. of individuals that started working again

Round 2 Round 3 Round 4 Round 5 Round 6

Seasonal/Casual worker 27 8 8 7 6
Contract ended 3 0 3 2 1
Covid-19 83 22 22 5 6
Temporarily absent 25 8 6 5 9
Retired 0 0 0 1 0
Being ill 2 8 2 1 5
Need to care for ill 1 1 1 0 0
Other 1 0 1 1 0
N/A 329 94 71 54 30
Total 471 141 114 76 57

Table A.2. Full regression estimates, total employment.

Variables Round 1 Round 2 Round 3 Round 4 Round 5 Round 6

Time -0.0709*** 0.313*** 0.368*** 0.643*** 0.886*** 0.961***
  (0.0196) (0.0463) (0.0539) (0.0807) (0.0975) (0.107)
Time*Cases -0.0360*** -0.0717*** -0.0673*** -0.0813*** -0.0987*** -0.107***
  (0.00654) (0.00837) (0.00879) (0.00980) (0.0110) (0.0118)
Time*Days -0.000364 -0.00707*** -0.00792*** -0.0178*** -0.0260*** -0.0296***
  (0.000640) (0.00134) (0.00142) (0.00281) (0.00377) (0.00391)
Time*Cases*Days -1.53e-06 0.00251*** 0.00197*** 0.00266*** 0.00338*** 0.00377***
  (0.000383) (0.000367) (0.000345) (0.000434) (0.000513) (0.000514)
Other HH members lost job 0.0119 -0.166*** -0.0858 -0.1000 -0.233*** -0.225***
  (0.0412) (0.0629) (0.0663) (0.0647) (0.0822) (0.0742)
HH received assistance 0.0449 0.0777 0.0571 0.0223 -0.0219 -0.000938
  (0.0358) (0.0634) (0.0537) (0.0495) (0.0480) (0.0459)
Constant 0.746*** 0.747*** 0.748*** 0.749*** 0.752*** 0.752***
  (0.00507) (0.00840) (0.00833) (0.00853) (0.00842) (0.00863)

Observations 4,693 4,694 4,694 4,693 4,694 4,694
R-squared 0.122 0.086 0.079 0.098 0.124 0.116
Number of pid1 2,347 2,347 2,347 2,347 2,347 2,347

Source: Own calculation from ESS 2018/2019 and HFPSH 2020. 
Note: Dependent variable = dummy equal to 1 if individual is employed. Coefficients estimated using a linear probability model with 
household fixed effects. Each post-COVID round is compared with the baseline. 95% confidence intervals in parenthesis. Sampling weights 
applied. Standard errors clustered at the household level. *** p<0.01, ** p<0.05, * p<0.

https://doi.org/10.36253/bae-13404


358

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Table A.3. Full regression estimates, total income.

Variables Round 1 Round 2 Round 3 Round 4 Round 5 Round 6

Time -0.549*** -0.599*** -0.502*** -0.237* -0.0341 0.203
  (0.0412) (0.0522) (0.0685) (0.124) (0.160) (0.186)
Time*Cases -0.0148 -0.00295 -0.0216** -0.0533*** -0.0677*** -0.0937***
  (0.0119) (0.0102) (0.0110) (0.0150) (0.0179) (0.0206)
Time*Days 1.58e-05 -0.00265 -0.00653** -0.0154** -0.0162* -0.0172*
  (0.00161) (0.00198) (0.00311) (0.00670) (0.00945) (0.0102)
Time*Cases*Days -0.000970 0.000198 0.000921 0.00199** 0.00188 0.00192
  (0.000864) (0.000564) (0.000685) (0.000947) (0.00119) (0.00124)
HH received assistance -0.0774 -0.141** -0.201*** -0.117 -0.163** -0.165**
  (0.0808) (0.0549) (0.0760) (0.0768) (0.0779) (0.0787)
Other HH members lost job -0.0758 -0.119 -0.123 -0.0720 -0.128 -0.108
  (0.0871) (0.0839) (0.0759) (0.0823) (0.0809) (0.0802)
Constant -0 -0 0 0 -0 0
  (0.0105) (0.0107) (0.0122) (0.0137) (0.0151) (0.0156)

Observations 4,691 4,693 4,694 4,691 4,693 4,685
R-squared 0.505 0.568 0.540 0.472 0.408 0.363
Number of pid1 2,347 2,347 2,347 2,347 2,347 2,347

Source: Own calculation from ESS 2018/2019 and HFPSH 2020. 
Note: Dependent variable: categorical variable of income change, ranging from -2 (total loss) to 1 (increase). Estimates are computed using 
a linear probability model with household fixed effects. Sampling weights applied. Standard errors clustered at the household level. *** 
p<0.01, ** p<0.05, * p<0.1.

https://doi.org/10.36253/bae-13404


359The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Ta
bl

e 
A

.4
. O

rd
er

ed
 p

ro
bi

t, 
to

ta
l i

nc
om

e 
ch

an
ge

.

Va
ria

bl
es

Ro
un

d 
1

Ro
un

d 
2

Ro
un

d 
3

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

Ye
ar

s o
f e

du
ca

tio
n

-0
.0

01
71

**
-0

.0
06

20
**

0.
00

73
0*

*
0.

00
06

11
*

-0
.0

02
11

**
-0

.0
05

95
**

0.
00

67
8*

*
0.

00
12

9*
*

-0
.0

01
60

-0
.0

02
91

0.
00

33
1

0.
00

11
9

(0
.0

00
86

9)
(0

.0
02

96
)

(0
.0

03
48

)
(0

.0
00

34
8)

(0
.0

00
98

3)
(0

.0
02

83
)

(0
.0

03
19

)
(0

.0
00

63
5)

(0
.0

01
30

)
(0

.0
02

41
)

(0
.0

02
73

)
(0

.0
00

96
8)

H
H

 h
ea

d 
is 

fe
m

al
e

-0
.0

01
76

-0
.0

06
47

0.
00

75
9

0.
00

06
43

0.
01

17
0.

03
04

-0
.0

35
8

-0
.0

06
40

0.
02

88
0.

04
46

-0
.0

55
4

-0
.0

18
1

(0
.0

10
2)

(0
.0

38
1)

(0
.0

44
5)

(0
.0

03
84

)
(0

.0
11

8)
(0

.0
28

4)
(0

.0
34

4)
(0

.0
05

87
)

(0
.0

24
7)

(0
.0

29
2)

(0
.0

41
6)

(0
.0

12
3)

A
ge

 o
f H

H
 h

ea
d

-0
.0

00
46

1*
-0

.0
01

67
*

0.
00

19
7*

0.
00

01
65

-0
.0

00
20

5
-0

.0
00

57
7

0.
00

06
57

0.
00

01
25

3.
79

e-
05

6.
89

e-
05

-7
.8

5e
-0

5
-2

.8
3e

-0
5

(0
.0

00
27

6)
(0

.0
00

97
5)

(0
.0

01
15

)
(0

.0
00

10
3)

(0
.0

00
32

3)
(0

.0
00

92
3)

(0
.0

01
05

)
(0

.0
00

19
8)

(0
.0

00
42

9)
(0

.0
00

77
9)

(0
.0

00
88

7)
(0

.0
00

32
1)

H
H

 h
ea

d 
is 

m
ar

rie
d

0.
00

80
2

0.
03

06
-0

.0
35

5
-0

.0
03

14
0.

00
92

5
0.

02
75

-0
.0

30
6

-0
.0

06
15

0.
01

62
0.

03
19

-0
.0

34
7

-0
.0

13
4

(0
.0

09
75

)
(0

.0
39

3)
(0

.0
44

9)
(0

.0
04

16
)

(0
.0

10
7)

(0
.0

33
7)

(0
.0

36
6)

(0
.0

07
82

)
(0

.0
18

4)
(0

.0
37

0)
(0

.0
39

2)
(0

.0
16

3)
H

H
 si

ze
-0

.0
00

39
7

-0
.0

01
44

0.
00

16
9

0.
00

01
41

0.
00

02
08

0.
00

05
86

-0
.0

00
66

7
-0

.0
00

12
7

0.
00

03
37

0.
00

06
13

-0
.0

00
69

8
-0

.0
00

25
2

(0
.0

02
06

)
(0

.0
07

44
)

(0
.0

08
77

)
(0

.0
00

73
3)

(0
.0

02
42

)
(0

.0
06

83
)

(0
.0

07
77

)
(0

.0
01

48
)

(0
.0

03
44

)
(0

.0
06

26
)

(0
.0

07
13

)
(0

.0
02

57
)

A
ss

et
 in

de
x

0.
00

19
3

0.
00

69
9

-0
.0

08
23

-0
.0

00
68

9
-0

.0
00

41
2

-0
.0

01
16

0.
00

13
2

0.
00

02
51

-0
.0

01
47

-0
.0

02
68

0.
00

30
5

0.
00

11
0

(0
.0

01
64

)
(0

.0
05

93
)

(0
.0

06
96

)
(0

.0
00

62
0)

(0
.0

01
80

)
(0

.0
05

03
)

(0
.0

05
74

)
(0

.0
01

09
)

(0
.0

02
67

)
(0

.0
04

69
)

(0
.0

05
39

)
(0

.0
01

96
)

H
H

 c
on

su
m

pt
io

n 
(lo

g)
-0

.0
09

01
-0

.0
32

6
0.

03
84

0.
00

32
1

0.
00

21
5

0.
00

60
5

-0
.0

06
89

-0
.0

01
31

0.
00

41
9

0.
00

76
1

-0
.0

08
67

-0
.0

03
13

(0
.0

06
62

)
(0

.0
23

3)
(0

.0
27

6)
(0

.0
02

41
)

(0
.0

07
56

)
(0

.0
21

2)
(0

.0
24

2)
(0

.0
04

62
)

(0
.0

09
75

)
(0

.0
17

8)
(0

.0
20

2)
(0

.0
07

31
)

%
 o

f m
em

be
rs

 w
ith

 jo
b 

co
nt

ra
ct

-0
.0

63
5*

**
-0

.2
30

**
*

0.
27

1*
**

0.
02

27
**

*
-0

.0
63

4*
**

-0
.1

78
**

*
0.

20
3*

**
0.

03
86

**
*

-0
.0

68
6*

**
-0

.1
25

**
0.

14
2*

*
0.

05
13

**

(0
.0

14
8)

(0
.0

49
4)

(0
.0

58
0)

(0
.0

06
29

)
(0

.0
14

3)
(0

.0
36

7)
(0

.0
42

3)
(0

.0
09

33
)

(0
.0

24
8)

(0
.0

52
2)

(0
.0

56
4)

(0
.0

20
3)

H
H

 re
ce

iv
es

 so
ci

al
 

as
sis

ta
nc

e
-0

.0
11

1
-0

.0
47

0
0.

05
30

0.
00

51
4

0.
00

67
4

0.
01

74
-0

.0
20

5
-0

.0
03

70
-0

.0
03

51
-0

.0
06

62
0.

00
74

0
0.

00
27

3

(0
.0

11
3)

(0
.0

54
9)

(0
.0

59
5)

(0
.0

06
74

)
(0

.0
18

8)
(0

.0
44

3)
(0

.0
53

8)
(0

.0
09

27
)

(0
.0

21
7)

(0
.0

42
3)

(0
.0

46
5)

(0
.0

17
5)

H
H

 h
as

 fo
rm

al
 c

re
di

t
-0

.0
01

74
-0

.0
06

41
0.

00
75

1
0.

00
06

39
0.

03
25

*
0.

06
62

**
-0

.0
85

1*
*

-0
.0

13
6*

*
0.

04
14

0.
05

31
**

-0
.0

72
3*

-0
.0

22
3*

*
(0

.0
11

3)
(0

.0
42

4)
(0

.0
49

4)
(0

.0
04

28
)

(0
.0

19
3)

(0
.0

26
8)

(0
.0

40
1)

(0
.0

06
10

)
(0

.0
27

5)
(0

.0
22

1)
(0

.0
38

2)
(0

.0
11

1)
H

H
 h

as
 sa

vi
ng

s
0.

02
89

**
*

0.
10

7*
**

-0
.1

26
**

*
-0

.0
10

2*
**

0.
01

63
*

0.
04

56
*

-0
.0

52
0*

-0
.0

09
89

*
0.

00
58

6
0.

01
06

-0
.0

12
1

-0
.0

04
37

(0
.0

08
87

)
(0

.0
29

8)
(0

.0
35

4)
(0

.0
03

29
)

(0
.0

09
57

)
(0

.0
25

8)
(0

.0
29

7)
(0

.0
05

68
)

(0
.0

12
5)

(0
.0

23
0)

(0
.0

26
2)

(0
.0

09
40

)
H

H
 h

as
 b

an
k 

ac
co

un
t

0.
00

01
99

0.
00

07
22

-0
.0

00
85

0
-7

.1
0e

-0
5

-0
.0

00
56

3
-0

.0
01

58
0.

00
18

0
0.

00
03

43
-0

.0
01

25
-0

.0
02

27
0.

00
25

9
0.

00
09

33
(0

.0
09

18
)

(0
.0

33
3)

(0
.0

39
2)

(0
.0

03
28

)
(0

.0
10

6)
(0

.0
29

7)
(0

.0
33

8)
(0

.0
06

42
)

(0
.0

14
2)

(0
.0

25
8)

(0
.0

29
4)

(0
.0

10
6)

H
H

 h
as

 fo
rm

al
 

in
su

ra
nc

e
-0

.0
08

23
-0

.0
32

9
0.

03
77

0.
00

34
3

-0
.0

19
9*

*
-0

.0
72

6*
0.

07
56

**
0.

01
69

*
-0

.0
32

1*
**

-0
.0

81
7*

*
0.

07
93

**
0.

03
46

**

(0
.0

08
32

)
(0

.0
36

0)
(0

.0
40

3)
(0

.0
04

00
)

(0
.0

08
35

)
(0

.0
37

7)
(0

.0
36

4)
(0

.0
09

78
)

(0
.0

11
7)

(0
.0

38
7)

(0
.0

32
8)

(0
.0

17
4)

D
ist

an
ce

 to
 m

ai
n 

ro
ad

-0
.0

00
18

1
-0

.0
00

65
7

0.
00

07
74

6.
47

e-
05

-0
.0

00
24

6
-0

.0
00

69
3

0.
00

07
90

0.
00

01
50

-0
.0

00
41

8*
-0

.0
00

76
1*

0.
00

08
67

*
0.

00
03

13
*

(0
.0

00
15

6)
(0

.0
00

56
1)

(0
.0

00
66

0)
(5

.6
7e

-0
5)

(0
.0

00
17

6)
(0

.0
00

49
5)

(0
.0

00
56

6)
(0

.0
00

10
6)

(0
.0

00
22

7)
(0

.0
00

40
5)

(0
.0

00
46

6)
(0

.0
00

16
4)

D
ist

an
ce

 to
 u

rb
an

 c
en

te
r

-4
.5

7e
-0

5
-0

.0
00

16
6

0.
00

01
95

1.
63

e-
05

-0
.0

00
13

2*
-0

.0
00

37
3*

*
0.

00
04

24
**

8.
06

e-
05

*
-0

.0
00

13
8

-0
.0

00
25

0
0.

00
02

85
0.

00
01

03
(5

.1
5e

-0
5)

(0
.0

00
18

8)
(0

.0
00

22
0)

(1
.8

9e
-0

5)
(6

.8
5e

-0
5)

(0
.0

00
18

7)
(0

.0
00

21
3)

(4
.3

3e
-0

5)
(9

.0
6e

-0
5)

(0
.0

00
16

1)
(0

.0
00

18
3)

(6
.7

7e
-0

5)
D

ist
an

ce
 to

 m
ar

ke
t

-0
.0

00
47

6
-0

.0
01

73
0.

00
20

3
0.

00
01

70
-0

.0
00

35
6

-0
.0

01
00

0.
00

11
4

0.
00

02
17

-0
.0

00
70

5
-0

.0
01

28
0.

00
14

6
0.

00
05

27
(0

.0
00

30
0)

(0
.0

01
08

)
(0

.0
01

27
)

(0
.0

00
10

8)
(0

.0
00

34
5)

(0
.0

00
96

7)
(0

.0
01

11
)

(0
.0

00
20

7)
(0

.0
00

46
6)

(0
.0

00
85

4)
(0

.0
00

97
2)

(0
.0

00
34

5)

(C
on

tin
ue

d)

https://doi.org/10.36253/bae-13404


360

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Va
ria

bl
es

Ro
un

d 
1

Ro
un

d 
2

Ro
un

d 
3

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

M
ic

ro
fin

an
ce

 in
st

itu
tio

n
-0

.0
03

41
-0

.0
12

4
0.

01
45

0.
00

12
3

0.
00

69
0

0.
01

92
-0

.0
22

0
-0

.0
04

12
0.

01
49

0.
02

65
-0

.0
30

6
-0

.0
10

8
(0

.0
07

61
)

(0
.0

27
8)

(0
.0

32
6)

(0
.0

02
80

)
(0

.0
08

66
)

(0
.0

23
9)

(0
.0

27
3)

(0
.0

05
24

)
(0

.0
11

8)
(0

.0
21

1)
(0

.0
24

1)
(0

.0
08

70
)

El
ec

tr
ic

ity
0.

00
87

6
0.

03
40

-0
.0

39
4

-0
.0

03
40

-0
.0

03
03

-0
.0

08
35

0.
00

95
7

0.
00

18
0

0.
00

11
4

0.
00

20
8

-0
.0

02
37

-0
.0

00
85

7
(0

.0
08

00
)

(0
.0

33
0)

(0
.0

37
6)

(0
.0

03
42

)
(0

.0
10

5)
(0

.0
28

1)
(0

.0
32

6)
(0

.0
06

05
)

(0
.0

14
0)

(0
.0

25
8)

(0
.0

29
2)

(0
.0

10
6)

Sa
fe

 w
at

er
0.

00
33

9
0.

01
19

-0
.0

14
1

-0
.0

01
14

0.
00

51
3

0.
01

36
-0

.0
15

8
-0

.0
02

90
0.

01
97

0.
02

98
-0

.0
37

3
-0

.0
12

2
(0

.0
09

53
)

(0
.0

32
3)

(0
.0

38
8)

(0
.0

03
02

)
(0

.0
11

2)
(0

.0
28

0)
(0

.0
33

3)
(0

.0
05

95
)

(0
.0

15
6)

(0
.0

19
7)

(0
.0

26
7)

(0
.0

08
46

)
To

ile
t

0.
02

58
*

0.
07

01
**

-0
.0

90
1*

*
-0

.0
05

81
**

0.
02

54
0.

05
22

**
-0

.0
66

9*
-0

.0
10

7*
*

0.
02

12
0.

03
07

-0
.0

39
2

-0
.0

12
7

(0
.0

15
2)

(0
.0

29
9)

(0
.0

42
7)

(0
.0

02
49

)
(0

.0
16

4)
(0

.0
23

9)
(0

.0
35

3)
(0

.0
05

05
)

(0
.0

19
3)

(0
.0

21
4)

(0
.0

31
4)

(0
.0

09
19

)
H

H
 in

co
m

e 
(lo

g)
-0

.0
08

05
**

*
-0

.0
29

2*
**

0.
03

44
**

*
0.

00
28

7*
**

-0
.0

09
95

**
*

-0
.0

28
0*

**
0.

03
19

**
*

0.
00

60
6*

**
-0

.0
11

9*
**

-0
.0

21
6*

**
0.

02
46

**
*

0.
00

88
8*

**
(0

.0
02

62
)

(0
.0

08
19

)
(0

.0
09

73
)

(0
.0

01
08

)
(0

.0
03

30
)

(0
.0

08
05

)
(0

.0
09

30
)

(0
.0

02
12

)
(0

.0
03

88
)

(0
.0

06
53

)
(0

.0
07

31
)

(0
.0

02
99

)
D

em
og

ra
ph

ic
 

D
ep

en
de

nc
y 

Ra
tio

0.
00

02
41

0.
00

08
73

-0
.0

01
03

-8
.6

0e
-0

5
-0

.0
01

34
-0

.0
03

76
0.

00
42

8
0.

00
08

14
-0

.0
05

81
-0

.0
10

6
0.

01
20

0.
00

43
4

(0
.0

04
99

)
(0

.0
18

0)
(0

.0
21

2)
(0

.0
01

78
)

(0
.0

05
46

)
(0

.0
15

4)
(0

.0
17

5)
(0

.0
03

35
)

(0
.0

08
26

)
(0

.0
14

5)
(0

.0
16

7)
(0

.0
06

09
)

Ec
on

om
ic

 D
ep

en
de

nc
y 

Ra
tio

0.
00

32
9

0.
01

19
-0

.0
14

0
-0

.0
01

17
0.

00
22

4
0.

00
63

0
-0

.0
07

17
-0

.0
01

36
0.

00
43

8
0.

00
79

7
-0

.0
09

07
-0

.0
03

28

(0
.0

02
23

)
(0

.0
08

03
)

(0
.0

09
46

)
(0

.0
00

80
6)

(0
.0

02
24

)
(0

.0
06

31
)

(0
.0

07
14

)
(0

.0
01

41
)

(0
.0

03
66

)
(0

.0
06

36
)

(0
.0

07
28

)
(0

.0
02

73
)

In
co

m
e 

di
ve

rs
ifi

ca
tio

n
0.

02
01

*
0.

07
28

*
-0

.0
85

7*
-0

.0
07

17
*

0.
01

87
0.

05
27

-0
.0

60
0

-0
.0

11
4

0.
01

47
0.

02
68

-0
.0

30
5

-0
.0

11
0

(0
.0

10
9)

(0
.0

37
9)

(0
.0

44
8)

(0
.0

04
07

)
(0

.0
12

1)
(0

.0
32

7)
(0

.0
37

6)
(0

.0
07

23
)

(0
.0

15
7)

(0
.0

28
6)

(0
.0

32
7)

(0
.0

11
6)

Ru
ra

l
-0

.0
29

7*
*

-0
.0

99
3*

**
0.

12
0*

*
0.

00
88

3*
*

-0
.0

33
1*

*
-0

.0
81

5*
**

0.
09

73
**

*
0.

01
73

**
*

-0
.0

23
9

-0
.0

40
2

0.
04

76
0.

01
65

(0
.0

12
5)

(0
.0

38
2)

(0
.0

46
9)

(0
.0

03
70

)
(0

.0
13

9)
(0

.0
27

5)
(0

.0
34

9)
(0

.0
06

55
)

(0
.0

16
3)

(0
.0

25
8)

(0
.0

31
1)

(0
.0

10
9)

H
H

 re
ce

iv
ed

 a
ss

ist
an

ce
 

sin
ce

 C
O

V
ID

-1
9 

ou
tb

re
ak

0.
02

30
0.

06
53

-0
.0

82
7

-0
.0

05
63

*
0.

02
69

*
0.

05
75

**
-0

.0
72

7*
*

-0
.0

11
7*

*
0.

04
41

*
0.

05
72

**
*

-0
.0

77
9*

*
-0

.0
23

4*
*

(0
.0

19
3)

(0
.0

43
3)

(0
.0

59
3)

(0
.0

03
35

)
(0

.0
15

0)
(0

.0
25

0)
(0

.0
34

9)
(0

.0
05

13
)

(0
.0

26
1)

(0
.0

20
5)

(0
.0

36
4)

(0
.0

09
75

)

O
bs

er
va

tio
ns

2,
34

4
2,

34
4

2,
34

4
2,

34
4

2,
34

6
2,

34
6

2,
34

6
2,

34
6

2,
34

7
2,

34
7

2,
34

7
2,

34
7

St
an

da
rd

 e
rr

or
s i

n 
pa

re
nt

he
se

s
**

* 
p<

0.
01

, *
* 

p<
0.

05
, *

 p
<0

.1

Ta
bl

e 
A

.4
. (

co
nt

d.
) O

rd
er

ed
 p

ro
bi

t, 
to

ta
l i

nc
om

e 
ch

an
ge

.

https://doi.org/10.36253/bae-13404


361The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Ta
bl

e 
A

.4
. (

co
nt

d.
) O

rd
er

ed
 p

ro
bi

t, 
to

ta
l i

nc
om

e 
ch

an
ge

.

Va
ria

bl
es

Ro
un

d 
4

Ro
un

d 
5

Ro
un

d 
6

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

Ye
ar

s o
f e

du
ca

tio
n

-0
.0

01
65

-0
.0

03
10

0.
00

27
8

0.
00

19
7

-0
.0

00
92

0
-0

.0
02

20
0.

00
15

1
0.

00
16

1
-0

.0
01

37
-0

.0
03

29
0.

00
20

4
0.

00
26

2
(0

.0
01

30
)

(0
.0

02
49

)
(0

.0
02

22
)

(0
.0

01
56

)
(0

.0
01

13
)

(0
.0

02
73

)
(0

.0
01

87
)

(0
.0

01
98

)
(0

.0
01

17
)

(0
.0

02
80

)
(0

.0
01

76
)

(0
.0

02
20

)
H

H
 h

ea
d 

is 
fe

m
al

e
0.

01
02

0.
01

80
-0

.0
16

8
-0

.0
11

5
0.

02
28

0.
04

77
-0

.0
35

7
-0

.0
34

7
0.

02
84

0.
05

82
-0

.0
40

4
-0

.0
46

2
(0

.0
22

8)
(0

.0
36

6)
(0

.0
35

8)
(0

.0
23

6)
(0

.0
22

5)
(0

.0
36

6)
(0

.0
31

1)
(0

.0
27

9)
(0

.0
24

1)
(0

.0
36

8)
(0

.0
29

7)
(0

.0
31

0)
A

ge
 o

f H
H

 h
ea

d
0.

00
06

73
0.

00
12

6
-0

.0
01

13
-0

.0
00

80
2

0.
00

07
35

0.
00

17
6*

-0
.0

01
21

*
-0

.0
01

29
*

0.
00

04
02

0.
00

09
65

-0
.0

00
59

9
-0

.0
00

76
8

(0
.0

00
47

3)
(0

.0
00

84
0)

(0
.0

00
75

4)
(0

.0
00

55
4)

(0
.0

00
44

7)
(0

.0
00

95
4)

(0
.0

00
67

4)
(0

.0
00

72
1)

(0
.0

00
40

0)
(0

.0
00

92
9)

(0
.0

00
57

8)
(0

.0
00

74
9)

H
H

 h
ea

d 
is 

m
ar

rie
d

0.
00

68
9

0.
01

34
-0

.0
11

8
-0

.0
08

54
0.

00
39

4
0.

00
96

3
-0

.0
06

51
-0

.0
07

07
0.

00
50

5
0.

01
24

-0
.0

07
55

-0
.0

09
93

(0
.0

20
2)

(0
.0

39
8)

(0
.0

34
4)

(0
.0

25
6)

(0
.0

18
0)

(0
.0

44
3)

(0
.0

29
6)

(0
.0

32
8)

(0
.0

18
2)

(0
.0

45
3)

(0
.0

27
0)

(0
.0

36
4)

H
H

 si
ze

-0
.0

02
96

-0
.0

05
55

0.
00

49
8

0.
00

35
3

-0
.0

04
08

-0
.0

09
75

0.
00

66
9

0.
00

71
4

-0
.0

02
19

-0
.0

05
26

0.
00

32
7

0.
00

41
9

(0
.0

03
39

)
(0

.0
06

25
)

(0
.0

05
64

)
(0

.0
03

99
)

(0
.0

03
11

)
(0

.0
07

14
)

(0
.0

04
99

)
(0

.0
05

23
)

(0
.0

03
01

)
(0

.0
07

19
)

(0
.0

04
48

)
(0

.0
05

71
)

A
ss

et
 in

de
x

-0
.0

01
26

-0
.0

02
37

0.
00

21
3

0.
00

15
1

-0
.0

02
49

-0
.0

05
95

0.
00

40
8

0.
00

43
6

-0
.0

02
28

-0
.0

05
48

0.
00

34
0

0.
00

43
6

(0
.0

02
61

)
(0

.0
04

73
)

(0
.0

04
30

)
(0

.0
03

04
)

(0
.0

02
58

)
(0

.0
05

64
)

(0
.0

03
93

)
(0

.0
04

29
)

(0
.0

02
59

)
(0

.0
05

79
)

(0
.0

03
64

)
(0

.0
04

74
)

H
H

 c
on

su
m

pt
io

n 
(lo

g)
0.

00
02

10
0.

00
03

94
-0

.0
00

35
4

-0
.0

00
25

0
0.

00
48

2
0.

01
15

-0
.0

07
92

-0
.0

08
44

0.
00

86
6

0.
02

08
-0

.0
12

9
-0

.0
16

5
(0

.0
09

68
)

(0
.0

18
1)

(0
.0

16
3)

(0
.0

11
5)

(0
.0

08
52

)
(0

.0
20

7)
(0

.0
14

1)
(0

.0
15

1)
(0

.0
08

21
)

(0
.0

19
8)

(0
.0

12
1)

(0
.0

15
9)

%
 o

f m
em

be
rs

 w
ith

 jo
b 

co
nt

ra
ct

-0
.0

68
4*

**
-0

.1
28

**
0.

11
5*

*
0.

08
15

**
-0

.0
55

7*
*

-0
.1

33
**

0.
09

14
**

0.
09

74
**

-0
.0

51
8*

*
-0

.1
24

*
0.

07
72

**
0.

09
90

**
(0

.0
26

1)
(0

.0
57

8)
(0

.0
49

1)
(0

.0
34

0)
(0

.0
22

9)
(0

.0
66

3)
(0

.0
43

9)
(0

.0
44

8)
(0

.0
23

1)
(0

.0
65

7)
(0

.0
39

3)
(0

.0
49

0)
H

H
 re

ce
iv

es
 so

ci
al

 
as

sis
ta

nc
e

-0
.0

01
83

-0
.0

03
50

0.
00

31
1

0.
00

22
2

0.
00

17
9

0.
00

42
0

-0
.0

02
92

-0
.0

03
07

-0
.0

03
23

-0
.0

08
04

0.
00

48
7

0.
00

64
0

(0
.0

23
1)

(0
.0

44
9)

(0
.0

39
5)

(0
.0

28
5)

(0
.0

20
6)

(0
.0

47
4)

(0
.0

33
3)

(0
.0

34
7)

(0
.0

20
2)

(0
.0

52
1)

(0
.0

30
9)

(0
.0

41
4)

H
H

 h
as

 fo
rm

al
 c

re
di

t
0.

01
65

0.
02

66
-0

.0
26

0
-0

.0
17

2
0.

00
19

7
0.

00
46

2
-0

.0
03

21
-0

.0
03

39
-0

.0
05

02
-0

.0
12

7
0.

00
76

4
0.

01
01

(0
.0

24
5)

(0
.0

32
2)

(0
.0

34
7)

(0
.0

22
0)

(0
.0

20
4)

(0
.0

46
5)

(0
.0

32
7)

(0
.0

34
2)

(0
.0

19
1)

(0
.0

51
7)

(0
.0

30
0)

(0
.0

40
8)

H
H

 h
as

 sa
vi

ng
s

0.
01

58
0.

02
89

-0
.0

26
0

-0
.0

18
6

0.
02

16
*

0.
04

97
*

-0
.0

34
3*

*
-0

.0
37

0*
0.

02
05

*
0.

04
73

*
-0

.0
29

5*
-0

.0
38

3*
(0

.0
12

3)
(0

.0
22

9)
(0

.0
20

2)
(0

.0
14

9)
(0

.0
11

3)
(0

.0
25

7)
(0

.0
17

4)
(0

.0
19

4)
(0

.0
11

5)
(0

.0
25

7)
(0

.0
15

7)
(0

.0
21

3)
H

H
 h

as
 b

an
k 

ac
co

un
t

-0
.0

31
3*

*
-0

.0
57

0*
*

0.
05

22
**

0.
03

60
**

-0
.0

19
7

-0
.0

46
4

0.
03

22
0.

03
39

-0
.0

26
4*

-0
.0

61
8*

*
0.

03
90

**
0.

04
92

**
(0

.0
15

6)
(0

.0
26

0)
(0

.0
24

4)
(0

.0
16

9)
(0

.0
12

8)
(0

.0
29

3)
(0

.0
20

4)
(0

.0
21

5)
(0

.0
13

5)
(0

.0
29

2)
(0

.0
18

8)
(0

.0
23

5)
H

H
 h

as
 fo

rm
al

 
in

su
ra

nc
e

-0
.0

16
5

-0
.0

36
5

0.
02

99
0.

02
31

-0
.0

19
8*

*
-0

.0
59

9*
0.

03
60

*
0.

04
37

*
-0

.0
08

28
-0

.0
21

6
0.

01
27

0.
01

72
(0

.0
12

8)
(0

.0
33

4)
(0

.0
25

2)
(0

.0
20

8)
(0

.0
10

0)
(0

.0
35

9)
(0

.0
19

4)
(0

.0
26

4)
(0

.0
13

9)
(0

.0
40

4)
(0

.0
22

4)
(0

.0
31

8)
D

ist
an

ce
 to

 m
ai

n 
ro

ad
-0

.0
00

55
9*

*
-0

.0
01

05
**

*
0.

00
09

40
**

*
0.

00
06

65
**

*
-0

.0
00

42
4*

*
-0

.0
01

01
**

0.
00

06
96

**
0.

00
07

42
**

-0
.0

00
43

1*
*

-0
.0

01
03

**
0.

00
06

41
**

0.
00

08
23

**
(0

.0
00

22
0)

(0
.0

00
40

2)
(0

.0
00

35
9)

(0
.0

00
25

4)
(0

.0
00

20
2)

(0
.0

00
49

1)
(0

.0
00

32
0)

(0
.0

00
36

8)
(0

.0
00

21
0)

(0
.0

00
49

2)
(0

.0
00

29
7)

(0
.0

00
39

8)
D

ist
an

ce
 to

 u
rb

an
 c

en
te

r
-0

.0
00

16
0*

*
-0

.0
00

30
1*

*
0.

00
02

70
**

0.
00

01
91

**
-0

.0
00

15
8*

*
-0

.0
00

37
7*

*
0.

00
02

59
**

0.
00

02
76

**
-0

.0
00

14
5*

*
-0

.0
00

34
9*

*
0.

00
02

16
**

0.
00

02
78

**
(7

.9
1e

-0
5)

(0
.0

00
13

8)
(0

.0
00

12
7)

(8
.7

7e
-0

5)
(7

.3
6e

-0
5)

(0
.0

00
14

8)
(0

.0
00

10
6)

(0
.0

00
11

4)
(7

.3
4e

-0
5)

(0
.0

00
15

2)
(9

.8
5e

-0
5)

(0
.0

00
12

5)
D

ist
an

ce
 to

 m
ar

ke
t

-0
.0

00
88

9*
*

-0
.0

01
67

**
0.

00
15

0*
*

0.
00

10
6*

*
-0

.0
00

75
5*

-0
.0

01
81

*
0.

00
12

4*
0.

00
13

2*
-0

.0
00

95
1*

*
-0

.0
02

28
**

0.
00

14
2*

*
0.

00
18

2*
*

(0
.0

00
45

1)
(0

.0
00

84
9)

(0
.0

00
75

6)
(0

.0
00

53
3)

(0
.0

00
39

6)
(0

.0
00

98
2)

(0
.0

00
66

3)
(0

.0
00

70
6)

(0
.0

00
41

5)
(0

.0
01

01
)

(0
.0

00
61

5)
(0

.0
00

79
3)

(C
on

tin
ue

d)

https://doi.org/10.36253/bae-13404


362

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Va
ria

bl
es

Ro
un

d 
4

Ro
un

d 
5

Ro
un

d 
6

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

M
ic

ro
fin

an
ce

 in
st

itu
tio

n
0.

00
55

2
0.

01
02

-0
.0

09
25

-0
.0

06
51

-0
.0

01
57

-0
.0

03
78

0.
00

25
9

0.
00

27
7

-0
.0

01
32

-0
.0

03
18

0.
00

19
7

0.
00

25
3

(0
.0

11
4)

(0
.0

21
1)

(0
.0

19
1)

(0
.0

13
3)

(0
.0

09
40

)
(0

.0
22

5)
(0

.0
15

4)
(0

.0
16

5)
(0

.0
10

3)
(0

.0
24

8)
(0

.0
15

3)
(0

.0
19

8)
El

ec
tr

ic
ity

-0
.0

00
41

1
-0

.0
00

76
8

0.
00

06
91

0.
00

04
88

-0
.0

04
58

-0
.0

10
7

0.
00

74
3

0.
00

78
3

0.
00

54
3

0.
01

34
-0

.0
08

18
-0

.0
10

7
(0

.0
14

7)
(0

.0
27

4)
(0

.0
24

6)
(0

.0
17

4)
(0

.0
13

5)
(0

.0
30

6)
(0

.0
21

8)
(0

.0
22

4)
(0

.0
12

7)
(0

.0
32

4)
(0

.0
19

3)
(0

.0
25

8)
Sa

fe
 w

at
er

0.
03

06
0.

04
31

**
-0

.0
45

5*
-0

.0
28

1*
*

0.
02

20
0.

04
21

*
-0

.0
32

9*
-0

.0
31

2*
0.

02
15

0.
04

20
*

-0
.0

29
7*

-0
.0

33
8*

(0
.0

18
7)

(0
.0

19
3)

(0
.0

24
0)

(0
.0

13
5)

(0
.0

14
9)

(0
.0

22
1)

(0
.0

19
8)

(0
.0

17
0)

(0
.0

13
8)

(0
.0

22
2)

(0
.0

17
6)

(0
.0

18
1)

To
ile

t
0.

02
13

0.
03

15
-0

.0
32

2
-0

.0
20

5
0.

01
78

0.
03

44
-0

.0
26

6
-0

.0
25

5
0.

02
35

0.
04

35
*

-0
.0

31
7

-0
.0

35
3*

(0
.0

19
1)

(0
.0

21
5)

(0
.0

25
7)

(0
.0

14
7)

(0
.0

16
8)

(0
.0

24
9)

(0
.0

22
3)

(0
.0

19
2)

(0
.0

16
8)

(0
.0

22
7)

(0
.0

19
9)

(0
.0

19
3)

H
H

 in
co

m
e 

(lo
g)

-0
.0

10
4*

**
-0

.0
19

4*
**

0.
01

74
**

*
0.

01
23

**
*

-0
.0

09
20

**
-0

.0
22

0*
**

0.
01

51
**

*
0.

01
61

**
*

-0
.0

07
62

**
-0

.0
18

3*
*

0.
01

13
**

0.
01

46
**

(0
.0

03
94

)
(0

.0
06

74
)

(0
.0

05
99

)
(0

.0
04

52
)

(0
.0

03
61

)
(0

.0
07

29
)

(0
.0

05
08

)
(0

.0
05

70
)

(0
.0

03
62

)
(0

.0
07

67
)

(0
.0

04
84

)
(0

.0
06

35
)

D
em

og
ra

ph
ic

 
D

ep
en

de
nc

y 
Ra

tio
-0

.0
00

78
7

-0
.0

01
48

0.
00

13
3

0.
00

09
38

0.
00

39
7

0.
00

94
8

-0
.0

06
51

-0
.0

06
94

-7
.2

8e
-0

5
-0

.0
00

17
5

0.
00

01
08

0.
00

01
39

(0
.0

07
62

)
(0

.0
14

2)
(0

.0
12

8)
(0

.0
09

04
)

(0
.0

06
38

)
(0

.0
16

0)
(0

.0
10

8)
(0

.0
11

6)
(0

.0
06

54
)

(0
.0

15
7)

(0
.0

09
74

)
(0

.0
12

5)
Ec

on
om

ic
 D

ep
en

de
nc

y 
Ra

tio
0.

00
64

0*
0.

01
20

*
-0

.0
10

8*
-0

.0
07

62
*

0.
00

72
3*

*
0.

01
73

**
-0

.0
11

9*
*

-0
.0

12
7*

*
0.

00
59

8*
0.

01
44

*
-0

.0
08

91
*

-0
.0

11
4*

(0
.0

03
83

)
(0

.0
06

74
)

(0
.0

06
07

)
(0

.0
04

43
)

(0
.0

03
36

)
(0

.0
07

25
)

(0
.0

04
98

)
(0

.0
05

54
)

(0
.0

03
40

)
(0

.0
07

65
)

(0
.0

04
71

)
(0

.0
06

27
)

In
co

m
e 

di
ve

rs
ifi

ca
tio

n
0.

01
15

0.
02

15
-0

.0
19

3
-0

.0
13

6
0.

00
29

0
0.

00
69

3
-0

.0
04

76
-0

.0
05

07
0.

00
71

5
0.

01
72

-0
.0

10
7

-0
.0

13
7

(0
.0

16
0)

(0
.0

30
2)

(0
.0

27
1)

(0
.0

19
1)

(0
.0

13
7)

(0
.0

32
9)

(0
.0

22
6)

(0
.0

24
0)

(0
.0

13
8)

(0
.0

33
2)

(0
.0

20
7)

(0
.0

26
3)

Ru
ra

l
-0

.0
18

4
-0

.0
31

9
0.

02
98

0.
02

05
-0

.0
14

8
-0

.0
33

0
0.

02
35

0.
02

43
-0

.0
05

40
-0

.0
12

7
0.

00
79

7
0.

01
01

(0
.0

15
8)

(0
.0

25
6)

(0
.0

24
4)

(0
.0

16
8)

(0
.0

13
7)

(0
.0

27
4)

(0
.0

20
3)

(0
.0

20
7)

(0
.0

13
7)

(0
.0

31
3)

(0
.0

19
9)

(0
.0

25
1)

H
H

 re
ce

iv
ed

 a
ss

ist
an

ce
 

sin
ce

 C
O

V
ID

-1
9 

ou
tb

re
ak

0.
00

88
8

0.
01

55
-0

.0
14

5
-0

.0
09

89
0.

01
41

0.
03

04
-0

.0
22

3
-0

.0
22

2
0.

01
33

0.
02

91
-0

.0
19

3
-0

.0
23

2
(0

.0
18

3)
(0

.0
29

1)
(0

.0
28

6)
(0

.0
18

8)
(0

.0
17

2)
(0

.0
30

9)
(0

.0
24

9)
(0

.0
23

2)
(0

.0
15

8)
(0

.0
29

7)
(0

.0
21

3)
(0

.0
24

2)

O
bs

er
va

tio
ns

2,
34

4
2,

34
4

2,
34

4
2,

34
4

2,
34

6
2,

34
6

2,
34

6
2,

34
6

2,
33

8
2,

33
8

2,
33

8
2,

33
8

St
an

da
rd

 e
rr

or
s i

n 
pa

re
nt

he
se

s
**

* 
p<

0.
01

, *
* 

p<
0.

05
, *

 p
<0

.1

Ta
bl

e 
A

.4
. (

co
nt

d.
) O

rd
er

ed
 p

ro
bi

t, 
to

ta
l i

nc
om

e 
ch

an
ge

.

https://doi.org/10.36253/bae-13404


363The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Ta
bl

e 
A

.5
. O

rd
er

ed
 p

ro
bi

t, 
fa

rm
 in

co
m

e 
ch

an
ge

.

Va
ria

bl
es

Ro
un

d 
1

Ro
un

d 
2

Ro
un

d 
3

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

Ye
ar

s o
f e

du
ca

tio
n

-0
.0

00
16

2
-0

.0
01

05
0.

00
11

0
0.

00
01

15
-0

.0
00

38
2

-0
.0

03
82

0.
00

32
3

0.
00

09
72

2.
50

e-
05

0.
00

04
65

-0
.0

00
37

3
-0

.0
00

11
7

(0
.0

00
91

0)
(0

.0
05

91
)

(0
.0

06
17

)
(0

.0
00

64
8)

(0
.0

00
63

5)
(0

.0
06

46
)

(0
.0

05
47

)
(0

.0
01

63
)

(0
.0

00
38

0)
(0

.0
07

02
)

(0
.0

05
63

)
(0

.0
01

77
)

H
H

 in
co

m
e 

(lo
g)

-0
.0

00
18

0
-0

.0
01

16
0.

00
12

1
0.

00
01

27
0.

00
27

7
0.

02
77

-0
.0

23
4

-0
.0

07
05

0.
00

12
8

0.
02

38
-0

.0
19

1
-0

.0
06

01
(0

.0
02

77
)

(0
.0

17
8)

(0
.0

18
6)

(0
.0

01
94

)
(0

.0
02

19
)

(0
.0

19
2)

(0
.0

16
2)

(0
.0

05
28

)
(0

.0
01

40
)

(0
.0

19
3)

(0
.0

15
6)

(0
.0

05
13

)
H

H
 c

on
su

m
pt

io
n 

(lo
g)

0.
00

91
5

0.
05

90
-0

.0
61

7
-0

.0
06

46
0.

00
43

3
0.

04
33

-0
.0

36
6

-0
.0

11
0

0.
00

11
3

0.
02

11
-0

.0
16

9
-0

.0
05

31
(0

.0
06

24
)

(0
.0

36
5)

(0
.0

37
9)

(0
.0

05
03

)
(0

.0
04

42
)

(0
.0

47
6)

(0
.0

39
4)

(0
.0

12
7)

(0
.0

02
69

)
(0

.0
53

9)
(0

.0
42

7)
(0

.0
13

8)
H

H
 h

ea
d 

is 
fe

m
al

e
-0

.0
09

46
-0

.0
69

6
0.

07
02

0.
00

88
1

0.
00

18
5

0.
01

79
-0

.0
15

3
-0

.0
04

41
-0

.0
02

73
-0

.0
56

0
0.

04
33

0.
01

55
(0

.0
08

17
)

(0
.0

67
3)

(0
.0

65
2)

(0
.0

10
5)

(0
.0

08
63

)
(0

.0
81

0)
(0

.0
70

3)
(0

.0
19

4)
(0

.0
04

65
)

(0
.0

96
9)

(0
.0

71
4)

(0
.0

30
0)

A
ge

 o
f H

H
 h

ea
d

0.
00

03
62

0.
00

23
3

-0
.0

02
44

-0
.0

00
25

5
0.

00
04

40
0.

00
44

1*
*

-0
.0

03
73

**
-0

.0
01

12
*

0.
00

02
07

0.
00

38
5*

-0
.0

03
08

*
-0

.0
00

97
0*

(0
.0

00
30

3)
(0

.0
01

73
)

(0
.0

01
83

)
(0

.0
00

21
0)

(0
.0

00
31

2)
(0

.0
02

11
)

(0
.0

01
81

)
(0

.0
00

64
3)

(0
.0

00
20

6)
(0

.0
02

01
)

(0
.0

01
66

)
(0

.0
00

57
4)

H
H

 h
ea

d 
is 

m
ar

rie
d

0.
00

56
5

0.
03

95
-0

.0
40

5
-0

.0
04

64
0.

00
30

6
0.

03
19

-0
.0

26
4

-0
.0

08
61

-0
.0

10
0

-0
.1

34
0.

11
6

0.
02

83
(0

.0
09

52
)

(0
.0

70
5)

(0
.0

71
2)

(0
.0

08
91

)
(0

.0
07

83
)

(0
.0

83
1)

(0
.0

67
2)

(0
.0

23
8)

(0
.0

09
57

)
(0

.0
85

6)
(0

.0
77

1)
(0

.0
17

4)
H

H
 si

ze
0.

00
24

3
0.

01
57

-0
.0

16
4

-0
.0

01
71

0.
00

05
26

0.
00

52
6

-0
.0

04
45

-0
.0

01
34

0.
00

03
62

0.
00

67
3

-0
.0

05
40

-0
.0

01
70

(0
.0

01
82

)
(0

.0
11

5)
(0

.0
11

9)
(0

.0
01

43
)

(0
.0

01
44

)
(0

.0
15

1)
(0

.0
12

7)
(0

.0
03

84
)

(0
.0

00
72

7)
(0

.0
14

9)
(0

.0
11

9)
(0

.0
03

73
)

A
ss

et
 in

de
x 

-0
.0

03
97

-0
.0

25
6

0.
02

68
0.

00
28

0
-0

.0
04

81
*

-0
.0

48
1*

*
0.

04
07

**
0.

01
22

*
-0

.0
00

42
9

-0
.0

07
99

0.
00

64
0

0.
00

20
1

(0
.0

03
26

)
(0

.0
19

5)
(0

.0
20

1)
(0

.0
02

74
)

(0
.0

02
65

)
(0

.0
21

1)
(0

.0
17

5)
(0

.0
06

47
)

(0
.0

01
17

)
(0

.0
21

4)
(0

.0
17

1)
(0

.0
05

47
)

La
nd

 is
 ir

rig
at

ed
0.

02
20

0.
10

5
-0

.1
18

-0
.0

09
01

0.
01

13
0.

09
08

-0
.0

82
7

-0
.0

19
3

-0
.0

01
32

-0
.0

26
7

0.
02

09
0.

00
70

3
(0

.0
21

7)
(0

.0
73

0)
(0

.0
89

0)
(0

.0
06

14
)

(0
.0

13
9)

(0
.0

92
4)

(0
.0

89
5)

(0
.0

17
0)

(0
.0

06
26

)
(0

.1
34

)
(0

.1
03

)
(0

.0
37

4)
O

rg
an

ic
 fe

rt
ili

ze
r

-0
.0

53
5*

*
-0

.2
21

**
*

0.
25

5*
**

0.
01

99
**

-0
.0

38
5*

-0
.2

16
**

*
0.

20
2*

**
0.

05
22

**
*

-0
.0

13
1

-0
.1

54
**

0.
12

9*
*

0.
03

81
*

(0
.0

24
4)

(0
.0

50
3)

(0
.0

64
4)

(0
.0

09
81

)
(0

.0
19

9)
(0

.0
51

0)
(0

.0
50

5)
(0

.0
18

6)
(0

.0
12

4)
(0

.0
75

8)
(0

.0
64

4)
(0

.0
22

6)
Ti

tle
 o

f t
he

 la
nd

-0
.0

08
55

-0
.0

52
5

0.
05

57
0.

00
52

6
-0

.0
10

1
-0

.0
89

2
0.

07
81

0.
02

12
-0

.0
08

94
-0

.1
30

0.
10

8
0.

03
06

(0
.0

10
6)

(0
.0

62
7)

(0
.0

67
9)

(0
.0

05
57

)
(0

.0
09

07
)

(0
.0

64
2)

(0
.0

58
5)

(0
.0

15
0)

(0
.0

09
19

)
(0

.0
84

7)
(0

.0
74

4)
(0

.0
19

0)
%

 o
f m

em
be

rs
 w

ith
 jo

b 
co

nt
ra

ct
0.

00
60

6
0.

03
91

-0
.0

40
8

-0
.0

04
27

0.
05

09
**

0.
50

9*
**

-0
.4

30
**

*
-0

.1
29

**
-0

.0
18

0
-0

.3
35

0.
26

8
0.

08
43

(0
.0

16
6)

(0
.1

07
)

(0
.1

12
)

(0
.0

12
0)

(0
.0

22
9)

(0
.1

93
)

(0
.1

64
)

(0
.0

54
7)

(0
.0

16
4)

(0
.2

21
)

(0
.1

75
)

(0
.0

61
6)

H
H

I_
cr

op
0.

02
90

**
0.

18
7*

**
-0

.1
96

**
*

-0
.0

20
5*

0.
02

71
**

0.
27

1*
**

-0
.2

29
**

*
-0

.0
68

8*
*

0.
01

64
0.

30
4*

**
-0

.2
44

**
*

-0
.0

76
7*

*
(0

.0
13

4)
(0

.0
71

2)
(0

.0
73

4)
(0

.0
12

0)
(0

.0
12

6)
(0

.0
77

2)
(0

.0
64

3)
(0

.0
27

5)
(0

.0
12

1)
(0

.0
94

3)
(0

.0
74

3)
(0

.0
32

1)
La

nd
 si

ze
 (h

ec
ta

re
s)

-0
.0

05
53

**
-0

.0
35

7*
*

0.
03

73
**

0.
00

39
0*

-0
.0

01
89

-0
.0

18
9

0.
01

60
0.

00
48

0
-0

.0
02

65
-0

.0
49

4*
*

0.
03

96
**

0.
01

24
**

(0
.0

02
80

)
(0

.0
15

0)
(0

.0
15

7)
(0

.0
02

34
)

(0
.0

01
78

)
(0

.0
15

9)
(0

.0
13

6)
(0

.0
04

13
)

(0
.0

02
09

)
(0

.0
19

7)
(0

.0
16

2)
(0

.0
05

62
)

In
co

m
e 

di
ve

rs
ifi

ca
tio

n
0.

00
66

6
0.

04
30

-0
.0

44
9

-0
.0

04
70

0.
00

41
2

0.
04

12
-0

.0
34

8
-0

.0
10

5
0.

00
06

94
0.

01
29

-0
.0

10
4

-0
.0

03
26

(0
.0

09
66

)
(0

.0
61

3)
(0

.0
64

3)
(0

.0
06

70
)

(0
.0

07
38

)
(0

.0
74

8)
(0

.0
63

4)
(0

.0
18

9)
(0

.0
04

45
)

(0
.0

83
4)

(0
.0

67
0)

(0
.0

20
8)

So
ci

al
 a

ss
ist

an
ce

0.
00

30
5

0.
01

88
-0

.0
19

9
-0

.0
01

95
-0

.0
04

07
-0

.0
43

7
0.

03
56

0.
01

22
-0

.0
03

97
-0

.0
93

8
0.

06
98

0.
02

80
(0

.0
12

3)
(0

.0
72

4)
(0

.0
77

6)
(0

.0
07

05
)

(0
.0

06
77

)
(0

.0
75

3)
(0

.0
58

6)
(0

.0
23

5)
(0

.0
04

21
)

(0
.0

89
0)

(0
.0

60
6)

(0
.0

32
4)

A
gr

ic
ul

tu
ra

l m
ac

hi
ne

ry
0.

00
84

9
0.

05
03

-0
.0

53
5

-0
.0

05
22

0.
00

48
3

0.
04

61
-0

.0
39

7
-0

.0
11

2
0.

01
35

0.
18

9*
**

-0
.1

62
**

*
-0

.0
40

2*
**

(0
.0

10
0)

(0
.0

54
4)

(0
.0

59
5)

(0
.0

05
13

)
(0

.0
07

25
)

(0
.0

61
0)

(0
.0

54
3)

(0
.0

14
0)

(0
.0

10
9)

(0
.0

58
7)

(0
.0

55
8)

(0
.0

14
2)

(C
on

tin
ue

d)

https://doi.org/10.36253/bae-13404


364

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Va
ria

bl
es

Ro
un

d 
1

Ro
un

d 
2

Ro
un

d 
3

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

Fo
rm

al
 c

re
di

t
0.

01
03

0.
05

80
-0

.0
62

6
-0

.0
05

68
0.

00
72

7
0.

06
26

-0
.0

55
3

-0
.0

14
6

0.
00

29
9

0.
04

82
-0

.0
39

8
-0

.0
11

4
(0

.0
12

4)
(0

.0
60

5)
(0

.0
67

8)
(0

.0
05

30
)

(0
.0

08
73

)
(0

.0
66

3)
(0

.0
60

5)
(0

.0
14

6)
(0

.0
05

25
)

(0
.0

72
6)

(0
.0

61
0)

(0
.0

16
8)

Sa
vi

ng
s

0.
01

55
*

0.
09

73
*

-0
.1

03
*

-0
.0

09
88

*
0.

00
19

6
0.

01
94

-0
.0

16
4

-0
.0

04
91

0.
00

89
7

0.
13

8*
*

-0
.1

13
**

-0
.0

34
2*

*
(0

.0
09

34
)

(0
.0

54
1)

(0
.0

58
1)

(0
.0

05
95

)
(0

.0
05

68
)

(0
.0

55
5)

(0
.0

47
3)

(0
.0

13
9)

(0
.0

07
19

)
(0

.0
58

2)
(0

.0
49

2)
(0

.0
15

5)
Ba

nk
 a

cc
ou

nt
-0

.0
04

10
-0

.0
26

3
0.

02
74

0.
00

29
6

-0
.0

06
37

-0
.0

65
2

0.
05

50
0.

01
66

-0
.0

06
67

-0
.1

28
**

0.
10

1*
*

0.
03

39
*

(0
.0

08
47

)
(0

.0
55

4)
(0

.0
57

5)
(0

.0
06

42
)

(0
.0

06
20

)
(0

.0
60

4)
(0

.0
51

3)
(0

.0
15

6)
(0

.0
05

49
)

(0
.0

63
1)

(0
.0

49
8)

(0
.0

18
8)

Fo
rm

al
 in

su
ra

nc
e

0.
00

24
5

0.
01

53
-0

.0
16

1
-0

.0
01

62
-0

.0
07

40
-0

.0
87

8
0.

06
92

0.
02

61
-0

.0
03

84
-0

.0
94

4
0.

07
09

0.
02

73
(0

.0
10

0)
(0

.0
59

9)
(0

.0
63

7)
(0

.0
06

27
)

(0
.0

05
63

)
(0

.0
71

3)
(0

.0
52

3)
(0

.0
25

0)
(0

.0
03

54
)

(0
.0

87
0)

(0
.0

61
1)

(0
.0

29
2)

D
ist

an
ce

 to
 ro

ad
-0

.0
00

10
1

-0
.0

00
65

3
0.

00
06

83
7.

14
e-

05
-3

.8
7e

-0
5

-0
.0

00
38

7
0.

00
03

27
9.

83
e-

05
1.

52
e-

05
0.

00
02

83
-0

.0
00

22
7

-7
.1

3e
-0

5
(0

.0
00

10
7)

(0
.0

00
69

5)
(0

.0
00

73
0)

(7
.5

0e
-0

5)
(7

.2
1e

-0
5)

(0
.0

00
72

1)
(0

.0
00

61
6)

(0
.0

00
17

8)
(5

.1
2e

-0
5)

(0
.0

00
90

1)
(0

.0
00

72
2)

(0
.0

00
23

0)
D

ist
an

ce
 to

 u
rb

an
 c

en
te

r
1.

70
e-

05
0.

00
01

10
-0

.0
00

11
5

-1
.2

0e
-0

5
1.

04
e-

05
0.

00
01

04
-8

.8
4e

-0
5

-2
.6

6e
-0

5
1.

17
e-

05
0.

00
02

18
-0

.0
00

17
5

-5
.5

0e
-0

5
(4

.0
0e

-0
5)

(0
.0

00
25

9)
(0

.0
00

26
9)

(2
.9

5e
-0

5)
(2

.9
4e

-0
5)

(0
.0

00
28

7)
(0

.0
00

24
3)

(7
.3

6e
-0

5)
(1

.8
7e

-0
5)

(0
.0

00
30

8)
(0

.0
00

24
8)

(7
.8

1e
-0

5)
D

ist
an

ce
 to

 m
ar

ke
t

-0
.0

00
45

8
-0

.0
02

95
*

0.
00

30
9*

0.
00

03
23

-0
.0

00
73

0*
*

-0
.0

07
30

**
*

0.
00

61
7*

**
0.

00
18

6*
**

-0
.0

00
47

0
-0

.0
08

74
**

*
0.

00
70

1*
**

0.
00

22
0*

**
(0

.0
00

27
9)

(0
.0

01
75

)
(0

.0
01

83
)

(0
.0

00
22

0)
(0

.0
00

34
9)

(0
.0

02
48

)
(0

.0
02

19
)

(0
.0

00
70

1)
(0

.0
00

32
7)

(0
.0

02
70

)
(0

.0
02

21
)

(0
.0

00
82

2)
M

ic
ro

fin
an

ce
 in

st
itu

tio
n

-0
.0

03
80

-0
.0

25
4

0.
02

63
0.

00
28

7
0.

00
30

4
0.

02
92

-0
.0

25
0

-0
.0

07
23

-0
.0

02
41

-0
.0

49
2

0.
03

87
0.

01
30

(0
.0

06
77

)
(0

.0
46

7)
(0

.0
48

2)
(0

.0
05

28
)

(0
.0

06
12

)
(0

.0
56

6)
(0

.0
48

7)
(0

.0
14

0)
(0

.0
03

56
)

(0
.0

65
1)

(0
.0

50
6)

(0
.0

18
0)

El
ec

tr
ic

ity
-0

.0
17

2*
-0

.1
03

**
0.

10
9*

*
0.

01
11

*
-0

.0
03

93
-0

.0
38

7
0.

03
29

0.
00

97
2

-0
.0

03
77

-0
.0

64
7

0.
05

24
0.

01
60

(0
.0

09
81

)
(0

.0
43

5)
(0

.0
48

5)
(0

.0
05

72
)

(0
.0

05
40

)
(0

.0
51

7)
(0

.0
44

2)
(0

.0
12

9)
(0

.0
03

93
)

(0
.0

57
3)

(0
.0

47
1)

(0
.0

13
9)

Sa
fe

 w
at

er
0.

03
79

**
0.

15
9*

**
-0

.1
85

**
*

-0
.0

11
6*

*
0.

02
24

0.
15

2*
*

-0
.1

44
**

-0
.0

29
6*

*
0.

00
47

8
0.

07
23

-0
.0

61
0

-0
.0

16
1

(0
.0

18
2)

(0
.0

49
1)

(0
.0

62
1)

(0
.0

05
42

)
(0

.0
15

1)
(0

.0
64

3)
(0

.0
67

0)
(0

.0
12

7)
(0

.0
06

44
)

(0
.0

76
4)

(0
.0

66
9)

(0
.0

15
7)

To
ile

t
0.

26
1

0.
22

2*
*

-0
.4

68
**

*
-0

.0
14

5*
*

0.
10

5*
0.

26
3*

**
-0

.3
27

**
*

-0
.0

40
6*

**
0.

01
78

0.
16

8
-0

.1
55

-0
.0

30
7

(0
.1

94
)

(0
.0

92
9)

(0
.1

07
)

(0
.0

06
35

)
(0

.0
59

9)
(0

.0
39

3)
(0

.0
74

4)
(0

.0
11

0)
(0

.0
33

9)
(0

.1
69

)
(0

.1
80

)
(0

.0
22

0)
D

em
og

ra
ph

ic
 

D
ep

en
de

nc
y 

Ra
tio

0.
00

34
6

0.
02

23
-0

.0
23

3
-0

.0
02

44
0.

00
15

1
0.

01
51

-0
.0

12
8

-0
.0

03
85

-0
.0

00
88

3
-0

.0
16

4
0.

01
32

0.
00

41
4

(0
.0

04
42

)
(0

.0
27

0)
(0

.0
28

4)
(0

.0
03

13
)

(0
.0

03
09

)
(0

.0
30

3)
(0

.0
25

7)
(0

.0
07

74
)

(0
.0

01
81

)
(0

.0
31

7)
(0

.0
25

5)
(0

.0
08

04
)

Ec
on

om
ic

 D
ep

en
de

nc
y 

Ra
tio

0.
00

01
86

0.
00

12
0

-0
.0

01
25

-0
.0

00
13

1
-0

.0
01

55
-0

.0
15

5
0.

01
31

0.
00

39
5

5.
26

e-
05

0.
00

09
78

-0
.0

00
78

4
-0

.0
00

24
7

(0
.0

01
89

)
(0

.0
12

2)
(0

.0
12

7)
(0

.0
01

33
)

(0
.0

01
65

)
(0

.0
14

9)
(0

.0
12

7)
(0

.0
03

87
)

(0
.0

00
83

6)
(0

.0
15

6)
(0

.0
12

5)
(0

.0
03

96
)

Ru
ra

l
-0

.0
01

60
-0

.0
10

1
0.

01
07

0.
00

10
8

0.
01

17
*

0.
14

9
-0

.1
09

*
-0

.0
51

9
0.

00
45

6
0.

11
1

-0
.0

82
0

-0
.0

34
0

(0
.0

13
8)

(0
.0

85
5)

(0
.0

90
4)

(0
.0

08
95

)
(0

.0
06

99
)

(0
.0

97
3)

(0
.0

59
1)

(0
.0

46
3)

(0
.0

04
32

)
(0

.1
13

)
(0

.0
75

8)
(0

.0
40

8)
H

H
 re

ce
iv

ed
 a

ss
ist

an
ce

 
sin

ce
 C

O
V

ID
-1

9 
ou

tb
re

ak
0.

05
50

**
0.

20
3*

**
-0

.2
45

**
*

-0
.0

13
6*

*
0.

05
33

**
0.

26
7*

**
-0

.2
78

**
*

-0
.0

41
9*

**
0.

00
49

4
0.

07
65

-0
.0

64
1

-0
.0

17
4

(0
.0

25
3)

(0
.0

53
4)

(0
.0

72
6)

(0
.0

06
07

)
(0

.0
22

2)
(0

.0
52

9)
(0

.0
59

0)
(0

.0
12

3)
(0

.0
05

37
)

(0
.0

61
7)

(0
.0

53
2)

(0
.0

13
7)

O
bs

er
va

tio
ns

67
8

67
8

67
8

67
8

53
5

53
5

53
5

53
5

50
3

50
3

50
3

50
3

St
an

da
rd

 e
rr

or
s i

n 
pa

re
nt

he
se

s
**

* 
p<

0.
01

, *
* 

p<
0.

05
, *

 p
<0

.1

Ta
bl

e 
A

.5
 (c

on
td

.).
 O

rd
er

ed
 p

ro
bi

t, 
fa

rm
 in

co
m

e 
ch

an
ge

.

https://doi.org/10.36253/bae-13404


365The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Ta
bl

e 
A

.5
 (c

on
td

.).
 O

rd
er

ed
 p

ro
bi

t, 
fa

rm
 in

co
m

e 
ch

an
ge

.

Va
ria

bl
es

Ro
un

d 
4

Ro
un

d 
5

Ro
un

d 
6

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

Ye
ar

s o
f e

du
ca

tio
n

8.
28

e-
05

0.
00

08
70

-0
.0

00
48

2
-0

.0
00

47
1

-0
.0

00
28

3
-0

.0
04

82
0.

00
15

6
0.

00
35

5
-1

.5
5e

-0
5

-0
.0

01
21

0.
00

02
87

0.
00

09
38

(0
.0

00
61

3)
(0

.0
06

37
)

(0
.0

03
52

)
(0

.0
03

46
)

(0
.0

00
43

2)
(0

.0
06

92
)

(0
.0

02
21

)
(0

.0
05

12
)

(7
.7

3e
-0

5)
(0

.0
06

06
)

(0
.0

01
44

)
(0

.0
04

70
)

H
H

 in
co

m
e 

(lo
g)

-0
.0

01
03

-0
.0

10
8

0.
00

59
9

0.
00

58
5

-0
.0

01
79

-0
.0

30
4

0.
00

98
1

0.
02

24
-0

.0
00

30
4

-0
.0

23
7

0.
00

56
3

0.
01

84
(0

.0
01

69
)

(0
.0

17
6)

(0
.0

09
69

)
(0

.0
09

63
)

(0
.0

01
54

)
(0

.0
19

9)
(0

.0
06

48
)

(0
.0

14
8)

(0
.0

00
31

7)
(0

.0
18

2)
(0

.0
04

50
)

(0
.0

14
2)

H
H

 c
on

su
m

pt
io

n 
(lo

g)
0.

00
50

4
0.

05
29

-0
.0

29
3

-0
.0

28
6

0.
00

24
5

0.
04

17
-0

.0
13

5
-0

.0
30

7
-0

.0
00

21
7

-0
.0

16
9

0.
00

40
1

0.
01

31
(0

.0
03

79
)

(0
.0

43
6)

(0
.0

23
3)

(0
.0

23
9)

(0
.0

02
72

)
(0

.0
50

4)
(0

.0
15

7)
(0

.0
37

3)
(0

.0
00

57
9)

(0
.0

43
9)

(0
.0

10
5)

(0
.0

34
0)

H
H

 h
ea

d 
is 

fe
m

al
e

0.
00

47
8

0.
04

60
-0

.0
27

3
-0

.0
23

6
0.

00
34

6
0.

05
31

-0
.0

19
3

-0
.0

37
2

-0
.0

00
20

0
-0

.0
16

1
0.

00
36

1
0.

01
27

(0
.0

11
4)

(0
.0

97
9)

(0
.0

62
1)

(0
.0

47
2)

(0
.0

06
29

)
(0

.0
83

5)
(0

.0
33

8)
(0

.0
55

8)
(0

.0
00

98
1)

(0
.0

81
7)

(0
.0

17
2)

(0
.0

65
5)

A
ge

 o
f H

H
 h

ea
d

0.
00

05
45

0.
00

57
2*

**
-0

.0
03

17
**

*
-0

.0
03

10
**

0.
00

02
75

0.
00

46
8*

**
-0

.0
01

51
**

-0
.0

03
45

**
5.

44
e-

05
0.

00
42

4*
*

-0
.0

01
01

**
-0

.0
03

29
**

(0
.0

00
35

3)
(0

.0
02

08
)

(0
.0

01
14

)
(0

.0
01

31
)

(0
.0

00
24

8)
(0

.0
01

80
)

(0
.0

00
62

7)
(0

.0
01

45
)

(4
.5

1e
-0

5)
(0

.0
01

82
)

(0
.0

00
48

8)
(0

.0
01

44
)

H
H

 h
ea

d 
is 

m
ar

rie
d

-0
.0

00
76

3
-0

.0
07

91
0.

00
44

4
0.

00
42

4
-0

.0
03

12
-0

.0
48

6
0.

01
75

0.
03

42
-0

.0
00

82
1

-0
.0

56
0

0.
01

55
0.

04
13

(0
.0

08
52

)
(0

.0
88

2)
(0

.0
50

0)
(0

.0
46

8)
(0

.0
05

65
)

(0
.0

81
8)

(0
.0

32
3)

(0
.0

54
9)

(0
.0

01
37

)
(0

.0
77

3)
(0

.0
24

6)
(0

.0
54

2)
H

H
 si

ze
0.

00
05

91
0.

00
62

1
-0

.0
03

44
-0

.0
03

36
0.

00
08

54
0.

01
45

-0
.0

04
69

-0
.0

10
7

8.
68

e-
05

0.
00

67
7

-0
.0

01
61

-0
.0

05
25

(0
.0

01
25

)
(0

.0
13

9)
(0

.0
07

66
)

(0
.0

07
47

)
(0

.0
00

73
1)

(0
.0

14
0)

(0
.0

04
44

)
(0

.0
10

3)
(0

.0
00

15
7)

(0
.0

11
5)

(0
.0

02
75

)
(0

.0
08

95
)

A
ss

et
 in

de
x 

-0
.0

01
91

-0
.0

20
1

0.
01

11
0.

01
09

-0
.0

00
49

0
-0

.0
08

34
0.

00
26

9
0.

00
61

4
-3

.4
8e

-0
5

-0
.0

02
71

0.
00

06
44

0.
00

21
0

(0
.0

01
72

)
(0

.0
15

9)
(0

.0
09

05
)

(0
.0

08
58

)
(0

.0
01

18
)

(0
.0

19
4)

(0
.0

06
31

)
(0

.0
14

2)
(0

.0
00

23
9)

(0
.0

18
7)

(0
.0

04
43

)
(0

.0
14

5)
La

nd
 is

 ir
rig

at
ed

-0
.0

04
61

-0
.0

56
4

0.
02

78
0.

03
32

-0
.0

05
32

-0
.1

31
*

0.
02

21
**

0.
11

4
-0

.0
01

48
-0

.2
27

**
*

-0
.0

24
0

0.
25

2*
**

(0
.0

07
88

)
(0

.1
05

)
(0

.0
45

4)
(0

.0
67

1)
(0

.0
04

50
)

(0
.0

78
3)

(0
.0

09
37

)
(0

.0
78

7)
(0

.0
01

06
)

(0
.0

50
0)

(0
.0

30
9)

(0
.0

76
3)

O
rg

an
ic

 fe
rt

ili
ze

r
-0

.0
07

00
-0

.0
65

7
0.

03
88

0.
03

40
0.

00
08

99
0.

01
56

-0
.0

04
90

-0
.0

11
5

0.
00

02
32

0.
01

86
-0

.0
04

27
-0

.0
14

5
(0

.0
09

97
)

(0
.0

74
6)

(0
.0

46
9)

(0
.0

37
6)

(0
.0

04
48

)
(0

.0
79

9)
(0

.0
24

7)
(0

.0
59

7)
(0

.0
00

84
4)

(0
.0

68
7)

(0
.0

15
5)

(0
.0

54
1)

Ti
tle

 o
f t

he
 la

nd
-0

.0
14

8
-0

.1
28

*
0.

07
79

0.
06

47
*

-0
.0

07
04

-0
.1

08
0.

03
98

0.
07

53
9.

23
e-

05
0.

00
71

9
-0

.0
01

68
-0

.0
05

59
(0

.0
12

6)
(0

.0
72

6)
(0

.0
50

0)
(0

.0
34

8)
(0

.0
07

23
)

(0
.0

73
1)

(0
.0

30
7)

(0
.0

48
8)

(0
.0

00
91

5)
(0

.0
71

1)
(0

.0
16

4)
(0

.0
55

6)
%

 o
f m

em
be

rs
 w

ith
 jo

b 
co

nt
ra

ct
-0

.0
31

4
-0

.3
30

*
0.

18
3*

0.
17

8*
-0

.0
05

80
-0

.0
98

6
0.

03
18

0.
07

26
-0

.0
01

67
-0

.1
30

0.
03

09
0.

10
1

(0
.0

20
5)

(0
.1

70
)

(0
.0

96
5)

(0
.0

93
8)

(0
.0

11
3)

(0
.1

82
)

(0
.0

58
6)

(0
.1

34
)

(0
.0

02
50

)
(0

.1
74

)
(0

.0
41

1)
(0

.1
36

)
H

H
I_

cr
op

0.
01

64
0.

17
2*

-0
.0

95
5*

-0
.0

93
3*

0.
00

46
2

0.
07

87
-0

.0
25

4
-0

.0
57

9
0.

00
18

7
0.

14
6*

-0
.0

34
6*

-0
.1

13
*

(0
.0

11
9)

(0
.0

88
4)

(0
.0

50
5)

(0
.0

50
0)

(0
.0

07
06

)
(0

.0
94

9)
(0

.0
30

6)
(0

.0
71

2)
(0

.0
01

64
)

(0
.0

77
4)

(0
.0

18
9)

(0
.0

61
8)

La
nd

 si
ze

 (h
ec

ta
re

s)
-0

.0
01

38
-0

.0
14

5
0.

00
80

5
0.

00
78

6
-0

.0
01

89
-0

.0
32

1*
*

0.
01

04
*

0.
02

37
**

-0
.0

00
32

5
-0

.0
25

3*
0.

00
60

1
0.

01
96

*
(0

.0
01

31
)

(0
.0

12
2)

(0
.0

06
90

)
(0

.0
06

55
)

(0
.0

01
66

)
(0

.0
15

9)
(0

.0
05

73
)

(0
.0

11
8)

(0
.0

00
28

3)
(0

.0
15

0)
(0

.0
03

96
)

(0
.0

11
5)

In
co

m
e 

di
ve

rs
ifi

ca
tio

n
0.

00
64

8
0.

06
81

-0
.0

37
8

-0
.0

36
9

-0
.0

01
44

-0
.0

24
5

0.
00

78
9

0.
01

80
-0

.0
01

07
-0

.0
83

3
0.

01
98

0.
06

46
(0

.0
07

38
)

(0
.0

74
8)

(0
.0

41
3)

(0
.0

40
8)

(0
.0

04
37

)
(0

.0
70

3)
(0

.0
22

7)
(0

.0
51

9)
(0

.0
01

12
)

(0
.0

68
9)

(0
.0

16
7)

(0
.0

53
9)

So
ci

al
 a

ss
ist

an
ce

-0
.0

04
63

-0
.0

55
5

0.
02

78
0.

03
23

0.
00

46
7

0.
06

72
-0

.0
25

4
-0

.0
46

4
0.

00
07

47
0.

05
09

-0
.0

14
1

-0
.0

37
5

(0
.0

06
37

)
(0

.0
79

7)
(0

.0
35

2)
(0

.0
50

8)
(0

.0
06

33
)

(0
.0

75
4)

(0
.0

32
4)

(0
.0

48
9)

(0
.0

01
45

)
(0

.0
80

0)
(0

.0
25

5)
(0

.0
56

1)
A

gr
ic

ul
tu

ra
l m

ac
hi

ne
ry

0.
00

82
3

0.
07

85
-0

.0
46

3
-0

.0
40

4
0.

01
27

0.
17

3*
**

-0
.0

69
1*

*
-0

.1
16

**
*

0.
00

10
6

0.
07

28
-0

.0
20

1
-0

.0
53

8
(0

.0
07

82
)

(0
.0

56
1)

(0
.0

36
0)

(0
.0

27
9)

(0
.0

10
6)

(0
.0

56
6)

(0
.0

28
1)

(0
.0

38
7)

(0
.0

01
15

)
(0

.0
54

7)
(0

.0
17

4)
(0

.0
38

8)

(C
on

tin
ue

d)

https://doi.org/10.36253/bae-13404


366

Bio-based and Applied Economics 12(4): 333-366, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-13404

Margherita Squarcina, Donato Romano

Va
ria

bl
es

Ro
un

d 
4

Ro
un

d 
5

Ro
un

d 
6

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
2

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

1
M

ar
gi

na
l 

eff
ec

ts
 fo

r 0
M

ar
gi

na
l 

eff
ec

ts
 fo

r +
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r -

2
M

ar
gi

na
l 

eff
ec

ts
 fo

r -
1

M
ar

gi
na

l 
eff

ec
ts

 fo
r 0

M
ar

gi
na

l 
eff

ec
ts

 fo
r +

1

Fo
rm

al
 c

re
di

t
-0

.0
01

72
-0

.0
19

0
0.

01
03

0.
01

05
-0

.0
00

40
3

-0
.0

07
00

0.
00

22
2

0.
00

51
8

-0
.0

00
36

2
-0

.0
30

9
0.

00
64

8
0.

02
48

(0
.0

05
51

)
(0

.0
62

2)
(0

.0
32

9)
(0

.0
34

8)
(0

.0
04

08
)

(0
.0

71
3)

(0
.0

22
3)

(0
.0

53
1)

(0
.0

00
74

3)
(0

.0
65

4)
(0

.0
12

1)
(0

.0
54

1)
Sa

vi
ng

s
0.

02
29

**
0.

18
8*

**
-0

.1
08

**
*

-0
.1

03
**

*
0.

01
30

0.
17

7*
**

-0
.0

60
5*

*
-0

.1
29

**
*

0.
00

34
7

0.
22

0*
**

-0
.0

60
1*

**
-0

.1
63

**
*

(0
.0

11
7)

(0
.0

49
6)

(0
.0

30
5)

(0
.0

29
7)

(0
.0

10
3)

(0
.0

61
4)

(0
.0

23
7)

(0
.0

46
0)

(0
.0

02
72

)
(0

.0
53

8)
(0

.0
19

5)
(0

.0
40

2)
Ba

nk
 a

cc
ou

nt
-0

.0
17

3*
-0

.1
76

**
*

0.
09

28
**

*
0.

10
1*

**
-0

.0
06

23
-0

.1
10

0.
03

32
0.

08
28

-0
.0

02
33

-0
.1

79
**

*
0.

03
57

**
*

0.
14

5*
**

(0
.0

09
53

)
(0

.0
57

1)
(0

.0
32

2)
(0

.0
35

9)
(0

.0
05

82
)

(0
.0

67
9)

(0
.0

20
8)

(0
.0

52
7)

(0
.0

01
83

)
(0

.0
52

8)
(0

.0
13

4)
(0

.0
46

2)
Fo

rm
al

 in
su

ra
nc

e
-0

.0
09

37
*

-0
.1

39
**

0.
05

91
**

*
0.

08
90

*
-0

.0
05

49
-0

.1
39

**
0.

02
64

**
0.

11
8*

-0
.0

01
41

-0
.2

00
**

*
0.

00
29

2
0.

19
8*

**
(0

.0
05

16
)

(0
.0

62
6)

(0
.0

20
6)

(0
.0

49
0)

(0
.0

04
28

)
(0

.0
69

5)
(0

.0
10

3)
(0

.0
68

8)
(0

.0
01

01
)

(0
.0

56
9)

(0
.0

22
3)

(0
.0

74
6)

D
ist

an
ce

 to
 ro

ad
-6

.2
4e

-0
5

-0
.0

00
65

5
0.

00
03

63
0.

00
03

55
-7

.6
6e

-0
5

-0
.0

01
30

*
0.

00
04

21
0.

00
09

59
*

-1
.2

5e
-0

5
-0

.0
00

97
5

0.
00

02
32

0.
00

07
56

(7
.1

3e
-0

5)
(0

.0
00

72
6)

(0
.0

00
41

2)
(0

.0
00

38
4)

(6
.4

2e
-0

5)
(0

.0
00

78
3)

(0
.0

00
27

4)
(0

.0
00

56
5)

(1
.3

7e
-0

5)
(0

.0
00

87
3)

(0
.0

00
21

7)
(0

.0
00

67
6)

D
ist

an
ce

 to
 u

rb
an

 c
en

te
r

-2
.9

3e
-0

6
-3

.0
8e

-0
5

1.
71

e-
05

1.
67

e-
05

3.
12

e-
05

0.
00

05
32

*
-0

.0
00

17
2*

-0
.0

00
39

1*
3.

54
e-

06
0.

00
02

76
-6

.5
5e

-0
5

-0
.0

00
21

4
(2

.1
3e

-0
5)

(0
.0

00
22

4)
(0

.0
00

12
5)

(0
.0

00
12

1)
(2

.6
5e

-0
5)

(0
.0

00
27

4)
(8

.9
0e

-0
5)

(0
.0

00
20

9)
(4

.9
0e

-0
6)

(0
.0

00
34

7)
(8

.5
1e

-0
5)

(0
.0

00
26

8)
D

ist
an

ce
 to

 m
ar

ke
t

-0
.0

00
61

7*
-0

.0
06

49
**

*
0.

00
35

9*
**

0.
00

35
1*

**
-0

.0
00

41
5

-0
.0

07
06

**
*

0.
00

22
8*

**
0.

00
52

0*
**

-0
.0

00
10

1
-0

.0
07

85
**

*
0.

00
18

6*
**

0.
00

60
9*

**
(0

.0
00

32
2)

(0
.0

01
63

)
(0

.0
01

08
)

(0
.0

00
91

7)
(0

.0
00

30
2)

(0
.0

01
78

)
(0

.0
00

79
9)

(0
.0

01
28

)
(7

.0
1e

-0
5)

(0
.0

01
84

)
(0

.0
00

67
1)

(0
.0

01
40

)
M

ic
ro

fin
an

ce
 in

st
itu

tio
n

-0
.0

05
83

-0
.0

67
4

0.
03

46
0.

03
86

-0
.0

09
12

-0
.1

94
**

*
0.

03
24

**
0.

17
0*

**
-0

.0
01

31
-0

.1
19

**
0.

01
67

**
0.

10
3*

*
(0

.0
05

15
)

(0
.0

59
3)

(0
.0

28
2)

(0
.0

36
2)

(0
.0

06
37

)
(0

.0
54

3)
(0

.0
14

9)
(0

.0
59

3)
(0

.0
00

93
7)

(0
.0

52
4)

(0
.0

07
93

)
(0

.0
50

4)
El

ec
tr

ic
ity

-0
.0

03
52

-0
.0

35
9

0.
02

02
0.

01
92

-0
.0

06
18

-0
.0

92
4*

0.
03

13
0.

06
73

*
0.

00
01

39
0.

01
10

-0
.0

02
59

-0
.0

08
52

(0
.0

05
13

)
(0

.0
50

3)
(0

.0
28

8)
(0

.0
26

6)
(0

.0
05

10
)

(0
.0

54
9)

(0
.0

20
1)

(0
.0

39
3)

(0
.0

00
62

8)
(0

.0
50

2)
(0

.0
11

8)
(0

.0
39

1)
Sa

fe
 w

at
er

0.
02

79
0.

18
2*

**
-0

.1
33

**
-0

.0
76

6*
**

0.
01

20
0.

14
0*

*
-0

.0
63

2*
-0

.0
88

6*
*

0.
00

22
2

0.
13

3*
*

-0
.0

47
7

-0
.0

87
8*

*
(0

.0
18

9)
(0

.0
62

3)
(0

.0
56

2)
(0

.0
23

7)
(0

.0
09

40
)

(0
.0

61
6)

(0
.0

33
8)

(0
.0

34
7)

(0
.0

02
07

)
(0

.0
65

9)
(0

.0
30

1)
(0

.0
38

3)
To

ile
t

0.
00

33
4

0.
03

20
-0

.0
18

9
-0

.0
16

4
0.

01
98

0.
17

9*
*

-0
.0

91
0

-0
.1

07
**

0.
00

82
5

0.
25

6*
**

-0
.1

20
**

-0
.1

44
**

*
(0

.0
11

8)
(0

.1
04

)
(0

.0
65

0)
(0

.0
50

4)
(0

.0
17

6)
(0

.0
90

3)
(0

.0
62

5)
(0

.0
42

5)
(0

.0
07

26
)

(0
.0

90
5)

(0
.0

61
2)

(0
.0

37
0)

D
em

og
ra

ph
ic

 
D

ep
en

de
nc

y 
Ra

tio
0.

00
17

9
0.

01
88

-0
.0

10
4

-0
.0

10
2

-0
.0

00
28

4
-0

.0
04

84
0.

00
15

6
0.

00
35

6
0.

00
03

74
0.

02
92

-0
.0

06
93

-0
.0

22
6

(0
.0

02
79

)
(0

.0
28

8)
(0

.0
15

8)
(0

.0
15

8)
(0

.0
01

92
)

(0
.0

32
2)

(0
.0

10
4)

(0
.0

23
8)

(0
.0

00
45

0)
(0

.0
28

8)
(0

.0
07

07
)

(0
.0

22
3)

Ec
on

om
ic

 D
ep

en
de

nc
y 

Ra
tio

-0
.0

00
51

3
-0

.0
05

39
0.

00
29

9
0.

00
29

2
-0

.0
00

11
1

-0
.0

01
88

0.
00

06
08

0.
00

13
9

0.
00

01
31

0.
01

02
-0

.0
02

42
-0

.0
07

90
(0

.0
01

37
)

(0
.0

13
9)

(0
.0

07
78

)
(0

.0
07

54
)

(0
.0

00
77

1)
(0

.0
12

9)
(0

.0
04

18
)

(0
.0

09
53

)
(0

.0
00

19
4)

(0
.0

13
3)

(0
.0

03
14

)
(0

.0
10

4)
Ru

ra
l

0.
00

21
3

0.
02

36
-0

.0
12

5
-0

.0
13

2
0.

00
10

5
0.

01
86

-0
.0

05
68

-0
.0

14
0

-0
.0

00
58

4
-0

.0
41

5
0.

01
14

0.
03

07
(0

.0
08

23
)

(0
.0

93
9)

(0
.0

47
9)

(0
.0

54
2)

(0
.0

05
31

)
(0

.0
98

0)
(0

.0
28

3)
(0

.0
75

0)
(0

.0
01

55
)

(0
.0

96
9)

(0
.0

29
7)

(0
.0

68
8)

H
H

 re
ce

iv
ed

 a
ss

ist
an

ce
 

sin
ce

 C
O

V
ID

-1
9 

ou
tb

re
ak

-0
.0

01
48

-0
.0

16
0

0.
00

87
1

0.
00

88
0

-0
.0

03
37

-0
.0

62
2

0.
01

75
0.

04
81

0.
00

01
67

0.
01

27
-0

.0
03

11
-0

.0
09

75
(0

.0
05

38
)

(0
.0

60
1)

(0
.0

32
0)

(0
.0

33
4)

(0
.0

03
97

)
(0

.0
68

0)
(0

.0
16

6)
(0

.0
55

3)
(0

.0
00

84
0)

(0
.0

61
3)

(0
.0

15
5)

(0
.0

46
7)

O
bs

er
va

tio
ns

50
6

50
6

50
6

50
6

49
7

49
7

49
7

49
7

49
6

49
6

49
6

49
6

St
an

da
rd

 e
rr

or
s i

n 
pa

re
nt

he
se

s.
**

* 
p<

0.
01

, *
* 

p<
0.

05
, *

 p
<0

.1

Ta
bl

e 
A

.5
 (c

on
td

.).
 O

rd
er

ed
 p

ro
bi

t, 
fa

rm
 in

co
m

e 
ch

an
ge

.

https://doi.org/10.36253/bae-13404

	Modeling conversion to organic agriculture with an EU-wide farm model
	Dimitrios Kremmydas*, Pavel Ciaian, Edoardo Baldoni
	A systematic literature review on the rural-urban economic well-being gap in Europe
	Cesare Meloni1,*, Benedetto Rocchi2, Simone Severini1
	The forest-based bioeconomy in Latvia: economic and environmental importance
	Vineta Tetere1,2,*, Jack Peerlings1, Liesbeth Dries1
	The impact of COVID-19 on household income and participation in the agri-food value chain: Evidence from Ethiopia
	Margherita Squarcina1,2,*, Donato Romano1
	Bioeconomy and resilience to economic shocks: insights from the COVID-19 pandemic in 2020
	Jesús Lasarte-López1,*, Nicola Grassano2, Robert M’barek1, Tévécia Ronzon1
	Acknowledgements