The Impact of Formalization program on Informal Household Enterprises Performance: Evidence from a quasi-experiment in Nigeria

This study uses a novel, large-scale of individuals and households obtained from the Access to Financial Services in Nigeria survey (A2F) to investigate the causal eﬀects of the BIS program on the performance of informal household enterprises. For the purpose of this study, I collected data for two periods, 2018 and 2020; meaning the period when the BIS formalization program was implemented. The survey was conducted by the Enhancing Financial Innovation & Access (EFInA) Team in partnership with the Central Bank of Nigeria and the National Bureau of Statistics.


Introduction:
Over the last decade, governments all over the world have taken steps to minimize the size of the informal sector due to the numerous macro and microeconomic effects of informal businesses.For example, the proliferation of unregistered businesses can result in low domestic tax mobilization, increased inequality, low productivity, and disruption of macroeconomic policy (Berdiev & Saunoris, 2016;Capasso & Jappelli, 2013;Schneider & Enste, 2000;Tanzi, 1999).Between 1991 and 2018, the size of informal sector activity in Nigeria averaged 57.1% of GDP (Medina & Schneider, 2019), ranking Nigeria third in Africa.This number is higher than that of Mauritius (22.4% of GDP), but lower than that of Zimbabwe (22.4% of GDP) (60.6% of GDP).Informality is most prevalent in household businesses, and it is associated with poor production circumstances and, as a result, decreased productivity (NBS, 2018).
This study investigates the causal effect of the formalization program on the performance of IHEs using recently collected survey data from Nigeria, which has the third highest levels of informal economy activity in Africa (Medina & Schneider, 2019).The aim is to fill the following knowledge gaps in the extant literature by investigating the following questions: how do formalization reforms affect IHEs, and what channels do these effects occur through?To do this, this study collect data for two period; 2018 and 2020, which correspond to the time frame during which the Business Incentive Strategy (BIS) formalization program was implemented.This allows for the measurement of two primary types of outcomes: annual value added (differences in turnover and intermediate costs, e.g., materials, labor, rent, and utilities), net-profits (annual value added minus total pay bill and taxes), and a number of intermediate outcomes, e.g., access to working equipment, scale of operation, and reported issues.
This study exploits two sources of variation: first, changes in the subsidized cost for business registration, which may incentivize businesses to become formal by 2020, and second, the BIS reform and the implementation timeframe.A commonly used measure of formality in the existing literature (Fajnzylber et al., 2011;McKenzie & Sakho, 2010;Rand & Torm, 2012) is the number of household enterprises that have government registration (that is, a business license; 0 for an informal firm and 1 for a formal business).This study hypothesized that the performance of formalized businesses is likely to increase as a result of their ability to expand their operations into more competitive environment and make use of more advanced equipment.
In this study, the difference-in-difference (DID) approach is used to estimate the data, focusing on a subsample of 11,821 informal household businesses out of a total of 23,642 observations (formal and informal), and comparing them to the 4,472 treatment units (formalized enterprises) and the 7,349 control groups (still informal enterprises).The estimation approach is extended to include double DID matching and propensity score matching to account for unobserved heterogeneity in enterprise owner characteristics and potential selection bias.In all these estimation procedures, a series of tests and obtain various results are employed to validate the findings.First, the results indicate that enterprise owners who formalize their business activity increase their value added by 26% and their net profits by 27% on average.Second, this study investigates how formality influences their business operating condition, scope of operation, and competitiveness in the market.The results show that formalization improves access to electricity (39%), internet usage (46%), enterprise size (26%), money borrowed (43%), investment (33%), bookkeeping (33%), product supply (36%), access to consumers (34%), competition (28%), and decreased outside premises (62%).This study finds no evidence, however, that formalization improves access to water.Third, the potential mechanism through which these impacts occur is investigated.The results show evidence that formalization affects the performance of IHEs through mechanisms such as electricity, enterprise size, mobile phone, and investment.Although these effects are minor, they are mostly concentrated among self-employed enterprise owners.Overall, the findings suggest that formality increases value added and net profits.These findings are highly robust to various estimating methodologies and largely consistent with our hypothesis.
This study primarily contributes to the extant literature on the micro-level effects of formalization on firms and its associated benefits.A few scholars study whether and how formalization have contributed to firm performance in Asia (Fajnzylber et al., 2011;Rand & Torm, 2012;Boly, 2018;Berkel & Tarp, 2022) and Latin America (Monteiro & Assuncao, 2012;Piza, 2018;McKenzie & Sakho, 2010), all with conflicting findings and unclear how household enterprises could benefits from formalization.Moreover, Africa is generally understudied in the existing literature, and we find no micro-level studies in Nigeria yet.This study extends this literature by studying how household enterprises respond to formalization reforms by employing quasi-experimental approaches (difference-in-difference and propensity score matching).Recent articles by Boly (2018), Moyo (2022), Berkel and Tarp (2022), and Cavotta and Dalpiaz (2022) solemnly focus on firm-level formalization and applied conventional estimation strategies that mostly do not addressed causality.While these studies share an emphasis on formalization and its benefits, the analysis in this study focus on household-level enterprises which is mostly contributes to informality in Nigeria (NBS, 2018) by employing DID for estimating causal effects, rather than firm-level.Although, this analysis is close in spirit to Rand and Torm (2012) and Demenet et al. (2016) on the nexus between formalization and informality.This study highlights that the results are likely driven by the socio-cultural norms in Nigeria, which shape household enterprise owners' characteristics with respect to business operating condition and environment compared to other country contexts in the formalization literature (e.g., Asia and Latin America).Furthermore, a significant contribution of this study is the focus on how households' enterprises in Nigeria engage in self-employment, which differs significantly across different socioeconomic circumstances, given our limited knowledge of such informal activities.Another strength of this investigation is how household enterprises respond to formalization reforms in the context of Nigeria, which has a substantial informal sector associated with poor operating conditions and, ultimately, lower productivity (NBS, 2018).Further to that, this study examines the channel of transmission patterns of those benefits associated with formalization, which is one of main contribution of this paper.
The paper is progresses as follows.Section 2 discusses relevant theoretical explanations, empirical literature, and the context of Nigeria's formalization program.Section 3 presents an overview of the data and descriptive evidence.The empirical strategy is presented in Section 4. Section 5 provides and discusses the results, and Section 6 conclude with policy recommendations and potential areas for future research.

Why enterprises choose to operate in the informal sector: Theoretical underpinning
The literature provides three different theoretical perspectives to explain the phenomenon of unregistered enterprises that do not pay their fair share of taxes (La Porta & Shleifer, 2008, 2014).There are still drawbacks to formalization, despite the fact that it is generally agreed that it would lead to better economic growth, jobs, efficiency in production, conditions for workers, and social protection (ILO, 2015;Tijdens et al., 2015;Gatti et al., 2014;Fajnzylber et al., 2011).
Widely known and well-studied models include the rational exit and the exclusion models (Perry et al., 2008).When explaining the exclusive model, De Soto (1989Soto ( , 2000) ) underlines the difficulty many potential entrepreneurs have in breaking into the market due to government regulations.They believe that informal enterprises may be unable to access official finance sources, limiting their access to loans for business expansion, or legal export channels, preventing them from reaching consumers in high-demand foreign and local markets.They agree with the idea that too many rules can discourage enterprises from becoming formal, and suggest eliminating obstacles to entrance, reducing bureaucracy, and enhancing legal frameworks.Policymakers and development experts generally agree that if registration costs were reduced, a greater number of businesses would formally register, suggesting that this is an area where policymakers should focus.This perspective is in line with the BIS program introduced by the Nigerian government to reduce the costs associated with registering a business, which is the primary topic of this study.
Those that advocate the rational exit model include Levy (2008) and Maloney (2004).
According to them, businesses operates informally when the expenses of doing so are higher than the benefits they receive.Businesses deciding whether or not to register may weigh the benefits (e.g., lower risk of paying bribes to government officials, access to formal financial services, legal protections, government contracts, or highly skilled labor) against the costs of complying with many business requirements, including those pertaining to labor laws, can incur significant costs in terms of both time and money.In this model, businesses evaluate the benefits and costs of a formality decision in the same way they would any other investment decision (Rauch, 1991).The dualism is the third type of theoretical model in the literature.
According to the dualism theory proposed by (Harris & Todaro, 1970;Lewis, 1954), the two sides of the economy-the informal and the formal-are separate and essentially unrelated.
According to La Porta and Shleifer (2014), informal businesses are a direct result of low income, while formal businesses are a consequence of the market system.This is due to the fact that informal businesses are often handled by inexperienced, undereducated individuals and are hence small and unproductive.They argue that the low production of informal enterprises makes them uncompetitive in the formal economy.For this reason, the informal sector is completely separate from the formal sector.Also, they produce goods using a variety of resources (labor, capital, and technology) and target people with limited purchasing power.
These three theoretical approaches have been the subject of extensive discussion in the literature, and each can be utilized to gain insight into the factors that contribute to the prevalence of informality.Therefore, under the dual economy model, changes in registration costs might not affect the size of the informal sector, in contrast to the first two models.Meanwhile, demand-side pressures play a substantial role in the growth of the informal sector, as low-wage informal sector workers purchased low-quality, inferior products manufactured by enterprises engaged in the informal sector.By improving access to financing, governments have the opportunity to shrink the size of the unregulated economy and help existing enterprises grow.

Empirical literature
In recent years, governments have implemented a number of policies and interventions aimed at decreasing the size of the informal economy, especially in developing countries where it plays a disproportionately large role in economic activity (De Mel et al., 2013;Jessen & Kluve, 2020).Williams and Round (2007), for example, argue that formalization is the most common approach in policymaking to dealing with informality.Among the many goals of formalization programs is the facilitation of informal businesses' access to credit and the provision of business training and other services geared toward their growth and expansion (Floridi et al., 2020).Economic growth, job creation, production efficiency, working conditions, and social protection were all seen as positive consequences in the formalization literature (ILO, 2015;Tijdens et al., 2015;Gatti et al., 2014;Fajnzylber et al., 2011;Berkel & Tarp, 2022).Contrary to these findings, a number of studies have documented negative association between formalization and firm performance (Rand & Torm, 2012;De Mel et al., 2013;Boly, 2018;Rocha et al., 2018;Campos et al., 2018;Benhassine et al., 2018;McCaig & Nanowski, 2019).The decision of an informal business to register is influenced by factors such as the quality of the institutions they deal with (Loayza et al., 2005;Williams & Kosta, 2020), the costs and benefits of registering (Diaz et al., 2018;De Mel et al., 2013), a lack of managerial skills (Mukorera, 2019), patriarchal norms (Thapa Karki et al., 2021), and human capital (Do and Vu, 2021).
Many microenterprises in developing countries do not have business registration licenses to operate and do not have a tax code with the government, despite efforts by the government to encourage firm formalization (Bruhn & McKenzie, 2014;Andrade et al., 2014).As a result, studies have investigated how formalization policies and interventions influence employment creation (Betcherman et al., 2010;Grimm and Paffhausen, 2015;Betcherman, 2015), institutions (Williams & Kosta, 2020), and new business entrants (Bruhn, 2011).Few prior researches have used quasi-experimental methodologies and administrative data to examine how the formalization of businesses affects firm performance.Demenet et al. (2016), for instance, use a difference-in-difference approach to evaluate how the Vietnam formalization program affected the performance and mode of operation of informal household businesses, and they find that formalization had a significant impact on annual value added, on average, by 20%, and on net-profits, on average, by 17%.Their research demonstrates that informal household businesses that go through the formalization process have access to better equipment, expand their operations, and face increased levels of competition.
The evidence presented here, however, suggests that self-employed individuals who formally established their enterprises did not enjoy appreciably improved working conditions or performance.The SIMPLES reforms of bureaucratic simplification and tax reduction in Brazilian microenterprises were also the subject of an investigation by Monteiro and Assuncao (2012), who used a difference-in-difference approach to analyze the results.They find that formalization increased by 13% in retailers, but had no significant impact on businesses in the transportation, construction, service, or manufacturing sectors.As a result, they conclude that formalization programs have varying impacts on various sectors.This is due to the disparity between the benefits of informality and the cost of being formal.
However, Rand and Torm (2012) find no evidence that formalization led to a higher percentage of wages in total value added when they surveyed roughly 2,500 informal enterprises in 10 provinces of Vietnam's manufacturing industry between 2007 and 2009.Authors show, however, that formalization increases both gross income and investment.Boly (2018) uses data from 2005-2013 to analyze the impact of formalization on the success of informal household businesses in Vietnam and comes to the same conclusion that there is no significant impact of access to funding.However, the authors show that formalizing a business leads to growth in profit, value added, market share, and access to modern equipment.Similarly, in Malawi, the formalization effort causes a rise in interest in becoming formal, but it has no significant effect on formal markets or firm performance (Campos et al., 2018).It has been estimated that the additional taxation that these businesses will pay over the next decade in Benin is less than the expense of formalizing their operations, and only a small fraction of businesses are likely to register (Benhassine et al., 2018).
Rocha et al. ( 2018) use a difference-in-difference approach to examine the IMP tax cut program in Brazil from January 2006 to December 2012 with the aim of determining whether or not the formalization program improves the performance of newly formal businesses, and they concluded that there is no significant effect of tax cut on firm formality.They further demonstrate that the effect is not the consequence of newcomers but instead the formalization of existing informal enterprises.To back up the claim that small businesses gain little from formalization, the authors find no evidence that firms that structured as a result of the policy could see a rise in revenue in the short-run.Firms are encouraged to register not just because the associated costs have been decreased but also because of the availability of various forms of financial assistance (De Mel et al., 2013).
Piza (2018) finds a similar conclusion, showing that the Brazilian SIMPLES tax simplification program has no substantial effect on firm formalization after controlling for measurement error and seasonal shocks.Their finding suggests revealing the potential gains connected with shift from informal activity to the formal sector in legal status.De Mel et al. (2013) analyze the situation in Sri Lanka and show that even small improvements in the perceived advantages of formality can have a sizable impact on formalization rates.This lends credence to the dualists' contention that informal and formal businesses will continue to coexist for some time despite incentives to attract them.

How does the rational exit theory relate to Nigeria's informality?
Nigeria's economy is mostly agricultural, with the majority of non-farm household enterprises operating from home on a relatively small scale.Because of their small characteristics, they tend to avoid operating within the legal framework and informality can easily arise in these types of circumstances.About 96% are self-employed on their farm, whereas 84% of selfemployed who do not work on forms and are not registered with the government (World Bank, 2015).This sector, which comprises of petty traders, artisans and craftsmen, roadside mechanics, and professionals, the majority of whom are self-employed or work for small-scale family-owned businesses.The labour market structure is dominated by MSMEs, which account for 96% of all businesses and provide almost 50% of national GDP (PwC, 2020).According to the survey, 73% of MSMEs are sole proprietorships, 14% are private limited liability companies, 13% are partnerships, 5% are faith-based organizations, 1% are cooperatives, and others (1%).
Following Perry et al. (2008) and Maloney (2004)'s theoretical perspectives that businesses conduct a cost-benefit analysis and the anticipated benefits of the decision to formalize, empirical studies find mixed results of the impacts on formalization reforms on informal enterprise registration; positive and insignificant effects (Rocha et al., 2018), positive and significant effects (De Mel et al., 2013), no significant effects (Fandl & Bustamante, 2016;Goldzmidt et al., 2018).In all of these cases, the decision to formalize is influenced by the types of benefits provided by the formalization program.Additionally, the majority of previous work modeled informality in the context of developed economies or emerging markets with strong institutional, regulatory, and policy frameworks that facilitate business formalization.However, given Nigeria's unique challenges and context, as well as the country's economic structure, there is a compelling case for considering a model that works within the context of informality in Nigeria.As a result, this study contributes to the small number of studies on informality in Nigeria, the bulk of those are anecdotal and based on case reports or small non -representative surveys covering largely urban areas.
Further to that, there is still a knowledge gap and contradictory findings regarding the effects of formalization policies and interventions on informal household enterprises.This study contributes to the literature by using a novel household-level dataset, a large-scale nationally representative sample of individuals and households from the Access to Financial Services in Nigeria survey (A2F) in the periods 2018 and 2020, and employs the difference-in-difference approach to effectively pin down the causal effect of business formalization on the performance of the informal household enterprises.

Context of the formalization program in Nigeria
This study investigates the impact of business formalization program on informal household enterprise performance.Business registration fees have been reduced by half, from N10,000 (about $25) to N5,000 (about $13), as part of a government strategy to encourage the registration of informal enterprises.However, the core of the issue is whether or not the relative benefits and perceived costs of each legal status influence the size of the informal sector.The expenses of formalization are confirmed by the research of Djankov et al. (2002) for the informal businesses.As part of the implementation strategy of the BIS program, government provides an online platform to register their businesses, as well as allow a third-party registration platform (mostly registered legal firm) to register on behalf of their clients.According to the CAC report (2019), the BIS initiative has registered over 2.8 million business names since it started.Thus, this has increased flexibility and reduced the bureaucratic bottleneck associated with business name registration.The essence of the reforms is to make MSMEs business registration and regulation as simple as possible for the overall benefit of the Nigerian economy.

Data source
This study uses a novel, large-scale of individuals and households obtained from the Access to Financial Services in Nigeria survey (A2F) 1 to investigate the causal effects of the BIS program on the performance of informal household enterprises.For the purpose of this study, I collected data for two period, 2018 and 2020; meaning in the period when the BIS formalization program was implemented.This allows measuring two main types of outcomes: annual value added (differences in turnover and intermediate costs such as materials, labor, rent, and utilities), netprofits (annual value added minus total pay bill and taxes), and a number of intermediate outcomes such as access to working equipment, scale of operation, and reported issues.
There are three reasons to use the survey data.First, the survey approach allows to captures both formal and informal firms as they transition towards formal status in both rural and urban locations.The second reason is that the survey enables for the tracking of the same units over time, including both formal and still informal units.Finally, the questionnaire enables for the precise identification and measurement of the impact of business formalization on final outcome variables such as value added, net profits, and tax registration.The survey also contained information regarding demographic characteristics (age, sex, education, motivation, etc.), labor market activity (employment, earnings, profits, etc.), scale of operation (enterprise size, borrowed money, book keeping, etc.), intensity of competition (supply, customers, etc.), access to equipment, and ownership of assets and value.
A total of 23,642 observations of both formal and informal household enterprises in 2018 and 2020 were obtained after balancing the panel between years and removing non-response.The subsample of 11,821 businesses that started out as informal is the main focus of the analysis, as they are compared to the 4,472 treated respondents and the 7,349 control respondents their businesses remained informal (see, Table 1).Through the unique household identifier, respondents were tracked and matched between the two years.For those who could not be matched as a result of change of location were dropped from the whole sample to avoid bias estimates.The recent BIS formalization program influenced the selection of these time periods.
For the period of the study, it is possible to separate the two.According to the International Labor Organization's (1993Organization's ( , 2003) ) definition, which is used here, the informal sector includes not only production and employment in the unincorporated or unregistered (informal sector), but also employment without social protection through work inside and outside the informal sector (informal employment).In addition, tax registration was employed as an alternative formalization measurement (McKenzie & Sahko, 2010) to check for the robustness of the estimates.Therefore, estimating the causal effects of business formalization is important for promoting measures addressing informality.

Descriptive evidence
Testing the parallel trend assumption is necessary for DID estimation, which holds that the difference between the control and treatment groups would be consistent over time if treatment were not involved.In spite of its importance for showing baseline trends, the idea of parallel trends has come under criticism as of lately (see, Ryan et al., 2018;Jaeger et al., 2020;Roth, 2022).Although Wing et al. (2018) state that the parallel trend assumption cannot be tested in a typical two-group two-period DID estimation, but it can be validated by a statistical test of mean differences.Because of this, a balance test in covariates is performed for 2018 prior to the formalization program's implementation and to statistically test the significant difference between the treated groups and the control groups during the baseline.
Table 1 compares the mean differences between the treatment groups and the control groups prior to the start of the formalization program in 2018.Except for the motive to start a business, it appears that there are significant differences among business location, enterprise age, and sectors.Interestingly, between the treated groups, differences in business migration increased by 26%.Furthermore, there is a 25% increase in the number of enterprises operating in Northern-Nigeria compared to Southern-Nigeria.At the same time, when compared to the control groups, the mean differences in sector of operation increased only marginally for the treated household enterprises.This finding implies that formalization can create opportunities for businesses operating in the formal sector.Table A1 displays the detailed descriptive statistics for both periods.Comparing the treated groups to the control groups during the two time periods, Table 2 displays the intermediate results in terms of business operating conditions.In comparison to informal enterprises, formalized ones were more likely to have working equipment (such as water, electricity, telephone, mobile phone, and internet), expand their operations (such as enterprise size, outside premise, borrowed money, investment, and bookkeeping), and have fewer problems (e.g., supply, customers, and competitors).In 2018, for example, treatment groups have a higher mean score (31%) in electricity than control groups (8%).But in certain categories following treatment in 2020, the mean difference actually got worse, with the exception of enterprise size between the treated groups (81%) and the control groups (80%).
Other variables mean differences are virtually the same in magnitude.The treated groups generally performed better in terms of economic opportunity to business condition than the control groups.Finally, to show graphically how the treatment and control groups perform in terms of annual value added as the primary outcome variable.Figure 1-2 illustrates the distribution and evolution of formalized vs informal enterprises.The results show that the treated units outperformed the control groups, and their value added grew significantly more.The mean differences in their distribution are evident in Figure 1, despite the fact that their value added is almost same in Figure 2.This is solely due to enterprises who have chosen to work in the official sector.

Empirical strategy
To estimate the causal effects of formalization on the performance of informal household enterprises, this study uses the standard difference-in-differences (DID) method, as employed by Demenet et al. (2016).The DID approach has recently received more attention in impact evaluation as a result of changes in policy or program interventions (Callaway and Sant' Anna, 2021).The following is the form of the model specification: where   is the outcome variable (annual value added and net profits) of informal household  and at period  . is an indicator of being assigned to treatment (business getting registered, yes=1, or 0 otherwise),   is the time dummy indicating the time after the BIS program,   is the set of control variables, and the error term   .Accordingly, the variable of interest and the effect to estimate is  3 , coefficient of the interaction term between treatment and time (  *   ).This study links business owners from the two time periods to their household enterprises through the use of persistent identifiers.
The second DID specification introduces fixed effects and controls for endogeneity linked to time-invariant local and unobservable characteristics   , such as the ability of the households and degree of compliance with regulations as specify below: Fixed effects, particularly unit-level fixed effects, are used in causal inference to adjust for unmeasured time-invariant confounders (see Mummolo & Peterson, 2018;Imai & Kim, 2019).
The ultimate goal of DID approach is to get an unbiased estimate of the treatment effect, containing two time periods -"pre" and "post", and two groups, "treatment" and "control" (Goodman-Bacon, 2021).This study estimates treatment effects on two outcome variables: annual value added (differences in turnover and intermediate costs such as materials, labor, rent, and utilities), net-profits (annual value added minus total pay bill and taxes), and a number of intermediate outcomes such as access to working equipment, scale of operation, and reported issues.In line with the existing literature (see McKenzie & Sahko, 2010;Rand & Torm, 2012;Demenet et al., 2016;Do & Vu, 2021), this study limits the set of covariates to includes various household characteristics that may influence formalization: age, sex, education, migration, and motivation.The analysis also controls for scale of operation (enterprise size, borrowed money, book keeping, etc.), intensity of competition (supply, customers, etc.), access to equipment, and ownership of assets and value.Overall, the empirical strategy relies on the DID estimator to identify the impact of formalization program to the informal household enterprises performance outcomes.Estimating DID's logic using a two-group, two-time period example highlights the method's potential to avoid the endogeneity concerns that arise when comparing groups of individuals with different characteristics (Meyer, 1995).

Identification
To estimate the causal effects of formalization on the performance of informal household enterprises, this study exploits two sources of variation.First, the analysis uses the BIS reform and the implementation timeframe that allows to track enterprises across the two time periods and identify whether they became formal in 2020.Household unique identifier greatly assist to track informal enterprises that are already existing in 2018, and became formal in 2020.For example, enterprises that started out informally in 2018, but by the time the survey was conducted in 2020, they transitioned into the formal sector and received the necessary permits to legally conduct business (see Figure 3 above for reference).This is important because the results may be different for new entrant that are formalized.As study by Fajnzylber et al. (2011) showed, newly established businesses that choose to operate formally generate greater income and profits, employ more employees, and require more investment capital.Bruhn (2011) examines the effects of Mexico's simplified business entrance regulation reform and finds that formalization led to a 5% rise in the number of registered businesses.However, the author reveals that the reform did not increase the likelihood that previously unregistered business owners would register their business.This indicates that the new entrants are the only ones affected by the reform.To isolate this effect, the analysis estimates the dummy variable representing whether or not household businesses registered their businesses, controlling for the fixed effects of enterprise location and time.An indicator of formalization is whether or not a household business is formally registered with the government (yes=1, no=0).
Second, the analysis considers changes in the subsidized fee for business registration, which may encourage enterprises to register.There are three essential assumptions that must be made in order to identify the causal effect.Under the first assumption, the BIS program must initially come as a complete surprise.The prevailing point of view is that the BIS initiative in Nigeria was announced unexpectedly and execution began soon after, this assumption holds true.
According to the second assumption, the BIS program was implemented at a time when there were no other shocks that could have had a comparable effect.Finally, the control group needs to be a good control to ensure that the treated groups behaved similarly to the control groups in the absence of the program.To validate our estimates, we perform various robustness and sensitivity checks in sections 5.2-5.4.First, we use propensity score matching and diffmatching to estimate our data.Following that, we test our findings using alternative measures of formalization and the outcome variable.

Impact of formalization on annual value added: Difference-in-Difference
This section investigates the effects of formalization on the performance of individual household enterprises as measured by annual value added.The results in Table 3 were estimated using the Diff-OLS and Diff-FE approaches given in the model specifications.Columns 1-3 show the estimates for Diff-OLS, we find that value added is positive and statistically significant at the 1% level for the treated enterprises.This implies that a 1% increase in household enterprise formalization resulted in a 26% increase in value added for the treated groups.Columns 4-6 show a similar level of magnitude for the Diff-FE estimates.Even with additional controls (enterprise age, region, sector, education, sex, migration, age, and motive to begin a business) of both informal household enterprises and head characteristics in panel A and B (columns 3 & 6), the results remained unchanged.This finding suggests that being formal results in higher value added.
Additionally, the coefficients for the North dummy (columns 2-3) are positive and statistically significant at the 1% level, indicating that Northern Nigerian household enterprises are likely to have higher value added than Southern Nigerian household enterprises.This result explains why the migration coefficients (columns 3 & 6) are positive and statistically significant at 10% and 1%, respectively.One possible explanation is that the Northern dwellers are predominantly "Hausa-Fulani" ethnic groups, which have a history of migration from one area to another in pursuit of greener pastures.

Impact of formalization on annual value added: Propensity Score Matching (PSM)
PSM are useful when evaluating the effect of a treatment on an outcome utilizing observational data and when selection bias owing to nonrandom treatment assignment is likely.Since the groups (treated and control) are not randomly assigned to register their businesses, the next requirement for the use of propensity score matching is that the outcome variable is conditionally mean independent of treatment based on the propensity score-P(Z) (Smith & Todd, 2005).This study employs the PSM approach, popularized by Heckman (1979) and Rosenbaum and Rubin (1983), in an attempt to reduce the selection bias and possible confounding that could otherwise be present in the model.The potential for selection bias was tested by comparing the baseline values of the outcome variable to those of the covariates at Table 1.Therefore, the propensity matching identifies groups deserving of a treatment and groups deserving of a control that have comparable chances of being chosen for a treatment.By matching each treated unit with a non-treated unit with similar characteristics, PSM approaches create an artificial group "counter factual", whose groups are as similar as feasible independent of treatment status, allowing for more accurate comparisons of results.One use of this technique is to compare the effects of registered versus unregistered informal household enterprises.To have enough confidence in the matching results, we need to know that the matching technique done by psmatch2 is fair with respect to the propensity score.
We extend Eq. ( 1) by constructing a probit model to estimate the propensity score, i.e. the probability that a household enterprise obtains a business license, as a function of a set of enterprise owner characteristics.This model assigns treatment units based on observable covariates, which can reduce confounding and selection bias (Austin, 2011).The PSM method has the advantage of compressing a potentially large number of observable factors into a single measure that can be used for matching.In accordance with (Wooldridge, 2015), we estimate the potential endogeneity of formalization in the following equation using a Control Function approach: where 1[.] is a binary of a normal conditional distribution function;   , the dependent variable, has a value of 1 if a household enterprise is registered (i.e., formal), and a value of 0 otherwise;   are control variables similar to those in Eq. ( 1);   corresponds to the set of exogenous variables that are omitted from Eq. ( 1), and are partially correlated with formalization;   = (  ,   ); and   is an error term.Our assumption is that when these informal enterprises observe some characteristics of formal enterprises and attribute such characteristics to formalization, they are more likely to formalize.
Following that, we rely on the conditional independence assumption, which stipulates that formalization assignment and performance are independent.If the matched sample is sufficiently balanced, the average treatment effect on the treated (ATT) can be used to determine the effect of the treatment, which is defined as the difference between potential outcomes with and without treatment for household businesses in the treatment units, as follows: where  1 is the annual value added of a household enterprise that formalizes and  0 , is the annual value added of a household enterprise that does not formalize.
The counterfactual parameter ( 0 |  = 1), can be difficult to be measured, but the annual value added of a household enterprise without business formalization, ( 0 |  = 0), can be measured and used in place of the counterfactual.Additionally, to prevent possible biased estimates of the pre-treatment difference between treatment and control units, we incorporate the effect of observed predictors (X) to a single estimate of the propensity score in the following way: (5) Lastly, we use a matching algorithm (radius-matching, nearest neighbor-matching, and kernelmatching) to reduce the level of bias between the treatment and control units as follows: The next step is to test the hypothesis some individual household businesses overlap between the treatment and control groups have similar propensity scores.With this information, we can match individual household between the treatment group and control group.In the density plot, the treated households' enterprises display an inverted U-shaped distribution, while the control groups show a right-skewed distribution.There is, thus, a large disparity between the two groups before they are matched.After matching, there is substantial overlap between the two groups' propensity scores, with a region of common support ranging from 0.0949 to 0.9998 (see Table 3), conditional on a set of several covariates.As can be seen in Figure 3, the two distributions are very similar.Figure 4 displays the nearly identical distributions.Moreover, the untreated, treated (on-support), and treated (off-support) individual household enterprises' range of common support is presented in Figure 5. Evidently, there is extent of overlapping between the treatment and control groups.One way to test for balance is to ensure that the means of the propensity scores is equivalent in the treatment and control groups across the five categories (Imbens, 2004).After the matching process, six off-support units are taken out of the observational data, and this can accurate estimates are obtained (Smith & Todd, 2005).Under the assumption of "selection on observables" (see Wooldridge, 2010, pp. 920-930), and when the matching shows an overlap, estimating the average treatment effects by reweighting on propensity score may offer a consistent and reliable estimate (Busso et al., 2014).Using a distance measure between each pair of data with respect to a given set of covariates, nearest-neighbor matching (NNM) identifies the closest observations to each subject.PSM can be used as an option to NNM, and matches people according on their predicted propensity to respond favorably to a treatment, or propensity score.To overcome the issue of sample selection bias, combining PSM and DID can produce a precise causal effect (Botosaru & Gutierrez, 2018).
Table 5 below shows balancing tests demonstrating that the matched sample has reduced difference between the treatment and control groups covariates, suggesting that this method can help make treated and control household enterprises more comparable to one another.Remarkably, after the matching the sample bias of the covariates is greatly reduced, with the lowest decrease reaching 50% and the highest amounting to 99.2%.Similar to how selection bias is decreased in terms of measured and tested covariates in Figure 6 graphical output from the pstest.The analysis proceeds by eliminating all except the most common support individual household enterprises from consideration.There is strong common support in favor of adopting PSM, proving the methods assumption to be correct.

Analysis of the Average Treatment Effects on the Treated (ATT): Comparison between Difference-in-Difference and Propensity Score Matching methods
Table 6 presents three PS-matching techniques; nearest neighbor, radius, and kernel regression matching methods to confirm the robustness in the findings.This compare treated and controls groups who have the same values for a set of covariates and likely to receive a treatment.Across all matching techniques, the average treatment effects on the treated for the effect of formalization on annual value added is between 0.11 to 0.12 units.At the confidence level of 1%, the impact of formalization significantly increased enterprise performance, implying that the reform had a strong positive impact for enterprises that formalize increased higher average value added.Table 7 shows the results separated into self-employed and enterprises with more than one worker to investigate whether the outcomes are being driven by larger enterprises.
The results are nearly same, with the exception that the size of effect decreases for the Diffmatching estimations (Radius-matching, Kernel-matching, and Nearest-neighbor matching).
At the 1% level, the coefficients are highly statistically significant, suggesting that more formalized enterprises increase value added.This estimate suggests a one-unit increase in enterprise size is associated with a 0.24% (column 1-4) and a 0.16% (column 5-6) increase in value added.The PSM results are consistent with the DID estimates.Notes: Nearest neighbor refers to matching each treated unit with the control unit that is the most similar to it in terms of pscore, Radius refers to matching each treated unit with the control units within a certain range of pscore, and Kernel connotes matching each treated unit with all control units, but giving more weight to those with more similar pscore.*Six untreated units are off common support (See Figure 5), *** p<0.01, ** p<0.05, * p<0.1 Following that, we explore how formalization affects a set of business operation condition (such as access to water, power, telephone, mobile phone, and internet) on household enterprises.Table 8 columns 2-5 show the Diff-OLS, Diff-FE, and Diff-matching estimations.These estimates suggest a significant and positive increase in access to electricity, telephone, mobile phone, and internet services.For example, the Diff-FE estimates in columns 2 and 5 imply that formalizing a household enterprise improves access to electricity by 39% and internet access by 46%, respectively, even after controlling for all characteristics that may affect enterprises performance.The coefficients are statistically significant at 5% and 1%.
Similarly, the significant level for electricity increases to 1% for Diff-matching estimates (Radius-matching, Kernel-matching, and nearest-neighbor matching).However, we find no significant results for water access, with the exception of Diff-matching (only Radius-matching and Kernel-matching), which is positive and statistically significant at the 1% level.This suggests that treated units are 59% and 46% more likely to have access to water, respectively.
Next, we exploit how formalization impacts the scale of operation of household enterprises measured by enterprise size, outdoor premises, borrowing of money, investment, and bookkeeping.Nigeria has a sizable informal industry that operates without a license and usually goes unnoticed by the authorities.In other words, being informal is likely to inhibit businesses from reaching their full potential (Fajnzylber et al., 2011).Table 9 presents the results of the Diff-OLS, Diff-FE, and Diff-matching regressions.For Diff-FE estimate that includes controlling for head characteristics, the coefficients are positive for enterprise size (at the 1% level), outside premises (at the 5% level), borrowed money (at the 5% level), investment (at the 5% level), and bookkeeping (at the 1% level).For example, leaving the informal sector raised household enterprise size by 26%, outside premises (decreased by 62%), borrowed money (43%), investment (36%), and bookkeeping (33%).However, after accounting for all controls, the significant level was lowered to 5% for the entire scale of operating components.We observe similar positive and statistically significant impacts at the 1% level for the Diffmatching estimates, but the magnitude effects are smaller than for the Diff-FE estimation.This investigation suggests that formalization can help business owners employ more employees, gain better access to decent premises, obtain funding, make regular investments, and manage effective written business accounts.
Next, in Table 10, we explore the environment competitiveness of the treatment groups separately in terms of potential supply issues (e.g., raw materials), customer access (marketing of goods and services), and competition intensity (marketing of goods and services).For Diff-FE estimates, we show that treated household enterprises are not constrained; rather, they experienced a positive and statistically significant increase in supply of goods and services to around 36%, customers (34%), and competitions (38%).A similar positive impact was observed for the Diff-FE estimation (Nearest-neighbor matching), as formalization leads to an increase of around 30% in supply, customers (18%), and competitors (28%).These findings confirmed the hypothesis that formalization leads to a better working environment, which is consistent with earlier studies (Fajnzylber et al., 2011;McKenzie & Sakho, 2010).induce enterprises to pay their taxes more frequently.The coefficients for North dummy continue to exert positive and statistically significant impact of household enterprises located in the Northern-Nigeria.This is expected because the region is predominantly populated with the "Hausa-Fulani" who are historically known for migrating from one location to another.
Next, we check whether our result is influenced by the enterprise size as reported in Table 12.We find a variation in the estimated results.For example, for the all sample, we find a sizable impact of 20%, 18% (for self-employed), and 23% (1 + employee) positive increased on net profits, respectively.This suggests more net profits could encouraged employing more workers.
Overall, this result also suggests that formalization may increase wages and reduce unemployment.Placebo effect on annual value added.-To further probe robustness checks, we perform the placebo exercise using fake treatment time as presented in Table 15.Here, we reassign the treatment to periods previous to the intervention when no treatment actually occurred.Expectantly, the magnitude of effects is dramatically disappeared and statistically insignificant on the annual value added in columns 1-4.The estimated results for the Diff-matching turns negative and statistically insignificant.This suggests that the formalization reform actually incentivize household enterprises to register their businesses and leads to an increase in value added for the treated groups.
Tax registration on annual value added.-Following McKenzie and Sahko (2010), we create a dummy variable to measure household enterprises that did not file for tax registration in 2018.
Obtaining a tax registration code allows an enterprise to pay their "fair share" to the tax authority, making it a formal entity.Table 13 shows the results of the Diff-OLS, Diff-FE, and Diff-matching regressions (only radius-matching).These estimates show significantly positive effects at the 1% level (columns 1-5), 5% level (column 6), and 10% level (column 7).After controlling for all owner characteristics, formalization, on average, increases value added by around 22% for both Diff-OLS and Diff-FE estimates.This finding is consistent with McKenzie and Sahko (2010), who showed that having a tax identification number raises profit by 53% for medium-sized enterprises while decreasing profits for small and big firms.However, when employing a common value of k-4 for nearest-neighbor matching, the coefficient becomes negative and insignificant.Overall, the evidence in Table 13 suggest that formalization increases value added.
Lastly, we explore if getting a tax registration code affects the treatment groups' operational conditions.The results of the Diff-OLS, Diff-FE, and Diff-matching (radiusmatching) regressions are shown in Table 14.After controlling for owner characteristics, the results for Diff-OLS estimation show significantly positive effects for access to equipment regarding access to electricity, with an increase of about 39% (at the 5% level) and 1% level, with an increase of about 13% for Diff-matching regressions (radius-matching).Similarly, internet accessibility increased to around 22% (at the 5% level) and 21% (at the 1% level) for diff-matching regressions (radius-matching).In respect of enterprise size, we find that owning a tax registration is strongly associated with a 22% increase (at the 5% level) for Diff-OLS estimates and a 21% increase (at the 1% level) for Diff-FE estimates.Similarly, for outdoor premises, Diff-OLS estimates reduced by 58% (at the 5% level) and Diff-matching regressions decreased by 36% (at the 1% level) (radius-matching).Furthermore, the ability of the treatment groups to get funding rose by 46% (at the 1% level) for Diff-OLS estimates and 50% (at the 5% level) for Diff-FE estimates.Furthermore, competitors grew by around 31% (at the 5% level) for Diff-FE estimates and 15% (at the 1% level) for Diff-matching regressions (radiusmatching).All of the owners' characteristics were controlled for in all of these estimates.When we compared tax registration to business registration as a measurement of formalization, we find that policy measures such as incentivizing registration may work better for business registration.

Potential Channel-Effects Analysis
This section of the study investigates the potential mechanisms through which formalization might influence enterprise performance.As Demenet et al. (2016) noted, selfemployed businesses gain less from registration than larger businesses do.One way to properly test which channel the causal effect play-out we think is to divide our sample into privatelyowned enterprises and enterprises employing two or more workers.Table 16 shows the results of the Diff-OLS and Diff-FE regressions.We find that electricity has a significantly positive effect on value added at the 5% level, with an increase of around 5% for self-employed treatment units using the Diff-FE estimate.However, for the entire sample, the significant level decreased to around 10%. (with 2 percent increased).Furthermore, the effect of mobile phones on value added is positive, at around 0.031 lower than the effect of electricity of 0.051.Similarly, enterprise size increased to about 0.012 (at the 10% significant level), and investment increased by about 0.061.(at the 5 percent significant level).When it comes to business owners with more than one employee, only enterprise size has a marginal effect of 0.010.(at the 10 percent significant level).This is the most interesting aspect of our study.Overall, the evidence in Table 16 suggests that electricity, mobile phone, enterprise size, and regular investment are mechanisms through which formalization influences the performance of household enterprises.These impacts are significant among self-employed treatment units in our sample.

Potential threat to causality
A performance of the business and its registration status (whether they register or not) could be influenced by unobserved time-varying heterogeneous factors.The choice to register a business or place restrictions on its activities in a given area may be influenced by factors that cannot be directly witnessed, such as changes in local policy or the state of the economy.So far as the author is aware, just one national agency (the Corporations Affairs Commission) is responsible for registering businesses in Nigeria.However, a company only needs to register for taxes in the state where it will be conducting business.Second, there's the problem of reverse causality.If improved results motivate informal businesses to register their operations, then the effect we observed would be nothing more than a correlation between growing one's business and becoming official, rather than an indication that these enterprises voluntarily transitioned out of the informal sector.This issue was resolved by leveraging the approach employed by Demenet et al. (2016) to construct a dummy variable to indicate the fraction of informal household enterprises whose value added increased in real terms in light of the overall economic situation.This only applied to 62% of the sample in the data.Table 17 shows the result of the Diff-OLS, Diff-FE, and Diff-matching (nearest-neighbor and radius-matching) regressions.The coefficients for both estimates are comparable to previous results, except that the magnitude impacts on the probability to formalize has increased on average to around 71% for the Diff-OLS estimate and 53% for the Diff-FE estimate.All of these findings show that being formal can leads to higher value added.

Conclusion
This study investigates the causal effect of formalization programs on the performance of IHEs.
It uses novel data to address two policy-relevant concerns about informal sector activities: how do formalization reforms affect IHEs, and through which channels may those benefits occur?This study uses the difference-in-difference approach to identify two sources of variation.To account for unobserved heterogeneity in enterprise owner characteristics and potential selection bias, double DID matching-approach and propensity score matching is employed to estimate the data.In this analysis, a series of tests and obtain several results are validated.
First, the results indicate that enterprise owners who formalize their business activity have a 26% increase in value added and a 27% increase in net profits on average.Second, this study examines how being formal affects their business operating condition, scope of operation, and competitiveness in the environment.The findings show that formalization improves access to electricity (39%), internet usage (46%), enterprise size (26%), money borrowed (43%), investment (33%), bookkeeping (33%), supply of a product (36%), access to consumers (34%), competitors (28%), and outside premises (62%).However, there is no evidence that formalization improves access to water.Third, this study investigates the possible channel through which these effects occur.The results show evidence in access to electricity, enterprise size, mobile phone, and investment as a mechanism through which formalization effects the performance of IHEs.Although these effects are small, they are concentrated primarily among self-employed enterprise owners.Overall, the findings indicate that being formal leads to increased value added and net profits.These results are highly robust to different estimating strategies and largely consistent with our hypothesis.
Consequence, this paper highlights the necessity of a policy mix that lowers the cost of registration, properly identifies possible benefits of business formalization, and increases incentives for those who remained informal.This is because the exclusion and rational exit viewpoints suggest that registration costs and benefits are important considerations for informal enterprises, particularly medium-sized enterprises.Furthermore, such policy reforms could give a way of addressing the business conditions and environment in which those enterprises operate.One method for lowering the size of the informal sector is to encourage unofficial enterprises to register.It is crucial to find the right policy combination to facilitate the transition from the informal to the formal sector, considering the complexity of the aspects that contribute to and sustain informality.Understanding these complex characteristics of informality and tracking its formalization require up-to-date and reliable data.Future research into the responses of household enterprise owners to changes in formalization reforms and the key drivers of these responses is necessary for the design of such relevant policy measures.

Figure 1 :
Figure 1: Distribution per year for all informal household enterprises

Figure 3 :
Figure 3: Quasi-Experimental Design Notes: The figure depicts the quasi-experimental design with a sample of individual household enterprises at each stage of the observational process.

Figure 4 :
Figure 4: Distribution of Propensity Scores at baseline before and after Kernel Matching, showing common support

Figure 5 :
Figure 5: Distribution of Propensity Score across Treatment and Control Groups

Figure 6 :
Figure 6: Graphical output of selection bias In January 1st to March 31st 2019, and then extended from May 13th to August 13th 2019, the Government of Nigeria launched the Business Incentive Strategy (BIS) program to enable informal enterprises to: (i) register their operations, (ii) open bank accounts, (iii) gain access to financing, and (iv) benefit from other government interventions such as grants.The program was a nation-wide initiative and one of the first direct attempts carried out by the Corporate Affairs Commission (CAC) saddled with this responsibility to formalize business activities.This is because informal business activities in Nigeria is generally home-based, with no fixed premises or operating outdoors.

Table 3 : Effects of formalization on annual value added Annual value added (log)
Notes:The dependent variable is annual value added, which is in log, and Post is a dummy indicating year after the formalization program, Treated is a dummy indicating individual household enterprises that formalized their businesses, and formalization (Treated*Post ) is the variable of interest that shows the causal effect.Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table 5 : Matching test of the PSM Model: Results of balancing test
Note: U (unmatched) and M (matched)

Table 11 : Effects of formalization on net profits Annual Net Profits (log)
The dependent variable is annual net profits, which is in log, and Post is a dummy indicating year after the formalization program, Treated is a dummy indicating individual household enterprises that formalized their businesses, and formalization (Treated*Post) is the variable of interest that shows the causal effect.Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Notes:

Table 13 : Effects of tax registration on annual value added
Notes:The dependent variable is annual value added, which is in log, and tax registration is a dummy indicating whether informal household enterprises registered with the tax authority, and formalization (Treated*Post) is the variable of interest that shows the causal effect.Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1