Policy Research Working Paper 10183 Missing Information Why Don’t More Firms Seek Out Business Advice? Miriam Bruhn Caio Piza Development Research Group & Development Impact Evaluation Group September 2022 Policy Research Working Paper 10183 Abstract This paper tests whether providing more information on sheet increased demand for business advice by 7 percentage business practices can lead firms to seek out advice and points, relative to 21 percent in the control group in the improve their practices. The authors collaborated with first six months, suggesting that information matters for a business advice provider in Brazil to implement a ran- seeking out advice. However, the control group catches domized experiment with 866 small firms. The treatment up over the next 12 months. The intervention did not groups received different versions of an information sheet affect business practices and performance outcomes, but that benchmarked business practices to other firms and it decreased the fraction of firms that report being happy listed five practices to improve. Receiving any information with their performance. This paper is a product of the Development Research Group and the Development Impact Evaluation Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http:// www.worldbank.org/prwp. The authors may be contacted at mbruhn@worldbank.org and caiopiza@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Missing Information: Why Don’t More Firms Seek Out Business Advice?* Miriam Bruhn – Development Research Group, World Bank, and IZA Caio Piza – Development Impact Evaluation Department, World Bank Keywords: Business practices; Benchmarking; Business training; Firm growth. JEL: O12, L20, M10. ∗ We thank the Knowledge for Change Program (KCP), DIME i2i, and the Brazil Country Office for supporting this project with funding. We thank Fabio Ono, Agnaldo Castanharo and Fernanda Robes from SEBRAE-PR for their collaboration and support. Beatrice Aline Zimmermann, Beatriz Machado Ribeiro, Caroline Nogueira, and Rafael Santos Dantas provided excellent field work and research assistance. We are grateful to David McKenzie, John Loeser, and various seminar participants for their valuable comments. 1. Introduction Seeking out business advice, for example through classroom training or consulting, can be a profitable investment. Business practices vary widely across firms (Bloom and Van Reenen 2007, 2010; Bloom, Sadun, and Van Reenen 2012; McKenzie and Woodruff 2016) and firms with better business practices have higher productivity, sales, and profits. Mounting evidence shows that training improves both business practices and firm performance. Random assignment to classroom business training increases profits and sales by 5 to 10 percent (McKenzie 2021). Offering more individually tailored and intense consulting services can have even larger effects on firm performance (Anderson and McKenzie 2020; Bianchi and Giorcelli 2020; Bloom et al. 2013; Bruhn, Karlan, and Schoar 2018; Iacovone, Maloney, and McKenzie 2022). Cost-effectiveness calculations suggest that firms can recover the cost of training in one to three years. Yet, many firms do not take up business training, even when it is heavily subsidized. The average take-up rate of free business training across fifteen studies was 65 percent (McKenzie and Woodruff 2014). Among firms that expressed interest in subsidized consulting, 53 percent participated when offered the service. Only 10 percent of those who were not offered the service hired other consultants (Bruhn, Karlan, and Schoar 2018). Empirical studies have identified lack of competition (Bloom et al. 2015) and family ownership (Lemos and Scur 2018; Bandiera et al. 2018) as reasons why firms do not improve their business practices. Gibbons and Henderson (2012) argue that lack of information is another barrier to adopting better business practices. Anecdotal evidence from studies on business training also points to lack of information. Bruhn, Karlan, and Schoar (2018) asked firms in Mexico why they did not hire a management consultant. Two major reasons were uncertainty about the benefits and simply not having thought about it. Anderson and McKenzie (2022) find that many small businesses in Nigeria do not know of the business service providers available in the market and what their services entail. Bloom et al. (2013) document that firms in India (i) did not believe the most common management practices could impact performance and (ii) were reportedly unaware of the least common practices. We implemented a randomized experiment to test empirically whether providing more information on business practices can lead firms to seek out advice and improve their performance. We collaborated with the Parana office of the Brazilian Micro and Small Business Support Service (SEBRAE-PR), a non-profit private organization that provides a range of both free and subsidized business services to small firms, including classroom training and one-on-one consultations. Our study sample consists of 866 small firms that participated in SEBRAE-PR’s Business-to-Business (NaN, short for Negocio-a-Negocio) program in 2018. During this program, a SEBRAE-PR agent applies a 2 questionnaire on 29 business practices and later delivers a dense 6-page report with recommendations for changes. We randomly assigned firms to a control group that received the 6-page report and to four treatment groups. The treatment groups received the 6-page report plus one of four versions of an information sheet that provides information about business practices in a more succinct and compelling format. Adding to the content of the 6-page report, the basic version of the information sheet benchmarks business practices relative to other firms in the same sector. Further, it lists five simple actions the firm can take to adopt good business practices and invites the firm to register for a free training session. The other three versions of the information sheet additionally include: (i) a motivating sentence, stating that firms who adopt fewer business practices are more likely to close based on SEBRAE data, (ii) pre-enrollment in the free training session, or (iii) both features (i) and (ii). To measure the effect of receiving the information sheet on take-up of business services, we used administrative data from SEBRAE-PR on the services they provided to firms in our sample during the 18 months following the intervention. About 28 percent of firms who received an information sheet used any SEBRAE service during the first few months after the intervention, compared to 21 percent of firms in the control group. However, this statistically significant difference disappears over time, as the control group caught up to the treatment group: 18 months after the intervention about 49 percent of firms in both the treatment and control group had used a SEBRAE-PR service. The short-run effect on take-up of SEBRAE-PR services is largest in the treatment group that received the most intense information treatment (including both the motivating sentence and pre-enrollment in a free training session). The effect on take-up is mainly driven by relatively light touch services, meaning that firms either went to a SEBRAE-PR service center and received materials on best business practices, or accessed material through SEBRAE’s website. Receiving the information sheet increased participation in group training sessions or one-on-one consulting only for firms who implemented below the median number of business practices at baseline, possibly because these firms realized they needed more hands-on help with improving their business practices. Given the relatively small absolute effect on take-up in the short-run (7 percentage points) and the lack of a medium-run effect, we do not expect the information sheet to affect business performance through take-up of SEBRAE services. However, we implemented a follow-up survey about one year after the intervention to check whether firms obtained business training from other sources or whether they improved business practices on their own. We do not find any effects on either the use of services from other sources or on a business practices index. Accordingly, the follow-up survey data shows no 3 effect of the information sheet on business sales or profits. We also accessed administrative follow-up data from the Brazilian Ministry of Labor and Employment and the Federal Tax Authority to investigate impacts on number of formal employees and business exit. We find no effects on these two outcomes either. Finally, the follow-up survey asked firms who did not use SEBRAE services after the intervention for their reasons. Most firms in both the treatment and control group reported lack of time as the main reason for not using these services, but treatment firms were statistically significantly less likely to report that they were happy with their business’ performance (16 percent in the treatment group vs. 29 percent in the control group). Our paper is related to recent randomized experiments that examine whether simply telling firms about the existence of best business practices (Dalton et al 2019) or benchmarking their performance helps them to improve by themselves (Grohmann, Menkoff, and Seitz 2022; Hou and Png 2022; Seither 2021). Although the overall evidence in these papers is mixed, the findings suggest that providing information can help improve firm performance at least in some settings, or when it is offered together with training. However, these papers don’t help explain why firms do not take-up more business assistance that already exists in the market. We test whether providing information on firms’ adoption of business practices increases their use of business services in the market, an intervention that is more scalable than also implementing training. In a related study, Anderson and McKenzie (2021) gave small firms in Nigeria information on the market for professional business service providers and provider’s quality, without including information on firms’ business practices or past performance. Distributing this information on providers had no significant impact on the use of business services. Our paper contributes to the literature by showing that lack of information on business practices is a reason why not more firms take up business services that exist in the market. When firms received information on how their business practices compared to other firms and on the benefits of improving practices, they were more likely to seek out business services that can help them improve their practices. However, after a few months, the control group started to use more business services, catching up to the treatment group. The information on business practices may thus have caused firms to speed up actions that they were going to take anyway. Eighteen months after the intervention, 51 percent of firms still did not use business services. Receiving the information made firms who did not use business services less happy with the business practices they were using, but firms claimed that they lack the time to use business services. Our findings thus highlight that time constraints might be 4 a critical barrier behind firms’ decision to not seek out business advice, a mechanism that has received little attention in the literature so far. This paper proceeds as follows. Section 2 describes the information intervention. Section 3 presents the impact evaluation design. Section 4 discusses the data and timeline. Section 5 examines the effects of receiving an information sheet on usage of SEBRAE services, business practices and performance. Section 6 concludes. 2. The information intervention SEBRAE is a non-profit that was founded in 1972 to foster entrepreneurship and promote micro and small business development. SEBRAE has regional offices in all 27 Brazilian states and over 700 service centers throughout the country. SEBRAE provides business advice through face-to- face or online consultations, group training courses, and one-to-one mentoring. Most of these services are free of charge. Each regional office runs its own programs. We collaborated with SEBRAE’s Parana office. Our information intervention augments SEBRAE-PR’s NaN program for small firms, defined to have up to R$360,000 (US$67,000) in annual sales. NaN is specific to SEBRAE-PR, although other regional offices, for example the Rio de Janeiro office, run similar programs. NaN consists of two firm visits. During the first visit, a SEBRAE consultant approaches firms asking them to participate in the program. The consultant then applies a questionnaire on 29 business practices in five areas: finance, marketing, strategic planning, organization, and human resources (HR) management (see Appendix 1). Many questions asked during the first NaN visit are like the ones that have been used to measure business practices in small firms in the academic literature, for example in McKenzie and Woodruff (2016). One to two weeks later, the consultant returns for a second visit and delivers a six-page feedback report summarizing which percentage of practices in each area the firm is currently using. For practices that are not currently used, the report suggests next steps and SEBRAE services that can help the firm implement these practices. The report is quite dense and technical (Appendix 2 contains a sample report). Data from previous rounds of NaN suggests that receiving the six-page report does not lead to much of a change in business practices. Firms that participated in NaN in both 2016 and 2017 adopted less than one additional business practice over time: the average number of business practices adopted was 12.0 in 2016 and 12.7 in 2017 (out of 29 business practices). This change is low compared to other programs. McKenzie (2021) reviews the literature on business training and concludes that, for every 20 practices that business training attempts to teach, the average firm invited to training implements 5 one additional practice. Furthermore, data from our control group show that only about 36 percent of firms that receive the six-page report get back in touch with SEBRAE during the following year to obtain more business advice or participate in training or mentoring. We tested whether providing information about business practices during the second NaN visit in a more succinct and compelling format can increase take-up of SEBRAE services. If attention is limited, as may well be the case among small firms in developing countries, providing simple information may be more effective than providing full information (Kremer, Rao and Schilbach 2019). Together with SEBRAE-PR, we designed an information sheet that was added to the six-page report in 2018. Appendix 3 contains a sample information sheet. We used four different versions of the information sheet. The basic version adds several pieces of concise information to the six-page report. It uses a scale of business practice adoption (labeled “scale of excellence”) to visualize how many out of the 29 business practices a firm is currently using. The information sheet also benchmarks business practices relative to other firms in the same sector, stating which percentage of firms adopt more practices than this firm and which percentage adopt the same number of practices. Further, it lists five simple actions the firm can take to start adopting more business practices. At the bottom of the page, the firm is invited to a free training session. The page gives the title, time, date, and location of the event and asks the firm to enroll by calling a phone number. The second version, pre-enrollment, includes the same information as the basic version, but the invitation to training lets the firm know that it was already enrolled in the session. The intention here was to overcome procrastination that firms may face when enrolling in a SEBRAE service. The third version, motivation, includes the same information as the basic version, but at the top of the page provides an additional sentence asking: “Did you know that being one level below your competitors (adopting between 1 and 5 fewer business practices) makes your firm on average 26 percent more likely to die within one year? 1” This statement was meant to provide tangible information on the benefits of adopting more business practices. It may also make the threat of competition more salient, thereby increasing firms’ incentive for adopting new business practices (Bloom et al. 2015). The fourth version, motivation plus pre-enrollment, includes the information in the basic version plus the motivating sentence and pre-enrollment in the training session. 1 This correlation is based on SEBRAE data. 6 3. Impact evaluation design Our starting sample consists of all firms that participated in NaN in 2018, in three out of six SEBRAE-PR offices. We limited the number of offices participating in the intervention to simplify the logistics. From the starting sample, we dropped firms that (i) had participated in a previous round of NaN (before 2018), (ii) adopted all 29 business practices, as measured by the questionnaire applied during the first NaN visit, or (iii) whose firm identification number (CNPJ) was not found in the 2015 RAIS database, which we use to measure employment, as described below. RAIS data is available with a time lag, and 2015 was the latest year we were able to access at the time of randomization. This selection criterion thus restricts our sample to firms that were at least three years old. The final sample includes 866 firms. We randomly assigned these firms to a control and four treatment groups. The control group received the standard six-page report described in section 2. The treatment groups received the standard six-page report, plus one of the four versions of the information sheet. Appendix figure A1 provides an overview of the five experimental groups. Randomization was done by computer, in weekly batches, to enable returning the report one to two weeks after the first NaN visit. Randomization was also stratified by SEBRAE-PR office and by whether the firm had previously used any SEBRAE service. 4. Data, timeline, and baseline summary statistics Our data comes from four different sources. Appendix figure A2 shows the timeline of the intervention and data sources. First, we have administrative data from SEBRAE-PR on use of SEBRAE services between January 2017 and March 2020, corresponding to about 12 months before and 18 months after the intervention. As outcome variables, we code dummies indicating whether a firm used a SEBRAE service within three, six, twelve, and eighteen months relative to month zero, which we define to be the month when the firm received the six-page report (and information sheet, for treatment firms). SEBRAE-PR offers many different services, which we classify into two broad groups: material and training, constructing separate dummy variables for each group. Material means a firm went to a SEBRAE-PR service center with a specific question and received written materials or media (DVDs, CDs) with content related to the question. Is also covers visits to the SEBRAE website where the firm logged into their SEBRAE account and accessed online materials. Training consists of training sessions for groups of firms, one-on-one consulting, and networking opportunities, such as trade fairs and match-making events for buyers and suppliers. 7 The baseline summary statistics in table 1 show that about 20 percent firms had used any SEBRAE service during the year before our intervention took place. More firms received material from SEBRAE (about 16 percent) than participated in training (about 9 percent). Also, from SEBRAE-PR, we use data from the questionnaire on 29 business practices applied during the first NaN visit (between April and September 2018) to construct our baseline measure of business practices: an index that corresponds to the percentage of business practices a firm uses. At baseline, the average firm used about half of the 29 business practices (table 1). Second, we obtained access to RAIS, a dataset from the Brazilian Ministry of Economy (Ministério da Economiado Trabalho) for one baseline and one follow-up year: 2015 and 2019. All formally registered firms in Brazil are required to report information about their employees to RAIS annually. RAIS data includes the number of formal employees, their characteristics, and wages, as well as firms’ opening date and sector. In 2015, which was the latest year of RAIS available at the start of our study, the average firm in our sample had about four employees, earning 1.6 times the minimum wage (table 1). The average employee was about 34 years old in 2015. About 45 percent of employees were male and 7 percent had a college degree or higher. The average firm was close to 13 years old in 2018, calculated based on the registration date reported in 2015. About half of the firms operated in the commerce sector, selling a wide range of products, with the most common ones being clothing and auto parts. Another 39 percent of firms performed services, most commonly restaurants, followed by accountants and medical practices. The remaining 11 percent of firms were mostly in manufacturing and construction. Although reporting to RAIS is mandatory, not all firms report to RAIS consistently year after year. Indeed, we only find 782 of the 866 study firms in the RAIS data at follow-up, in 2019. Some firms may have closed, while others may simply not have reported to RAIS that year. Third, to verify which firms closed after the intervention, we consulted the registry of firms maintained by the Federal Tax Authority (RFB). This data shows that 32 firms had officially closed by the end of 2019. From the RFB data, we also obtained firms’ starting capital as an additional background characteristic reported in table 1 (data on sales is not available for confidentiality reasons). The average value of starting capital was about USD 10,000. 2 Finally, we conducted a follow-up survey about one year after the intervention, between July and September 2019. This survey collected information on business performance (sales and profits), as well as business practices. The original plan was to revisit study firms during the 2019 round of NaN to reapply the questionnaire on 29 business practices. However, NaN was discontinued after 2018 for budget reasons. We applied a different questionnaire on business practices, based on McKenzie 2 Using an exchange rate of 0.19 BRL per USD. 8 and Woodruff (2016), during the follow-up survey, for a more direct comparison to the literature. The follow-up survey contained 22 questions on business practices, 16 of which correspond closely to the NaN questionnaire. We use these 16 questions to construct our follow-up measure of business practices (the stars in appendix 1 indicate the corresponding 16 NaN questions). A total of 683 firms participated in the follow-up survey, giving a response rate of 79 percent, with no statistically significant difference across experimental groups. 5. Results To measure the effects of receiving an information sheet, we estimate an ANCOVA specification controlling for randomization strata s: , = + , + ,0 + ∑ ′ ′ =1 , 1( = ) + , (1) where denotes firms and denotes the follow-up period ( = 0 is the baseline period). We first pool all treatment groups, so that the indicator variable is equal to one if the firm was randomly assigned to the treatment group and equal to zero otherwise. When the baseline value of the outcome variable is not available, we estimate a simple difference in means at time t, controlling for strata fixed effects. Next, we look at separate effects for each treatment group: , = + ∑4 ′ =1 ,, + ,0 + ∑′ =1 , 1( = ) + , (2) where denotes the four different treatment groups. The indicator variable is equal to one if the firm is treatment group , and equal to zero otherwise. Finally, we examine heterogeneous treatment effects to test whether the effects of the information are larger for firms with below median percentage of business practices adopted at baseline: , = + , ∗ ,0 + , ∗ ℎ,0 + ,0 + ,0 + ∑ ′ ′ =1 , 1( = ) + , (3) where ,0 is an indicator variable equal to one if the firm used below the median percentage of business practices at baseline (50 percent of practices) and equal to zero otherwise. Similarly, ℎ,0 is an indicator variable equal to one if the firm used above the median percentage of business practices at baseline and equal to zero otherwise. We did not stratify on baseline business practices in the randomization. However, table A1 in the appendix shows that most baseline characteristics of the treatment and control groups are not statistically significantly different in the subsamples of firms with below vs. above median business practices at baseline. We estimate many different treatment effects given our four treatment groups, analysis of heterogenous treatment effects, and outcomes that measure usage of SEBRAE services at different 9 points in time. To deal with multiple hypothesis testing, we calculate sharpened q-values that hold constant the false discovery rate (Anderson 2008). In each table, we report which coefficients have sharpened q-values at 1-, 5-, and 10-percent statistical significance levels. 5.1 Effects on demand for business advice Panel A of table 2 shows the results from estimating equation 1 with use of any SERBRAE service as the outcome variable. During the three months following the intervention, firms that received any information sheet were 7.7 percentage points more likely to seek out advice from SEBRAE than firms in the control group (column 1 in table 2). This increase corresponds to 46 percent of the fraction of control firms that used a SEBRAE service within three months of the intervention (16.7 percent). The effect of receiving an information sheet drops slightly, to 7.3 percentage points, when we consider the longer period of six months after the intervention (column 2). The effect then dissipates 12 to 18 months after the intervention as more firms in the control group seek out advice from SEBRAE (columns 3 and 4), with 49.4 percent of control group firms having used a SEBRAE service by 18 months after the intervention. Figure 1 shows this pattern more clearly. Treatment firms are more likely to use a SEBRAE service for about nine months after the intervention, but then the control group catches up. We find no effect of the information sheet on usage of SEBRAE services a year or more after the intervention. This pattern suggests that the information intervention caused firms that were going to use a SEBRAE service anyway to use this service a few months earlier, perhaps by increasing the salience of business practices, without leading to an increase in usage of services in the longer run. Panel B of table 2 reports separate effects for each treatment group, i.e., the results of estimating equation 2. The coefficients in column 1 show that the most intense treatment, where the information sheet included both a motivating sentence about the benefits of good business practices and pre- enrollment to a SEBRAE training session, had the largest effect on use of SEBRAE services. In the three months after receiving the most intense information treatment, firms were 14.7 percent more likely to use a SEBRAE service than control firms, corresponding to nearly a doubling in firms using SEBRAE services (from 16.7 percent of control firms to 31.4 percent of firms that received the most intense information treatment). However, even the effect of the most intense information treatment wanes over time (columns 3 and 4 in panel B of table 2). Finally, in panel C of table 2, we examine whether the effect of the information sheet varied with baseline adoption of business practices, by estimating equation 3. We find weak evidence that the impact of the information sheet on using any type of SEBRAE service was greater for firms with below 10 median business practices at baseline than for firms with above median business practices. These results suggest that the information sheet was particularly useful for firms lagging in terms of managerial practices. In table A2, we examine whether the information intervention affected the number of SEBRAE services used. That is, here we use the number of services instead of dummy variables for using any SEBRAE services as the outcome variables. Panel C of table A2 shows that the intervention increased the number of SEBRAE services used for firms with below median business practices at baseline, but this effect is not statistically significant after accounting for multiple hypothesis testing. Appendix tables A3 and A4 look at two different types of SEBRAE services: material and training. The control group averages in panel A of appendix tables A3 and A4 show that firms were more likely to access SEBRAE material (38 percent within 18 months of the intervention) than to receive SEBRAE training (19 percent within 18 months). More firms thus sought out relatively light touch services, meaning they either went to a SEBRAE-PR service center and received materials on best business practices, or accessed material through SEBRAE’s website. The effect of the information sheet is also mostly concentrated on these types of services. In the full sample, we find a statistically significant effect of the information sheet only on accessing material from SEBRAE, and not on receiving SEBRAE training (panels A and B of appendix tables A3 and A4). A possible reason why the intervention did not increase the participation in SEBRAE training for the average firm is that these more intense services require a larger time commitment. In contrast, the results in panel C of appendix table A4 show that the information sheet increased the probability of participating in SEBRAE training for firms with below median business practices at baseline. Perhaps these firms realized that they need more hands-on help with improving their business practices and were thus willing to dedicate more time to using SEBRAE services. Overall, we find a positive, but temporary effect of receiving an information sheet on the probability of using a SEBRAE service. The size of the effect (7.7 percentage point during the first 3 months after the intervention) is large compared to the fraction of firms using a SEBRAE service in the control group (16.7 percent) but represents only a small fraction of the treatment group. In fact, less than 30 percent of treatment firms use a SEBRAE service within the first six months after the intervention. Given this relatively low take-up rate and the fact that usage of SEBRAE services in the control group catches up to the treatment group after a few months, we do not expect the intervention to affect business practices or performance. Nevertheless, we measure these effects in the following subsection. 11 5.2 Effects on business practices and performance While the effect of the information sheet on use of SEBRAE services was small in absolute magnitude and temporary, firms may have relied on other mechanisms to improve their business practices. First, firms may have sought out business development services from providers other than SEBRAE. Panel A of table 3 show that, during the follow-up survey, 15.4 percent of control firms reported using services from other providers. This number was not statistically significantly different in the treatment group, indicating that the intervention did not cause firms to seek out business development services from other sources. Second, firms may have improved their business practices without help from SEBRAE or other service providers. Some of the actions listed on the information sheet were concrete and can in principle be implemented without help, such as “keep a register of suppliers” or “conduct a customer satisfaction survey.” However, column 2 in table 3 shows no effect of receiving an information sheet on business practices. Given that the information sheet had no effect on business practices, we do not expect it to affect business performance. Indeed, columns 3 through 6 in table 3 show no robust effect of the intervention on business performance, measured using follow-up survey variables (sales and profits), as well as administrative data on the number of employees and business closures. 5.3 Reasons for not using SEBRAE services During the follow-up survey, we asked firms that had not used a SEBRAE service during the months between the intervention and follow-up survey why they did not use these services. The most given reason was lack of time, with about 60 percent of firms giving this reason (table 4). The second most given reason was that the firm was happy with their business performance: 28.6 percent of control firms gave this reason, dropping to only 15.8 percent of treatment firms. Our interpretation of this effect is that the information sheet made firms realize that they were not using best business practices. At the same time, firms report being time constrained, which also aligns with the finding from section 5.1 that even the firms that sought out SEBRAE services predominantly accessed light touch services. Since table 4 only includes firms that did not use SEBRAE services, the finding that treatment firms are less likely to be happy with their business performance than control firms may be driven by differential selection in the treatment and control groups. However, the results in table A5 suggest that the effect on being happy with the firm’s performance is not only due to selection. Table A5 shows the answers to the related question: “what prevents you from adopting additional business practices?” that was posed to all firms that participated in the follow-up survey. As in table 4, treatment firms are 12 statistically significantly less likely than control firms to say that they are happy with their choices, although the difference here is smaller than in table 4. The strongest and longest lasting effect of the information intervention was thus to make firms aware that they were not using best business practices. However, for the firms included in table 4, this realization was not enough to seek out help from SEBRAE with improving their practices. 6. Conclusion In collaboration with SEBRAE-PR, we implemented a randomized experiment to test whether providing more information on business practices can lead small firms to seek out advice and improve their practices. Firms in the treatment group received an information sheet that benchmarked business practices to other firms and listed five practices to improve. The results show that the information led firms to demand more advice only in the short-run, without yielding medium-run effects on business practices or performance. Our findings differ from Seither (2021) who finds that benchmarking firms’ performance led low performers to make more effort and raise prices, leading to higher sales and profit. A possible explanation for the difference in findings is that benchmarking sales may provide stronger incentives for improving business performance than benchmarking business practices. Our information intervention decreased the fraction of firms that report being happy with their performance, but it only led a small fraction of firms to try to improve their business practices through advice. The most reported reason for not seeking out advice was time constraints, which is consistent with our finding that even the firms that used SEBRAE services mostly sought out light touch services. Time constraints may be particularly binding in small businesses, where the owner cannot easily delegate tasks to others. We believe that studying the role of time constraints in preventing firms from seeking out business advice and improving their practices is an area for future research. It is not clear how to alleviate time constraints, but potential solutions could include offering more flexible training schedules, and online options. 13 References Anderson, Michael. 2008. “Multiple Inference and Gender Differences in the Effects of Early Intervention: A Reevaluation of the Abecedarian, Perry Preschool, and Early Training Projects.” Journal of the American Statistical Association, 103(484): 1481–1495. Anderson, S. and McKenzie, D. 2022. “Improving Business Practices and the Boundary of the Entrepreneur: A Randomized Experiment Comparing Training, Consulting, Insourcing and Outsourcing.” Journal of Political Economy, 130(1): 157-209. Anderson, S. and McKenzie, D. 2021. “What Prevents More Small Firms from Using Professional Business Services? An Information and Quality-Rating Experiment in Nigeria.” Policy Research Working Paper No. 9614, World Bank, Washington, DC. Bandiera, O., Lemos, R., Prat, A., and Sadun, R. 2018. “Managing the Family Firm: Evidence from CEOs at Work.” Review of Financial Studies, 31(5): 1605–1653. Bianchi, N. and Giorcelli, M. 2022. “The Dynamics and Spillovers of Management Intervention: Evidence from the TWI Program.” Journal of Political Economy, 130(6). Bloom, N. and Van Reenen, J. 2007. “Measuring and Explaining Management Practices Across Firms and Countries.” Quarterly Journal of Economics, 112(4): 1351-1408. Bloom, N. and Van Reenen, J. 2010. “Why Do Management Practices Differ Across Firms and Countries?” Journal of Economic Perspectives, 24(1): 203-224. Bloom, N., Eifert, B., Mahajan, A., McKenzie, D. and Roberts, J. 2013. “Does Management Matter? Evidence from India.” The Quarterly Journal of Economics, 128(1): 1-51. Bloom, N., Propper, C., Seiler, S., and Van Reenen, J. 2015. “The Impact of Competition on Management Practices in Public Hospitals.” Review of Economic Studies, 83: 87-117. Bloom, N., Sadun, R. and Van Reenen, J. 2012. “The Organization of Firms Across Countries.” The Quarterly Journal of Economics, 127(4): 1663-1705. Bruhn, M., Karlan, D. and Schoar, A. 2018. “The Impact of Consulting Services on Small and Medium Enterprises: Evidence from a Randomized Trial in Mexico.” Journal of Political Economy, 126(2): 635-687. Dalton, P., Ruschenpohler, J. Uras, B. and Zia, B. 2019. “Learning to Grow from Peers: Experimental Evidence from Small Retailers in Indonesia.” Policy Research Working Paper No. 8933, World Bank, Washington, DC. Gibbons, R. and Henderson, R. 2012. “Relational Contracts and Organizational Capabilities.” Organization Science, 23: 1350-1364. Grohmann, A., Menkhoff, L., and Seitz, H. 2022. “The Effect of Personalized Feedback on Small Enterprises’ Finances in Uganda.” Economic Development and Cultural Change, 70(3). 14 Hou, Y., and Png, I. P. L. 2022. “Learning to Let Go: A Field Experiment on Benchmarking and Entrepreneur Exit.” Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4062428 Iacovone, L., Maloney, W. and McKenzie, D. 2022. “Improving Management with Individual and Group-Based Consulting: Results from a Randomized Experiment in Colombia.” Review of Economic Studies, 89(1): 346-371. Kremer, M., Rao, G. and F. Schilbach F. 2019. “Behavioral Development Economics,” in Handbook of Behavioral Economics: Applications and Foundations 1, Elsevier, vol. 2, 345–458. Lemos, R. and Scur, D. 2018. “All in the Family? CEO Succession and Firm Organization.” PEDL Research Papers. McKenzie, D. 2021. “Small Business Training to Improve Management Practices in Developing Countries: Reassessing the Evidence for "Training Doesn't Work." Oxford Review of Economic Policy, 37(2): 276-301. McKenzie, D. and Woodruff, C. 2014. “What Are We Learning from Business Training Evaluations Around the Developing World?” World Bank Research Observer, 29(1): 48-82. McKenzie, D. and Woodruff, C. 2016. “Business Practices in Small Firms in Developing Countries.” Management Science, 63(9): 2967-2981. Seither, J. 2021. “Keeping up with the Joneses: Economic Impacts of Overconfidence in Microentrepreneurs.” NOVAFRICA Working Paper. 15 Figure 1: Use of SEBRAE services over time Cumulative fraction with any SEBRAE service 0 .1 .2 .3 .4 .5 0 5 10 15 20 Month relative to intervention Any information sheet Control 95% confidence intervals 16 Table 1: Baseline summary statistics by experimental group No motiv, No motiv, Motiv, Motiv, Control no enroll pre-enroll no enroll pre-enroll Mean/SD Mean/SD Mean/SD Mean/SD Mean/SD (1) (2) (3) (4) (5) Panel A: SEBRAE data (2017/2018) Used SEBRAE service during the year before the intervention Any type of service 0.202 0.145 0.212 0.192 0.197 [0.403] [0.353] [0.410] [0.395] [0.399] Material 0.155 0.139 0.165 0.158 0.174 [0.363] [0.347] [0.372] [0.366] [0.380] Training 0.089 0.023*** 0.124 0.079 0.045* [0.286] [0.151] [0.330] [0.271] [0.208] Business practices used (%) 0.493 0.450 0.505 0.499 0.518 [0.255] [0.273] [0.272] [0.266] [0.259] Panel B: RAIS and RFB data Number of employees (2015) 3.869 4.098 3.412 4.124 3.798 [3.525] [5.871] [3.324] [7.177] [5.016] Average wage (minimum wages, 2015) 1.596 1.624 1.631 1.656 1.606 [0.489] [0.551] [0.560] [0.595] [0.486] Average employee age (2015) 33.967 34.206 33.484 33.97 33.682 [8.673] [8.852] [8.441] [8.708] [8.372] Share of male employees (2015) 0.446 0.458 0.486 0.477 0.497 [0.405] [0.409] [0.412] [0.401] [0.407] Share of employees with college or 0.072 0.07 0.081 0.083 0.084 higher (2015) [0.189] [0.182] [0.220] [0.208] [0.209] Firm age (2018) 12.577 13.509 12.476 12.983 11.185 [9.984] [11.078] [9.228] [8.610] [8.557] Commerce sector 0.482 0.48 0.447 0.497 0.427 [0.501] [0.501] [0.499] [0.501] [0.496] Service sector 0.387 0.416 0.412 0.339 0.416 [0.488] [0.494] [0.494] [0.475] [0.494] Starting capital (USD) 9771.951 7140.508 10526.56 9947.013 9573.12 [20061.252] [13389.208] [42617.959] [22146.332] [20069.107] N 168 173 170 177 178 Notes: This table shows baseline summary statistics using administrative data from SEBRAE-PR, for 2017 and 2018, and RAIS, for 2015, the latest available year at the start of the study. Starting capital comes from the Federal Tax Authority (RFB). Column 1 displays means and standard deviations for the control group. Columns 2 through 5 display means and standard deviations for each treatment group. Material denotes receipt of written or recorded material on business practices, either during a visit to a SEBRAE service center, or to SEBRAE's website. Training includes classes, consulting, and networking events. Business practices are only available for 846 observations, since recording errors led to missing observations for firms in one week. Stars refer to differences in means of the respective treatment group to the control group. ***, **, and * indicate significance at the 1, 5, and 10 percent level. 17 Table 2: Effect of the information intervention on use of SEBRAE services Used any SEBRAE service within 3 months 6 months 12 months 18 months (1) (2) (3) (4) Panel A: Pooled treatment effect Any information sheet 0.077**† 0.073** 0.017 -0.009 (0.031) (0.033) (0.040) (0.043) R2 0.284 0.250 0.197 0.165 Control mean 0.167 0.208 0.357 0.494 Panel B: Effects for each treatment group (1) No motiv, no pre-enroll 0.050 0.074* 0.030 -0.027 (0.039) (0.043) (0.050) (0.054) (2) No motiv, pre-enroll 0.060 0.056 -0.002 -0.023 (0.040) (0.043) (0.051) (0.055) (3) Motiv, no pre-enroll 0.056 0.049 -0.003 -0.030 (0.039) (0.042) (0.050) (0.054) (4) Motiv, pre-enroll 0.140***†† 0.114**† 0.042 0.043 (0.042) (0.044) (0.050) (0.053) R2 0.290 0.253 0.198 0.168 F-test p-value (2=1) 0.799 0.683 0.532 0.945 F-test p-value (3=1) 0.884 0.580 0.518 0.962 F-test p-value (4=1) 0.036 0.382 0.816 0.181 F-test p-value (1=2=3=4) 0.137 0.490 0.750 0.427 Panel C: Heterogenous treatment effects by baseline business practices (BPs) Any information sheet*Low BPs 0.128**†† 0.107** 0.093* 0.042 (0.040) (0.045) (0.055) (0.061) Any information sheet*High BPs 0.016 0.029 -0.074 -0.081 (0.051) (0.052) (0.062) (0.064) R2 0.291 0.256 0.205 0.176 F-test p-value (Low BPs=High BPs) 0.092 0.275 0.048 0.172 Control mean low BPs 0.093 0.151 0.267 0.419 Control mean high BPs 0.253 0.278 0.468 0.595 N 866 866 866 866 Notes: This table shows OLS estimates for the effect of receiving the information sheet on use of SEBRAE services. The outcome variables are dummy variables equal to one if the firm used any type of SEBRAE service within the stated number of months after the intervention and zero otherwise. Panel A pools all treatment groups. Panel B shows separate treatment effects for each of the four different versions of the information sheet. Panel C shows heterogenous treatment effects for firms using below and above the median percentage of baseline business practices. BP stands for business practices. Low BP refers to below the median (50%), while high BP refers to above the median. All regressions control for the fraction of firms that used a SEBRAE service within 12 months before the intervention and strata fixed effects. The regressions in Panel C additionally control for a dummy variable equal to one if the firm had low BPs at baseline. ***, **, and * indicate significance at the 1, 5, and 10 percent level. To deal with multiple hypothesis testing, †††, ††, and † indicate sharpened q-values statistically significant at the 1, 5, and 10 percent level. 18 Table 3: Effect of the information intervention on business practices and performance Service Business Number of Business other than practices IHS sales IHS profits employees closed SEBRAE used (%) (RAIS) (RFB) (1) (2) (3) (4) (5) (6) Panel A: Pooled treatment effect Any information sheet -0.035 -0.015 -0.101 -0.461 -0.466 -0.007 (0.036) (0.018) (0.154) (0.300) (0.500) (0.017) R2 0.171 0.291 0.131 0.174 0.334 0.067 Control mean 0.154 0.658 11.330 9.875 5.904 0.042 Panel B: Effects for each treatment group (1) No motiv, no pre-enroll -0.007 -0.030 -0.201 -1.075** -0.499 -0.000 (0.044) (0.024) (0.178) (0.484) (0.702) (0.022) (2) No motiv, pre-enroll -0.103** -0.009 -0.191 -0.511 -0.393 -0.018 (0.040) (0.024) (0.176) (0.412) (0.606) (0.020) (3) Motiv, no pre-enroll -0.036 -0.010 0.100 0.031 -0.409 -0.009 (0.043) (0.024) (0.176) (0.317) (0.671) (0.022) (4) Motiv, pre-enroll 0.002 -0.010 -0.119 -0.339 -0.563 0.000 (0.046) (0.022) (0.178) (0.367) (0.607) (0.022) R2 0.182 0.292 0.140 0.188 0.320 0.068 F-test p-value (2=1) 0.011 0.388 0.941 0.269 0.874 0.357 F-test p-value (3=1) 0.464 0.421 0.039 0.014 0.908 0.667 F-test p-value (4=1) 0.838 0.374 0.551 0.125 0.927 0.989 F-test p-value (1=2=3=4) 0.018 0.787 0.149 0.078 0.992 0.746 Panel C: Heterogenous treatment effects by baseline business practices (BPs) Any information sheet*Low BPs 0.010 -0.042 -0.245 -0.731* -0.148 -0.042 (0.044) (0.027) (0.302) (0.403) (0.728) (0.031) Any information sheet*High BPs -0.083 0.012 0.018 -0.229 -0.480 0.027 (0.061) (0.027) (0.147) (0.504) (0.536) (0.019) R2 0.185 0.293 0.158 0.180 0.445 0.074 F-test p-value (Low BPs=High BPs) 0.231 0.169 0.464 0.450 0.712 0.068 Control mean low BPs 0.075 0.622 11.324 10.015 5.617 0.070 Control mean high BPs 0.242 0.699 11.344 9.705 6.167 0.013 N 683 665 542 516 782 866 Notes: This table shows OLS estimates for the effect of receiving the information sheet on business practices and performance. The outcome variables in columns 1 through 4 are from the follow-up survey, conducted in 2019. The number of observations varies due to nonresponse. The outcome in column 1 is equal to one if the firm reported having received business development services from a provider other than SEBRAE and zero otherwise. The outcome in column 2 is the percentage of 16 business practices the firm reported using. IHS stands for inverse hyperbolic sine. Number of employees in column 5 is from 2019 RAIS. It is missing for firms that did not report to RAIS in 2019. The outcome in column 6 is a dummy variable equal to one if the firm was officially closed by end of 2019 according to the Federal Tax Authority (RFB) and zero otherwise. Panel A pools all treatment groups. Panel B shows separate treatment effects for each of the four different versions of the information sheet. Panel C shows heterogenous treatment effects for firms using below and above the median percentage of baseline business practices. BP stands for business practices. Low BP refers to below the median (50%), while high BP refers to above the median. All regressions control for strata fixed effects. The regressions in Panel C additionally control for a dummy variable equal to one if the firm had low BPs at baseline. Columns 2 also controls for percentage of business practices used in 2018, and column 5 for number of employees in 2015. Baseline outcome variables are not available for columns 1, 3, and 4. All firms were open in 2018, so that the baseline outcome for column 6 is zero for all firms. ***, **, and * indicate significance at the 1, 5, and 10 percent level. To deal with multiple hypothesis testing, †††, ††, and † indicate sharpened q-values statistically significant at the 1, 5, and 10 percent level. 19 Table 4: Self-reported reasons for not using a SEBRAE service Fraction of firms mentioning this reason at follow-up (multiple mention) Control Treatment Lack of time 0.571 0.624 I am happy with my business' performance 0.286 0.158*** Other 0.194 0.198 Lack of interest 0.102 0.131 High cost 0.031 0.027 Received services from other provider 0.020 0.037 N 98 298 Notes: This table shows reasons that firms reported on the follow-up survey for not using a SEBRAE service after the information intervention. These reasons were only collected for firms who had not used a SEBRAE service between the time of the intervention and the follow-up survey, according to SEBRAE-PR administrative data. Stars refer to differences in fractions of the treatment group and the control group. ***, **, and * indicate significance at the 1, 5, and 10 percent level. 20 Appendix Figure A1: Experimental design Appendix Figure A2: Intervention and data timeline 2015 2016 2017 2018 2019 2020 J F MAM J J A S O N D J F MAM J J A S O N D J F M Baseline questionnaire Information intervention SEBRAE-PR services data RAIS data RFB data Follow-up survey Notes: RAIS is a dataset on employees of registered firms from the Brazilian Ministry of Labor and Employment. We use data from the Federal Tax Authority (RFB) to measure if a firm has officially closed. 21 Appendix Table A1: Baseline summary statistics by business practices at baseline Percentage of business practices used at baseline Below median Above median Control Treatment Control Treatment Mean/SD Mean/SD Mean/SD Mean/SD (1) (2) (3) (4) Panel A: SEBRAE-PR data (2017/2018) Used SEBRAE service during the year before the intervention Any type of service 0.151 0.179 0.266 0.194 [0.360] [0.384] [0.445] [0.396] Material 0.093 0.161 0.228 0.157 [0.292] [0.368] [0.422] [0.364] Training 0.070 0.060 0.114 0.078 [0.256] [0.237] [0.320] [0.269] Business practices used (%) 0.295 0.264* 0.709 0.717 [0.151] [0.152] [0.146] [0.135] Panel B: RAIS data Number of employees (2015) 3.953 3.301 3.785 4.275 [3.850] [3.678] [3.225] [5.519] Average wage (minimum wages, 2015) 1.580 1.604 1.615 1.641 [0.465] [0.496] [0.522] [0.573] Average employee age (2015) 34.216 35.011 33.873 32.670 [8.824] [9.317] [8.677] [7.639] Share of male employees (2015) 0.453 0.574** 0.440 0.395 [0.413] [0.407] [0.406] [0.384] Share of employees with college or 0.055 0.050 0.092 0.102 higher (2015) [0.160] [0.165] [0.218] [0.226] Firm age (2018) 12.884 13.217 12.392 11.841 [9.092] [10.053] [11.046] [8.651] Commerce sector 0.500 0.485 0.456 0.449 [0.503] [0.501] [0.501] [0.498] Service sector 0.372 0.327 0.405 0.452 [0.486] [0.470] [0.494] [0.498] Starting capital (USD) 9228.393 6778.421 10374.000 9638.463 [22576.267] [9777.351] [17413.359] [19118.309] N 86 336 79 345 Notes: This table shows baseline summary statistics using administrative data from SEBRAE-PR, for 2017 and 2018, and RAIS, for 2015, the latest available year at the start of the study. Starting capital comes from the Federal Tax Authority (RFB). Columns 1 and 2 displays means and standard deviations for control firms and treatment firms using below the median percentage of business practices at baseline (50%), grouping all four different treatments together. Columns 3 and 4 display means and standard deviations for firms with percentage of business practice above the median at baseline. Material denotes receipt of written or recorded material on business practices, either during a visit to a SEBRAE service center or to SEBRAE's website. Training includes classes, consulting, and networking events. Business practices are only available for 846 observations, since recording errors led to missing observations for firms in one week. Stars refer to differences in means of the treatment group to the control group in each subsample. ***, **, and * indicate significance at the 1, 5, and 10 percent level. 22 Appendix Table A2: Effect of the information intervention on number of SEBRAE services used Number of SEBRAE services used within 3 months 6 months 12 months 18 months (1) (2) (3) (4) Panel A: Pooled treatment effect Any information sheet 0.114 0.146 0.170 0.204 (0.120) (0.162) (0.251) (0.342) R2 0.222 0.302 0.273 0.232 Control mean 0.518 0.732 1.435 2.155 Panel B: Effects for each treatment group (1) No motiv, no pre-enroll 0.047 0.106 0.259 0.305 (0.149) (0.194) (0.314) (0.478) (2) No motiv, pre-enroll 0.004 0.079 0.120 0.046 (0.157) (0.207) (0.350) (0.451) (3) Motiv, no pre-enroll 0.105 0.040 -0.185 -0.133 (0.162) (0.198) (0.300) (0.409) (4) Motiv, pre-enroll 0.299* 0.361 0.495 0.606 (0.164) (0.222) (0.349) (0.455) R2 0.226 0.305 0.277 0.235 F-test p-value (2=1) 0.785 0.887 0.697 0.607 F-test p-value (3=1) 0.723 0.723 0.152 0.356 F-test p-value (4=1) 0.133 0.233 0.505 0.556 F-test p-value (1=2=3=4) 0.327 0.469 0.208 0.367 Panel C: Heterogenous treatment effects by baseline business practices (BPs) Any information sheet*Low BPs 0.275* 0.387** 0.591* 0.800* (0.153) (0.195) (0.324) (0.419) Any information sheet*High BPs -0.101 -0.156 -0.350 -0.547 (0.197) (0.280) (0.418) (0.586) R2 0.237 0.313 0.283 0.241 F-test p-value (Low BPs=High BPs) 0.139 0.119 0.085 0.069 Control mean low BPs 0.279 0.430 0.919 1.395 Control mean high BPs 0.797 1.089 2.051 3.063 N 866 866 866 866 Notes: This table shows OLS estimates for the effect of receiving the information sheet on number of SEBRAE services used. The outcome variables are the number of SEBRAE services used within the stated number of months after the intervention. Panel A pools all treatment groups. Panel B shows separate treatment effects for each of the four different versions of the information sheet. Panel C shows heterogenous treatment effects for firms using below and above the median percentage of baseline business practices. BP stands for business practices. Low BP refers to below the median (50%), while high BP refers to above the median. All regressions control for number of SEBRAE services used within 12 months before the intervention and strata fixed effects. The regressions in Panel C additionally control for a dummy variable equal to one if the firm had low BPs at baseline. ***, **, and * indicate significance at the 1, 5, and 10 percent level. To deal with multiple hypothesis testing, †††, ††, and † indicate sharpened q-values statistically significant at the 1, 5, and 10 percent level. 23 Appendix Table A3: Effect of the information intervention on accessing SEBRAE material Accessed SEBRAE material within 3 months 6 months 12 months 18 months (1) (2) (3) (4) Panel A: Pooled treatment effect Any information sheet 0.069** 0.068** 0.022 0.012 (0.028) (0.031) (0.038) (0.041) R2 0.285 0.242 0.205 0.187 Control mean 0.131 0.167 0.304 0.381 Panel B: Effects for each treatment group (1) No motiv, no pre-enroll 0.035 0.066 0.023 -0.013 (0.035) (0.041) (0.049) (0.051) (2) No motiv, pre-enroll 0.037 0.038 -0.002 -0.019 (0.036) (0.040) (0.049) (0.051) (3) Motiv, no pre-enroll 0.057 0.042 -0.001 0.007 (0.036) (0.039) (0.048) (0.052) (4) Motiv, pre-enroll 0.147***††† 0.126***† 0.068 0.073 (0.040) (0.043) (0.049) (0.052) R2 0.296 0.247 0.208 0.191 F-test p-value (2=1) 0.962 0.493 0.605 0.909 F-test p-value (3=1) 0.552 0.559 0.615 0.703 F-test p-value (4=1) 0.006 0.184 0.361 0.092 F-test p-value (1=2=3=4) 0.027 0.182 0.437 0.251 Panel C: Heterogenous treatment effects by baseline business practices (BPs) Any information sheet*Low BPs 0.080** 0.056 0.055 0.054 (0.038) (0.045) (0.054) (0.057) Any information sheet*High BPs 0.057 0.078 -0.027 -0.053 (0.046) (0.048) (0.059) (0.063) R2 0.288 0.245 0.209 0.193 F-test p-value (Low BPs=High BPs) 0.707 0.751 0.318 0.220 Control mean low BPs 0.093 0.151 0.244 0.314 Control mean high BPs 0.177 0.190 0.380 0.468 N 866 866 866 866 Notes: This table shows OLS estimates for the effect of receiving the information sheet on accessing SEBRAE material (either via a service center or SEBRAE's website). The outcome variables are dummy variables equal to one if the firm accessed SEBRAE-PR material within the stated number of months after the intervention and zero otherwise. Panel A pools all treatment groups. Panel B shows separate treatment effects for each of the four different versions of the information sheet. Panel C shows heterogenous treatment effects for firms using below and above the median percentage of baseline business practices. BP stands for business practices. Low BP refers to below the median (50%), while high BP refers to above the median. All regressions control for the fraction of firms that accessed SEBRAE material within 12 months before the intervention and strata fixed effects. The regressions in Panel C additionally control for a dummy variable equal to one if the firm had low BPs at baseline. ***, **, and * indicate significance at the 1, 5, and 10 percent level. To deal with multiple hypothesis testing, †††, ††, and † indicate sharpened q-values statistically significant at the 1, 5, and 10 percent level. 24 Appendix Table A4: Effect of the information intervention on receiving SEBRAE training Received SEBRAE training within 3 months 6 months 12 months 18 months (1) (2) (3) (4) Panel A: Pooled treatment effect Any information sheet 0.019 0.018 0.002 -0.002 (0.018) (0.018) (0.023) (0.034) R2 0.235 0.279 0.220 0.150 Control mean 0.042 0.048 0.083 0.185 Panel B: Effects for each treatment group (1) No motiv, no pre-enroll 0.025 0.018 0.020 0.027 (0.023) (0.023) (0.030) (0.044) (2) No motiv, pre-enroll 0.013 0.009 -0.029 -0.043 (0.023) (0.022) (0.026) (0.041) (3) Motiv, no pre-enroll 0.017 0.024 0.010 -0.003 (0.022) (0.024) (0.029) (0.042) (4) Motiv, pre-enroll 0.020 0.019 0.005 0.012 (0.022) (0.022) (0.028) (0.041) R2 0.236 0.279 0.224 0.153 F-test p-value (2=1) 0.606 0.692 0.078 0.096 F-test p-value (3=1) 0.732 0.825 0.745 0.492 F-test p-value (4=1) 0.814 0.959 0.599 0.721 F-test p-value (1=2=3=4) 0.964 0.933 0.249 0.345 Panel C: Heterogenous treatment effects by baseline business practices (BPs) Any information sheet*Low BPs 0.056***†† 0.060***†† 0.040 0.025 (0.017) (0.018) (0.027) (0.046) Any information sheet*High BPs -0.025 -0.032 -0.035 -0.028 (0.032) (0.032) (0.038) (0.052) R2 0.243 0.289 0.231 0.158 F-test p-value (Low BPs=High BPs) 0.024 0.011 0.115 0.454 Control mean low BPs 0.000 0.000 0.035 0.128 Control mean high BPs 0.089 0.101 0.139 0.253 N 866 866 866 866 Notes: This table shows OLS estimates for the effect of receiving the information sheet on receiving SEBRAE training, including classes, consulting, and networking events. The outcome variables are dummy variables equal to one if the firm received SEBRAE training within the stated number of months after the intervention and zero otherwise. Panel A pools all treatment groups. Panel B shows separate treatment effects for each of the four different versions of the information sheet. Panel C shows heterogenous treatment effects for firms using below and above the median percentage of baseline business practices. BP stands for business practices. Low BP refers to below the median (50%), while high BP refers to above the median. All regressions control for the fraction of firms that received SEBRAE-PR training within 12 months before the intervention and strata fixed effects. The regressions in Panel C additionally control for a dummy variable equal to one if the firm had low BPs at baseline. ***, **, and * indicate significance at the 1, 5, and 10 percent level. To deal with multiple hypothesis testing, †††, ††, and † indicate sharpened q- values statistically significant at the 1, 5, and 10 percent level. 25 Appendix Table A5: Self-reported reasons for not adopting more business practices Fraction of firms mentioning this reason at follow-up (multiple mention) Control Treatment Lack of time 0.434 0.408 Difficulty of finding skilled workers 0.301 0.265 I am happy with my choices 0.294 0.205** Lack of funds 0.294 0.367 Owner's cautiousness 0.191 0.229 Lack of management support services 0.154 0.117 Owner's lack of technical knowledge 0.14 0.112 N 136 547 Notes: This table shows reasons that firms reported on the follow-up survey for not adopting more business practices. These reasons were collected for all firms that participated in the follow-up survey. Stars refer to differences in fractions of the treatment group and the control group. ***, **, and * indicate significance at the 1, 5, and 10 percent level. 26 Appendix 1 – NaN business practice questionnaire No. Area Question 1* Finance Você controla as entradas e saídas diárias de dinheiro? Do you keep track of daily money inflows and outflows? 2* Finance Você possui um controle de entradas, saídas e saldo de dinheiro para os meses futuros? Do you project income, expenses, and cash balance for future months? 3* Finance Você sabe quanto dinheiro deve ter em caixa para cobrir suas despesas enquanto não recebe os pagamentos? Do you know how much money you need to have to cover your expenses in case you don’t receive payments? 4 Finance Você controla as datas e valores de todos os tributos a serem pagos? (Impostos e também Taxas) Do you record the dates and amount for all taxes and fees you need to pay? 5* Finance Você calcula os gastos (custos e despesas) mensalmente? Do you calculate costs and expenses every month? 6* Finance Você calcula o resultado (lucro ou prejuízo) do seu negócio periodicamente? Do you calculate business profits regularly? 7* Finance Você sabe qual é o seu ganho de dinheiro em cada produto ou serviço vendido? Dou you know how much profit you make on each product or service sold? 8* Finance Você tem um controle do estoque dos produtos e insumos? (incluindo datas de validade, quantidade, etc.) Do you track your inventory of products and inputs (including expiration dates, quantity, etc.)? 9 Marketing Sua empresa tem um cadastro com informações de seus clientes em algum lugar? Do you keep a registry of information on your clients? 10* Marketing A empresa sabe quais são os diferentes grupos de clientes que frequentam seu estabelecimento e faz ações direcionadas a cada um destes grupos? Do you know what types of customers buy from your firm and do you cater to these different groups specifically? 11* Marketing Você analisa regularmente as necessidades dos clientes para orientar suas ações de marketing e definição de serviços e produtos? Do you regularly analyze your customers’ need to target marketing campaigns and adapt services and products? 27 12* Marketing Você faz alguma pesquisa para saber a satisfação dos clientes? Do you conduct customer satisfaction surveys? 13* Marketing Você planeja ações para promover e divulgar seus produtos e serviços? Do you engage in marketing campaigns? 14 Marketing Sua empresa oferta novos produtos/ serviços para seus clientes? (Observa critérios para escolha dos produtos e teste dos produtos novos) Do you offer new products or services to your clients (purposefully developing and testing new products)? 15* Marketing Você avalia seus concorrentes para saber o que deve melhorar? Do you assess your competitors to find out how you can do better? 16 Strategic Você planeja os objetivos que deseja alcançar, tendo alguma forma de registro de Planning onde quer chegar? Do you strategically set and plan your goals? 17* Strategic Você tem seus objetivos e metas escritos e bem definidos? Planning Are your objectives and goals written down and clearly defined? 18* Strategic Você toma nota das decisões e resultados de seu planejamento, observando Planning situações em que houve resultados diferentes dos que você queria, para aprimorar o que deu certo e não repetir erros do passado? Do you keep track of decisions and results of your planning, paying attention to situations where the results were different than desired, to gauge what went well and which errors not to repeat? 19 Strategic Existem ações com prazos definidos para que os objetivos planejados pela Planning empresa para o futuro possam ser alcançados? Have you defined concrete actions with deadlines to help reach the firm’s goals? 20* Organization Você possui um controle de quem são os fornecedores de seus produtos e prestadores de serviço, facilitando sua pesquisa de preços? Do you keep a register of suppliers and service providers for your products to help research prices? 21 Organization Os documentos importantes de sua empresa e contas estão bem organizados? Are your important documents and accounts well organized? 22 Organization As atividades realizadas regularmente possuem um procedimento escrito de como elas devem ser feitas? Do you have written instructions for how frequently occurring tasks should be executed? 23 Organization A organização da sua empresa permite que você trabalhe de maneira eficiente e ágil? 28 Is your firm organized in a way that allows you to work efficiently and nimbly? 24 Organization Sua empresa possui um layout (arranjo físico) definido para facilitar a entrega do produto/ serviço, comercialização e atendimento aos clientes? Are your premises designed in a way that facilitates delivery and sales of your product or service, as well as customer service? 25 Human Você sabe quais são as obrigações trabalhistas de sua empresa, da contratação resources até o desligamento? Are you aware of your firms’ obligations towards its employees, from hiring to firing? 26 Human Você já definiu que conhecimentos e características as pessoas devem ter para resources realizar as atividades em sua empresa, incluindo você e sócios? Have you assessed which knowledge and characteristics staff should have to successfully carry out tasks in your firm, including yourself and partners? 27 Human Se você, seu sócio ou algum funcionário faltar ou tirar férias, está definido um resources responsável para ficar no lugar desta pessoa? If you, your partner, or an employee is not able to come to work or takes vacation, is it clearly established who will act for this person? 28 Human Você faz uma avaliação dos problemas recorrentes ou potenciais em seu negócio, resources visando evitar que os mesmos ocorram? Do you try to anticipate recurring or potential problems in your business with the goal of preventing them from happening in the first place? 29 Human Se for necessário delegar alguma atividade (por motivo de doença, expansão, resources etc.), você já sabe quais são as atividades que pode delegar? If you had to delegate some tasks (due to illness, expansion etc.) do you know which tasks you can delegate? * Corresponds to a business practices question asked in our follow-up survey, based on McKenzie and Woodruff (2016). 29 Appendix 2: Sample NaN report 30 31 32 33 34 35 Appendix 3: Sample information sheet 36