Red tape can be a significant barrier to having informal firms formalize and eventually benefit from any business support programs provided by governmental agencies. While the relationship between this formalization, access to finance and a firm’s performance has been implied by anecdotal evidence (de Soto 1989 and 2000), a recent survey of empirical evidence suggests that such programs may in fact achieve the opposite and not necessarily nudge firms to formalize. In fact, the evidence suggests that even with a significant reduction in red tape most informal firms decide to remain informal (see Bruhn and McKenzie 2013 for a
survey).
One of the few studies that found large effects of a tax simplification program on formalization rates and firms’ performance is Fajnzylber et al. (
2011). In the paper the authors used two different quasi-experimental evaluation methods: a regression discontinuity (RD) design and RD combined with difference-in-differences (DID) to estimate the impact of the Brazilian program SIMPLES from November 1996. A bit rusty on these topics? I would suggest a quick read up
here and
here.
In Brazil, a tax simplification program (SIMPLES - Sistema Integrado de Pagamento de Impostos e Contribuições das Microempresas e Empresas de Pequeno Porte) was designed with the objective to simplify the tax system for SMEs in the country. The program used two official thresholds based on annual revenues, one for micro firms and one for small firms. The system combined six different federal taxes and social contributions into one single monthly-based rate and by doing so, reduced the tax burden. However in the beginning, only some sectors were covered by the new law. Eligible firms had to belong to one of the following sectors: retail trade, manufacturing, transportation, civil construction, and other services that do not require professionals with regulated occupations such as auditors, architects, dentists, engineers and others. (For a more detailed discussion of the program see Fajnzylber et al. 2011 and Monteiro and Assunção 2012).
Using the pooled sample of firms (micro and small) from the program, Fajnzylber et al. (2011) found that the program ended up increasing formalization by 11 percentage points or by 50% and used the program as an instrumental variable to assess the impact of formalization on firms’ performance. And they found very large effects on revenues and profits. Interestingly, Monteiro and Assunção (2012) used exactly the same dataset ( Economia Informal Urbana, ECINF 2007), though a different sample composition, and a DID evaluation approach to estimate the impact of the program and found limited impact on formalization rates, a result that contrasts with the findings obtained in Fajnzylber et al. (2011).
In a recent paper I attempted to reconcile these conflicting findings. Re-reading the papers I realized that Fajnzylber et al. (2011) used November 1996, the date of the provisional measure launching the program, as the cutoff. Monteiro and Assunção ( 2012) used the month the law was approved by the Brazilian congress as cutoff, i.e., December 1996. This subtle difference explains the main discrepancies of the results found in both studies.
Figures 1 and 2 below show the proportion of eligible formal firms before and after the program. In figure 1 the cutoff is defined in November 1996 and in figure 2 December 1996 is used instead. Ideally we would like to compare firms created just before and just after the law passed. For this purpose, the time (in months) the firms are in business is used as a running variable.
Thus, the figures used information of firms created between February 1996 and October 1997. This is equivalent to a window of about 10 months on each side of the cutoff point. Firms 0 to 11 months old were created between November 1996 and October 1997 and therefore were supposedly created after the program. Firms 11-20 months olds were created before the program. As can be seen, there is no apparent discontinuity in the formalization rate when the cutoff is defined in December 1996. This supports why Monteiro and Assunção (2012) did not find large effects on formalization.
In Brazil, a tax simplification program (SIMPLES - Sistema Integrado de Pagamento de Impostos e Contribuições das Microempresas e Empresas de Pequeno Porte) was designed with the objective to simplify the tax system for SMEs in the country. The program used two official thresholds based on annual revenues, one for micro firms and one for small firms. The system combined six different federal taxes and social contributions into one single monthly-based rate and by doing so, reduced the tax burden. However in the beginning, only some sectors were covered by the new law. Eligible firms had to belong to one of the following sectors: retail trade, manufacturing, transportation, civil construction, and other services that do not require professionals with regulated occupations such as auditors, architects, dentists, engineers and others. (For a more detailed discussion of the program see Fajnzylber et al. 2011 and Monteiro and Assunção 2012).
Using the pooled sample of firms (micro and small) from the program, Fajnzylber et al. (2011) found that the program ended up increasing formalization by 11 percentage points or by 50% and used the program as an instrumental variable to assess the impact of formalization on firms’ performance. And they found very large effects on revenues and profits. Interestingly, Monteiro and Assunção (2012) used exactly the same dataset ( Economia Informal Urbana, ECINF 2007), though a different sample composition, and a DID evaluation approach to estimate the impact of the program and found limited impact on formalization rates, a result that contrasts with the findings obtained in Fajnzylber et al. (2011).
In a recent paper I attempted to reconcile these conflicting findings. Re-reading the papers I realized that Fajnzylber et al. (2011) used November 1996, the date of the provisional measure launching the program, as the cutoff. Monteiro and Assunção ( 2012) used the month the law was approved by the Brazilian congress as cutoff, i.e., December 1996. This subtle difference explains the main discrepancies of the results found in both studies.
Figures 1 and 2 below show the proportion of eligible formal firms before and after the program. In figure 1 the cutoff is defined in November 1996 and in figure 2 December 1996 is used instead. Ideally we would like to compare firms created just before and just after the law passed. For this purpose, the time (in months) the firms are in business is used as a running variable.
Thus, the figures used information of firms created between February 1996 and October 1997. This is equivalent to a window of about 10 months on each side of the cutoff point. Firms 0 to 11 months old were created between November 1996 and October 1997 and therefore were supposedly created after the program. Firms 11-20 months olds were created before the program. As can be seen, there is no apparent discontinuity in the formalization rate when the cutoff is defined in December 1996. This supports why Monteiro and Assunção (2012) did not find large effects on formalization.
Since firms were allowed to register their business under the new regime despite the fact the law had not yet been approved, I decided to scrutinize the results obtained with November 1996 as cutoff. Because the identification strategy used in Fajnzylber et al. (2011) relies on a jump in the formalization rate among eligible firms created after the program, a couple of validity checks that are now standard in the RD literature were carried out. A McCrary density test, a balance check around the cut-off point and placebo tests were carried out to make sure the discontinuity observed around November 1996 (see fig.1) was due to the program.
In fact, the results identified some issues with the microdata that could cast some doubt on the plausibility of the identification strategy used in both studies. In particular, the placebo tests using November 1995, 1994 and 1993 as cutoff points suggest the jump observed around November 1996 might be picking some other thing such as some seasonality effect. Figure 3 summarizes the placebo tests with a 95 per cent confidence interval.
The dark gray bar is the proportion of formal firms in the ‘control’ group whereas the light gray bar shows the effect over the control group’s mean. The first and second estimates use November 1996 as a threshold, but the second estimate controls for a data heaping (or spikes) observed in October 1996. The paper has a more detailed discussion of these issues and their implication for RD estimates. The other three estimates are placebo effects.
The placebo results indicate the program was not effective in increasing formalization rates of small firms, however one should be cautious before jumping to that conclusion. The analysis suggests that the data used in both Fajnzylber et al. (2011) and Monteiro and Assunção (2012) do not allow for a clean identification strategy that could inform the impact of SIMPLES on firms’ formalization decisions and performance. The impact
of this program on formalization rate of firms thus remains an open question.
Join the Conversation