Unintended consequences of tax reform in developing countries

Corporate tax reform has been a big topic of conversation among economists here in the U.S. with the passage of the recent tax bill on Capitol Hill. Proponents argued that lowering the corporate tax rate from 35 to 20 percent would increase U.S. competitiveness and increase after-tax profits for corporations and in turn increase investments and returns to labor.

A large component of the discussion surrounding the corporate tax rate in developed, wealthy nations revolves around international tax competition and large multinational firms with the capacity to move their operations across borders to take advantage of lower tax structures (see Devereux and Loretz, 2012, for a review of the relevant literature on tax competition). Much of the study of corporate taxes has been based in these high enforcement settings and has been a study of corporate firms.

Setting appropriate tax rates is arguably more challenging in developing countries with low enforcement and large informal sectors. In these settings firms (noncorporate firms in particular) have two additional methods of responding to changes in the tax rate: they can move into the large existing informal sector, thereby reducing the overall tax base, or they can engage in tax evasion, given the low enforcement capabilities of the government. Evidence from these settings can be informative as to the elasticity of noncorporate firms to tax rates in a setting that is distinct from that of developed countries where there are fewer tax evasion opportunities. Waseem (2017) recently published a paper in the Journal of Public Economics ("Taxes, Informality, and Income Shifting: Evidence from a Recent Pakistani Tax Reform") analyzing firm responses to an increase in the tax rate for noncorporate firms in Pakistan that looks at this issue.

He finds that the responses by firms to the tax reform were so large that the Pakistani government was collecting less in in revenue three years after the reform than it was prior to the reform. On the intensive margin, firms impacted by the increase in tax rate were reporting significantly smaller earnings. On the extensive margin, fewer firms were registered with the government and reporting positive profits, indicating a shrinking of the formal tax base.

The policy in question was enacted by the Pakistani government in 2009 and raised the tax rate on partnerships from a bracketed system with average tax rates that varied progressively from 0% to 25% (a tax rate that continued to be applied to sole proprietorships) to a flat tax rate of 25%. This change was enacted in order to reduce the variation in tax rate between partnerships and corporate firms, the latter of which was taxed at and remained taxed at 35% after the tax reform, and to reduce the disincentives for incorporation of new firms.

The implications of this study for developing countries are stark. Policies such as this one need to account for the "unintended" effects of tax reform on formalization and tax evasion as well as the intended effects on incentives to incorporate. High variation in the tax rates across noncorporate firms in this particular case (specifically between partnerships and sole proprietorships) led to significant levels of tax evasion, income shifting, and movement into informality that counteracted higher tax revenues collected from the partnership firms still remaining in the formal tax base and from incorporated firms.

Estimation strategy

The author employs a difference-in-differences method to take advantage of the tax reform in Pakistan in 2009. He uses administrative data of income tax returns filed from 2006-2011 to compare the outcomes of partnership and non-partnership firms (corporations and sole proprietorships): log change in reported earnings and log change in the number of filers. The former represents the intensive margin response of firms that remained in the tax base and the latter the extensive margin response of firms that exited the tax base.

The model estimated for the intensive margin elasticity is:

Δlog zit =α+β Partnershipi +ε Δlog(1−τit)+Xi δ+λt +uit

This regresses the log change in the reported earnings of a firm i at time t on its status as a partnership, the log change in its net-of-tax rate, a set of control variables, and year fixed effects.

To account for potential endogeneity between the change in the tax rate and change in reported earnings (i.e. not only does the change in the tax rate impact the change in reported earnings but reported earnings also impacts the tax rate), the author uses an instrumental variables strategy regressing the change in the tax rate on a dummy variable for whether the observation is a partnership firm in the post-reform period in a first-stage regression that allows him to isolate the variation in the tax rate change that is due to the tax reform for partnerships.



Key identification issue here is whether partnerships would in fact grow at the same rate as corporations and sole proprietorships in the absence of the reform. Based solely on intuition it would appear to me that larger, better financed, and more regulated firms (corporations and to an extent partnerships) would have different responses to shocks than smaller, less regulated firms with fewer financing options (sole proprietorships and to an extent partnerships). It may also be the case that corporations make up a disproportionate amount of certain industries (e.g. those with higher fixed costs and entry barriers or those that are more concentrated) than others, leading there to be differential responses to shocks in which certain industries are hit harder than others. This is especially relevant given that the reform took place in the immediate aftermath  of the 2007-2008 financial crisis.

To address the parallel trends assumption the author provides visual evidence of the trends in reported earnings and firm entry over time. It certainly appears that there is a parallel trend in reported earnings (though with only three data points it is challenging to confirm these similarities as a trend).

Waseem (2017) - Fig 4

He also shows the composition of the various industries in terms of the three firm types in the following charts. It appears that the composition is roughly balanced across partnerships and sole proprietorships but not as much so for partnerships and corporations (as I hypothesized in a previous paragraph). Helpfully, the author breaks out the results into two sets: one set using corporations as the control group and the other using sole proprietorships so we can investigate this further.

Waseem (2017) - Fig A1

The results reported in the paper are highly significant and robust to several sensitivities. This is the case for both regression iterations that rely on sole proprietorships as controls as well as those that rely on corporates as controls. The results indicate that the reported earnings and firm participation in the tax base are highly elastic to the partnership tax rate to a degree of 2:1 (for every percent decrease in the net-of-tax rate there is a close to two percent decrease in reported earnings).

Interesting implications

1. First, because this natural experiment exploits a change in the partnership tax rate rather than the corporate tax rate, the author finds the elasticity of reported earnings of likely domestic small to medium sized firms rather than large multinationals. The options for these firms in response to the tax rate (if at all) would be to report lower earnings, exit into the informal sector, or go out of business. This elasticity is distinct from one associated with corporate firms which more than likely have the option to invest and produce across borders (assuming mobility of capital) if faced with relatively high effective tax rates. 
2. Estimating the costs of shifting between business forms (sole proprietorship-partnership and partnership-corporation) would allow policymakers to set more appropriate tax rates. While the costs of moving from partnership to sole proprietorships are clearly low, "how low?" is a yet unanswered question that if answered could allow for more effective tax rate setting. If the variation between business forms is too high with respect to income shifting costs then you'll see greater income shifting with implications on overall revenue collection. 
3. In Pakistan, firms engaged in income shifting from former partnerships to sole proprietorships to take advantage of differential tax rates. Income shifting is not unique to developing countries - the recently passed tax bill in the U.S. will have its own income shifting implications to watch over the next few years. 
1. For example, the bill will allow sole proprietors and partnerships and other pass-through entities to deduct up to 20 percent of their revenue from their taxable income. It incentivizes individuals to classify their income as business rather than individual income and to move towards contract-based labor arrangements to reap larger tax benefits. This article provides a good summary on how this would work with various examples. Such a shift towards sole proprietorship and partnership income will have implications for revenue collection. It will also, as pointed out by journalists here, steepen the rise of the gig economy and the good and bad associated with it including fewer benefits and protections for workers.

Sources

1. Waseem, M. (2017). Taxes, informality, and income shifting: Evidence from a recent Pakistani tax reform. Journal of Public Economics. 
2. Devereux, M.P., & Loretz, S. (2012). What do we know about corporate tax competition? (Working Paper 12/29). Oxford University Centre for Business Taxation.