Macroprudential FX Regulations: Sacrificing Small Firms for Stability?

Nontradable2 firms in emerging economies issue large amounts of dollar debt, exposing their balance sheets to exchange rate movements and credit default risk3. This is worrisome since by exposing their own balance sheets they are indirectly exposing the asset portfolio of banks that lend to them. Regulatory authorities have responded implementing macroprudential policies on banks’ use of dollar funding as a source of bank lending, particularly to nontradable firms.4

However, the unintended distributional effects of these regulations on firms’ financing are not well understood and have remained unexplored in the literature. Macroprudential FX policies may impose disproportionate costs on small, financially constrained firms, for which currency mismatch is arguably a means for relaxing their borrowing constraints and for financing investment.5 In this column I provide a theoretical explanation and verify empirically that Macroprudential FX policies might increase financing disparities between small and large nontradable firms.

I propose a mechanism leading to heterogeneous responses of nontradable firms’ outcomes to a tax on banks’ dollar lending. This mechanism relies on the well documented empirical fact that dollar debt is cheaper than local currency debt, after correcting for expected exchange rate depreciation.6 And, therefore, some firms might find it beneficial to expose their balance sheets to currency risk by issuing dollar debt.

Under this mechanism, firm’s optimal decision on debt denomination is driven by two opposing forces: if they denominate their debt in local currency, firms avoid insolvency risk. On the other hand, if firms issue dollar debt and take currency mismatch risk, they can take advantage of the cheaper cost of borrowing in dollars. For financially constrained firms (e.g. small firms), currency mismatch allows them to relax their borrowing constraints and increase their leverage and investment possibilities. Meanwhile, for unconstrained firms (e.g. large firms), the trade-off between leverage gains and insolvency risk is irrelevant.

Imposing a macroprudential tax on lender’s dollar funding, ultimately increases firm’s cost of borrowing in dollars. If dollar debt becomes more expensive, the firm could find it optimal to switch away from dollar debt to more expensive but risk-free local currency debt. Alternatively, if the gains of taking on mismatch risk are still high enough after the tax, the firm could find it optimal to keep on issuing dollar debt, and pay the tax. In either case, firm’s cost of borrowing increases after the tax is implemented. In an equilibrium where firms are small, borrowing constraints become tighter after tax. Then, not only does the tax might affect the currency composition of firms’ debt, but also generates real effects in the economy. By contrast, in an equilibrium of large firms, issuing dollar debt is not a means for relaxing borrowing constraints; the tax only generates a change in the currency composition of firm’s debt but does not affect total borrowing.

To verify this empirically, I take advantage of an unprecedented and arguably unexpected macroprudential FX policy intervention by the Central Bank of Peru (BCRP). In December 2014, it announced it was implementing a policy that would increase the reserve requirements (tax) on banks’ dollar liabilities. This increase was heterogeneous across banks, depending on the following rule: the BCRP required that by December 2015 (deadline), banks in the financial system had to reduce their stock of dollar loans7 to, at most, 90% of its own stock in September 2013 (benchmark). Otherwise, banks had to pay a tax on their FX liabilities proportional to the difference between their stock of dollar loans at the deadline and at the benchmark.

Based on this rule, two sources of variation determine banks’ exposure to the policy: First, for a given increase in the tax rate, banks that at the announcement more strongly rely on dollar funding are more exposed to the tax. Second, for a given degree of reliance on dollar funding, banks that at the announcement are further from the regulatory benchmark are also more exposed to the tax. These two sources of bank exposure are strongly correlated: banks that at the announcement rely more heavily on dollar funding are also further from the regulatory benchmark.8 Therefore, this regulation can be understood more simply as a progressive tax on banks’ dollar liabilities: the tax rate increases as the base increases.

Figure 1 shows the evolution of aggregate loans before and after a Macroprudential FX policy intervention by the Central Bank of Peru in December 2014. The blue dashed line shows the evolution of the normalized stock of dollar loans for the sample of nontradable firms and for the smallest (micro) and largest size categories (large). The red line shows the evolution of the normalized stock of total loans (in both dollars and soles). This plot shows that the growth rate of total loans was almost the same as that of dollar loans before the policy announcement. However, they start differing afterwards. In line with the proposed mechanism, this might suggest that firms were switching away from dollar borrowing to soles borrowing to avoid the higher costs of the tax. This substitution, if any, is clearly not complete for micro firms (panel a). However, large firms seemed to have completely avoided the tax by switching to soles debt, thus keeping their total financing unaffected relative to the pre-tax trend (panel b).

 Figure1b: Large firms

To formally verify this hypothesis, I exploit the cross-sectional variation in banks’ exposure to the tax to identify the bank lending channel of the macroprudential tax to nontradable firms. Simultaneously, I test whether firms borrowing from differently exposed banks respond heterogeneously to this supply shock.

Figure 2 shows the cumulative reduction in the monthly growth rate of dollar and total loans, for a given a marginal increase in bank exposure to the policy. The cumulative effect on dollar and total loans for the micro size segment become statistically significant at the 5% level, three months after the announcement. And after that, the cumulative effect on total loans departs from the effect on dollar loans. This suggests that firms begin to switch from dollar loans toward soles loans around four months after policy announcement. By the deadline, one year after the policy announcement, a 1% increase in bank exposure results in a cumulative reduction in the growth rate of dollar loans of 12.1 percentage points. The reduction in total loans is 8.7 percentage points.

For large firms, the cumulative effect of the policy on dollar and total loans is negative but not significantly different from zero in any of the periods after the treatment. Also, there seem to be a degree of substitution to soles loans starting 5 months after the announcement. In contrast to the micro segment, the overall cumulative effect on total loans is not statistically significant.

 Figure 2b: Large firms

I replicate my empirical strategy at the firm level to account for the possibility of firms borrowing in soles also from other banks, to avoid the burden of the regulation. I find that micro firms remain significantly negatively affected by the tax, while firms in larger size categories are able to exploit their multiple relationships with differently exposed banks to increase their debt in soles and remain unaffected by the tax. Additionally, I show that firms that are mostly affected by the policy are not hedged against exchange rate risk through FX derivatives.

Finally, I conduct a back-of-the-envelope calculation to show that the aggregate average effect of the policy in the growth rate of total loans is a reduction of 2.9 percentage points. For the micro segment, this reduction is 9 percentage points, for large firms, this reduction is of 1.6 percentage points.

In this column I provide evidence of a potential trade-off between small firms’ growth and financial stability that has not been studied in the literature. My results taken together show that policies aimed at achieving financial stability through the restriction of the bank lending channel in foreign currency, might end up disproportionally hurting small firms’ financing possibilities with potential real implications. It is worth noting that I am not taking a stance on the optimality of these policies, for which I would need to account for the likelihood and the size of the counterfactual growth losses of small firms in an episode of a large exchange rate depreciation. The findings in this study have important implications for policy design.