Putting a Price Tag on the Cost of Loan Subsidies

Written by Armando Castelar.

Brazil is in the middle of its worst fiscal crisis in decades. In 2015, the primary budget deficit (i.e., before interest payments)peaked at 1.9% of GDP, while the overall deficit reached 10.3% of GDP. Market analysts project equally scary deficits for this and the next years, the financing of which will raise the public debt to an eye-watering 82.1% of GDP at the end of 2017, up from 66.2% in 2015 and 51.7% in 2013.

The explosive dynamics of the public sector accounts is the main reason for Brazil’s deep economic crisis, which will cause GDP to contract by about 8% in 2015-16. Although fiscal adjustment is essential, disagreement about how to reduce the budget deficit has delayed significant action in this front. The private sector wants the government to cut expenditures, arguing that the tax burden is already too high. The government, in turn, has reacted by raising taxes and asking Congress to create yet new taxes. In particular, it argues that there is little scope to lower expenditures, for most of them are either earmarked (e.g., for education and health) or obligatory (personnel, social security benefits etc.). Indeed, the government says, cutting expenditures is not only difficult, but also unfair, for it would require cutting down on social programs, thus disproportionately affecting the poor.

Luckily, the Ministry of Finance (MoF) has just released a study that allows for a more nuanced view of this debate by highlighting an often underappreciated source of government outlays: loans by Brazil’s National Development Bank, BNDES.[1] The study puts a price tag on the cost of two of the three main types of subsidies firms received when they borrow from BNDES. In the study’s parlance, these comprise:

1.   Credit or implicit subsidies. These type 1 subsidies derive from the difference between the interest rate the government pays when borrowing from the market and the rate it charges on loans extended to BNDES, which are then used to fund the bank’s own lending. Those subsidies have risen significantly since 2009, as the Treasury rapidly expand the value of loans to the bank, which reached R$ 523.9 billion at the end of 2015 (Figure 1). In present value terms, those subsidies added R$ 114.8 billion to the government’s outlays in 2008-15 and the MoF estimates that they will it cost the taxpayer an additional R$ 175.0 billion in 2016-60. Overall, they amount to roughly 4.9% of GDP, almost seven times the annual revenue estimated for the financial transaction tax (CPMF) the government is trying to create. About 70% of these subsidies will be disbursed, in present value terms, in 2010-20, when they will average 0.31% of GDP per year.

Figure 1: Treasury loans to BNDES (R$ mn)

Source: Central Bank.

BNDES was not alone in operating this type of quasi-fiscal instrument: the Treasury also lent Type 1 funds at subsidized rates to Banco do Brasil (BB) and Caixa Econômica Federal (CEF), two other state-owned banks. BB focuses on agriculture, while CEF favors loans to housing and infrastructure. The Treasury loans extended to these banks, though, were substantially smaller; overall, I estimate a ballpark figure of 0.4% of GDP for the corresponding subsidies for the overall period, in present value terms.

2.  Financial or explicit subsidies. These type 2 subsidies result from the difference between the interest rate BNDES charges for especially favored loans, extended through the Program to Sustain Investment (PSI), and the cost of these loans to the bank, including fund and operational costs and a profit margin. The Treasury directly pays for those subsidies, through transfers to BNDES, the reason why the MoF labels them “explicit”. The PSI started in 2009, with an authorization for BNDES to extend PSI loans of up to R$ 43 billion, but this ceiling rapidly climbed, reaching R$ 452 billion in 2015. In present value terms, those subsidies cost the taxpayer R$ 46.7 billion in 2009-15 and, according to MoF estimates, the Treasury will pay an additional R$ 25.1 billion in 2016-41. Overall, the government will spend 1.2% of GDP to fund those subsidies. In present value terms, the Treasury will disburse 95% of those subsidies in 2010-20, when they will average 0.11% of GDP per year.

Large firms receive most subsidies channeled through BNDES. For instance, the top 1% of the borrowers who benefitted from the PSI in 2009-14 accounted for 53% of the program’s loan value and probably an even larger share of the subsidies (see here). BNDES reckons this distribution is similar to that of its overall lending (see here) so the distribution of type 1 subsidies follows an analogous pattern.

Although a recent addition in the case of BNDES, there is a long history of type 2 subsidies being channeled through other state-owned banks. Thus, a substantial share of the loans that BB extends to the agricultural sector carry below-market interest rates. The Treasury then compensates BB through an annual transfer of budgetary resources. In 2013-15, annual loan subsidies to the agricultural sector amounted on average to R$ 8.8 billion (0.15% of GDP per year). The distribution of those subsidies across farmers is also concentrated on the top ones. Moreover, it is not clear why they are necessary, considering Brazil’s comparative advantage in agriculture.

There is a third type of subsidy, unaccounted for in the MoF calculations: roughly, 27% of BNDES’s overall funding comes from taxes, through the FAT (Worker’s Support Fund). These funds cost to BNDES the same as the loans provided by the Treasury and amount to 40% of their value. Differently from the Treasury loans, though, BNDES does not need to pay them back to the FAT, barring critical circumstances, so the present value of the implied subsidies per R$ is proportionately larger.

A similar mechanism generates funding for loan subsidies channeled through CEF. CEF manages the FGTS, a fund to which workers compulsorily contribute with 8.25% of their monthly salaries, and to which they have access when fired from a job, retiring, getting married or falling seriously sick. The FGTS pays a negative real return (see here). CEF then transforms the negative return subtracted from the workers’ savings to its clients in the form of below market interest rates. This is a loan subsidy in the sense that the workers are unwittingly subsidizing CEF programs by accepting a negative real return for their retirement and unemployment savings.

Tax exemptions on interest income accruing from certain savings instruments make possible a fourth type of loan subsidy. The most traditional of these instruments is the “Caderneta de Poupança”, a simple savings account that pays a monthly interest rate of 0.5%, plus a base rate, the TR, which also indexes the FGTS. In 2015, the TR was 1.8% and the annual return for the Caderneta added 8.1%. The government exempts this return from income and other taxes. Had the government levied a 15% tax on this income, it would have collected R$ 1.7 billion in 2015. So this loan subsidizes citizens’ savings – theoretically a noble and worthwhile policy—but the flip side of this tax exemption is the obligation of financial institutions to use the majority of the funds deposited on those savings accounts to finance agriculture (BB) and housing (all other banks). More recently, the government extended this privilege to other financial instruments (e.g. CRIs, LCIs, CRAs, LCAs, and infrastructure debentures). Financial institutions have to use the money raised through the sale of these securities to finance housing, agriculture or infrastructure.

A fifth type of loan subsidy comes from public sector guarantees on certain credit operations. This is the case, for instance, of the CCR (Convênio de Crédito Recíproco), an agreement through which central banks guarantee export credit operations across Latin American countries that are members of LAIA (Latin American Integration Association).

Governments in most countries provide one type or the other of loan subsidies to its firms and households. In this regard, there is nothing unique about Brazil. Still, it is remarkable that Brazilian taxpayers spend so much money with such unclear goals, in a regressive fashion, and with absolutely no formal assessment of its impact. At a time when the fiscal cost of such subsidies is rising so dramatically, it makes sense to reassess their utility and check how much bang taxpayers are really getting for their buck spent on loan subsidies.

[1] The Valor Econômico newspaper was the first to report on the study, on February 10, 2016 (see here).