Republic of Poland: Selected Issues
Author: International Monetary Fund. European Dept.
Publication Date:
05 Jul 2016
eISBN:
9781475525632
Language:
English
Keywords:
ISCR; CR; bank; Poland; asset tax; tax; profitability of bank; tax payment; estimate of the impact; private sector credit; bank asset tax; Bank soundness; Credit
Download PDF (570.9 KB)

Republic of Poland: Selected Issues

Abstract

Republic of Poland: Selected Issues

Macro-Financial Implications of the Bank Tax1

While the Polish banking sector is still generally healthy, bank profitability has recently declined, reflecting narrowing interest margins and other costs on banks. Going forward, the new bank asset tax is expected to further dampen bank profits and a prolonged low-interest-rate environment is likely to continue to weigh on bank returns.2 Using bank-level data, this chapter analyzes the potential impact on bank profitability of the recently implemented bank asset tax. Furthermore, to examine macro-financial linkages, it undertakes an aggregate analysis to estimate implications for credit growth and economic activity. The chapter concludes with a summary of findings and policy recommendations.

A. Introduction

1. The Polish banking sector performs well on a number of financial soundness indicators. Banks are well capitalized and the ratio of non-performing loans to gross loans, though still elevated, compares well within the region. Yet, return on average assets is below that in some other countries with similar capital adequacy ratios and NPLs, such as the Czech Republic. Nonetheless, returns to average equity in Polish banks are close to the regional average (Figure 1).

Figure 1.
Figure 1.

Selected Countries: Financial Soundness Indicators

Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A003

2. Bank profitability has recently declined. The return on average assets has been trending down, starting in 2014 Q33—with low interest rates being the main driving factor. In addition, most Polish banks have seen an increase in costs in 2015, in particular in the fourth quarter as the bankruptcy of SK bank and the associated payment of guaranteed deposits by the Bank Guarantee Fund (BFG) (about PLN 2 billion) impacted bank balance sheets. As a result, costs for banks grew at a much higher rate than income, resulting in declining efficiency (as measured by the cost-to-income ratio) (Figure 2).

Figure 2.
Figure 2.

Poland: Factors Contributing to the Recent Drop in Bank Profitability

Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A003

3. Going forward, the new bank tax as well as some other factors would further reduce bank profits. Starting in February 2016, the bank asset tax is levied on the total value of assets (less PLN 4 billion, own funds, and purchased sovereign debt) at a monthly rate of 0.0366 percent (or 0.44 percent per year).4 This tax may have created incentives for banks to increase holdings of (exempted) local government bonds. Nonetheless, back-of-the-envelope calculations suggest that tax outlays paid by banks in 2016 may still double relative to 2015 levels. In addition, costs on banks are expected to increase, owing to increased contributions to BFG and contributions to finance the Borrower Support Fund (Figure 3).

Figure 3.
Figure 3.

Poland: New Factors Contributing to the Decline in Bank Profitability

Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A003

4. Continued weakening of bank profitability would have implications for the real economy. As costs on banks, including related to the bank asset tax, would reduce bank net profits, this could potentially prompt banks to reduce their supply of loans. With reduced credit availability, private sector investment and consumption would decline, slowing economic activity.

5. This chapter assesses the impact of the new bank asset tax on bank profitability, credit, and overall economic growth. The analysis is carried out in two stages. First, in a bank-level analysis this chapter assesses the impact of the bank asset tax on bank profitability. Second, to assess the impact on overall economic activity through macro-financial linkages, an aggregate-level analysis assesses implications for overall credit to the non-financial sector and real GDP growth.

B. Bank-Level Analysis: Impact on Bank Profitability

6. Econometric analysis points to a significant impact from the bank asset tax on bank profitability. To quantify the impact of the new tax on banks, econometric analysis is employed on annual data for 55 Polish banks and other depository corporations (see methodological details below).5 Specifically, using panel regressions, individual bank profitability was projected for 2016 under two scenarios:

  • Scenario 1 assumes that interest rates and taxes remain at the 2015 level;

  • Scenario 2 assumes that taxes increase corresponding to the new asset tax.

7. In both scenarios, Polish banks’ profitability is expected to decline further. The analysis indicates that even in Scenario 1, the annual return on assets could decline by 12 percent in 2016 relative to 2015, largely owing to higher costs in 2015. In Scenario 2, the new bank asset tax could reduce profitability by an additional 18 percent.6 If, in order to reduce the tax base, banks responded by increasing their holdings of government securities by, for example, 10 percent relative to the 2015 level, the difference in bank profitability relative to Scenario 1 would be 15 percent instead of 18 percent as shown in Scenario 2. Thus, even large increments in bank holdings of sovereign debt are likely to have only limited mitigating impact on profitability (Figure 4).

Figure 4.
Figure 4.

Poland: Profitability of Banks: Impact Analysis

Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A003

Sources: Bankscope and IMF staff calculations.Note: Based on results of System GMM regression. Impact is assessed on the annual levels of returns on average assets.

8. Aggregate regression analysis estimates the impact on the real economy. While the strength of the bank-level analysis includes the ability to control for bank-level characteristics, which may be important for explaining the evolution in profitability, it cannot fully capture dynamic effects. Specifically, with annual data, important higher-frequency information may be omitted, limiting the strength of the time-series analysis. In addition, data on lending are not available for all banks. For these reasons, the impact on economic activity is estimated using aggregate quarterly data.

C. Aggregate Analysis: Macro-Financial Implications

9. The aggregate analysis suggests that the bank tax could weaken overall credit growth. An aggregate vector autoregression (VAR) model is used to assess macro-financial implications of the asset tax (see Appendix I). First, a baseline scenario of no bank asset tax is projected from the VAR. Under this baseline, average credit growth in 2016 is expected to decline by 1 percentage points relative to 2015. An increase in tax outlays would reduce credit growth by between 3 (Approach 1) and 4 (Approach 2) percentage points at end- 2016 relative to the end-year baseline. The following two approaches were used:

  • Approach 1 estimates the impact on credit growth, taking advantage of the bank-level estimate of the impact of the bank tax on profitability, discussed above;

  • Approach 2 estimates the impact on credit growth by including tax payments directly in the VAR instead of the indirect effect through bank profitability.

10. As a result, real sector activity would weaken. In order to quantify the impact on overall economic activity, a standard OLS regression framework is used with real GDP and credit growth rates (see Appendix I). Credit growth rates are then varied, corresponding to the two estimates from the VAR analysis above. As shown in Figure 5, after accounting for the impact of the bank tax based on the impact on credit growth found under Approach 1, real GDP growth by end-2016 would be about 0.3 percentage points below a no-tax scenario. Based on Approach 2, the impact could be as large as 0.4 percentage points below a no-tax scenario.

Figure 5.
Figure 5.

Poland: Effect of Asset Tax on the Real Economy

Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A003

D. Summary of Findings and Policy Recommendations

11. This chapter finds that the bank tax can significantly reduce bank profits, slowing down credit and economic growth. Specifically, the bank asset tax could reduce bank profitability by 18 percent, significantly contributing to a total decline in profitability of 30 percent relative to 2015. Moreover, the analysis points to macro-financial implications. Based on aggregate data, credit growth at end-2016 may contract by between 3 and 4 percentage points relative to a projection without the bank asset tax. In turn, the effect of the bank asset tax on the economy would be significant. Staff estimates suggest that the bank tax alone could reduce GDP growth by up to 0.4 percentage points by end-2016.7

12. However, the analysis in this chapter may not fully account for all economic implications of the bank tax. For example, to the extent that the bank tax creates an incentive for banks to increase their holdings of government bonds, this could result in a shift from holdings of NBP bills to government bonds, potentially complicating the conduct of monetary policy. It could also crowd out private sector credit and intensify adverse feedback loops between the sovereign and banks, should public finances come under pressure. Moreover, policy uncertainty associated with potential introduction of additional distortionary sectoral taxes could further sour investor sentiment, increasing financing costs and weighing on public finances and growth. To mitigate the impact of the bank tax on profitability, banks could also respond by shifting lending away from corporate to more risky consumer lending, where spreads are higher, or shift lending activities to less-regulated non-banks. In turn, corporates may be able to obtain loans from other segments such as leasing companies.

13. The negative economic effects of the bank asset tax could be mitigated through better design. To the extent that a tax on banks is desired to raise revenue, a Financial Activities Tax (FAT) would be less distortionary for the private sector and the economy in general than a tax on assets (IMF, 2010). Denmark, Iceland, and Israel, for example, have imposed a FAT. A FAT, levied on the sum of profits and remuneration of financial institutions, is similar to a value added tax (VAT) from which the financial sector is currently exempt. Unlike the current bank asset tax, a FAT does not distort the structure of the activities undertaken by financial institutions themselves, as the tax depends on the level of profit, not on how the profit is earned or on the volume of turnover. For example, a 10–13 percent levy on bank profits and remunerations could generate the same revenue of PLN 4 billion expected from the current bank asset tax.8

References

  • Arellano, M., and S. Bond. 1991. Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies 58: 277–97.

    • Search Google Scholar
    • Export Citation

  • Bloomberg, April 20, 2016. New Bank Tax Smothers Polish Loan Growth as Economic Risks Rise.

  • Borio, Gambacorta, and Hofmann, 2015, “The Influence of Monetary Policy on Bank Profitability,” BIS Working Paper No. 514.

  • IMF, 2010, “A Fair and Substantial Contribution by the Financial Sector,” Final Report for the G-20.

  • Narodowy Bank Polski, 2016, “Financial Stability Report,” Warsaw, February.

Appendix I. Methodology

This chapter presents three types of analyses carried out to measure the macro-financial implications of the new bank asset tax. First, bank-level analysis provides the estimated impact of the asset tax on bank profitability. Second, to quantify the impact of the asset tax on the real economy, changes in credit growth are examined at the aggregate level. Subsequently, using the quantified impact on credit growth, this chapter estimates the association with real GDP growth.

1. Bank-level Analysis

The data used in this study are from the Bankscope database, with the final sample consisting of 39 commercial banks and 16 other depository corporations. The final sample was based on annual data, covering the period 2005-15.

Methodology

This chapter follows the framework of bank profitability, which depends on bank characteristics and the interest rate, as is common in the literature.

ROAAit=β1ROAAit−1+β2(Net loans / Assets)it−1+β3(Equity / Assets)it−1       β4(Cost / Income)it−1+β5Log(assets)it−1+β6Net Interest Marginit    +β7GDP growthit+β8(Tax / Pretax income)it    +β9(Government securities/assets)it+Year dummies+εit

As elsewhere in the literature (Borio and others, 2015), bank characteristics include 1st lags of (i) returns on average assets to capture the persistence of the time series; (ii) net loans to assets to capture net liquid financial assets; (iii) equity to assets to capture banks’ leverage; and (iv) cost to income ratio to capture bank efficiency. To capture the effect of the macroeconomic environment, the contemporaneous GDP growth rate and net interest margin (NIM) are included. Tax relative to pretax income is included to capture the effect of the tax outlays on banks’ returns. Since government securities are exempt from the tax base, holdings of government securities relative to assets are also included.

Similar to Borio and others (2015), this chapter employs the system GMM approach. The attractiveness of this approach is that it deals with the potential endogeneity issues that may arise from a given specification and produces unbiased and consistent estimates. System GMM also allows for dynamic processes. Errors may have heteroscedasticity and serial correlation. The instruments used in the regression are “internal” and consist of the lags of the instrumented variables, collapsed together (Arrellano and Bond, 1991).

Assessing the impact on bank profitability

Table 1 reports results from the main regression. As expected, bank liquidity and leverage are positively associated with profitability. Bank efficiency is negatively associated with bank profitability. The bank size, measured by the log of assets, has a counterintuitive sign. Still, in regressions with different instrument specifications, the sign of the bank size had a negative and expected sign. Net interest margin, which reflects banks’ reaction to changes in policy interest rates, has positive effect on banks profitability as expected. Bank profitability is also positively associated with higher GDP growth. Taxes are negatively associated with bank profitability, whereas holdings of government securities have positive impact. In addition, the diagnostics tests, as reported in the bottom section of the table, suggest well-identified internal instruments.

Table 1.

Poland: Commercial Bank Performance

article image

To forecast the impact of the asset tax on bank profitability a few assumptions were made. Specifically, assumptions were made on contemporaneous variables, NIM, tax to pretax income, GDP growth, and holdings of government securities. For the NIM, an assumption was of no change in 2016 relative to 2015.1 For the tax variable, in Scenario 1 it was assumed that banks would pay the same ratio of tax outlays relative to pretax income. In Scenario 2, the increment in the tax outlay was calculated as per asset tax regulations on an annual basis. Data from individual banks in 2015 showed little change in the pretax income relative to 2014. Therefore, it was assumed that there was no change in the pretax income in individual banks in 2016 relative to 2015. Staff projections of GDP growth were used for 2016.

Based on the latest available data from 2015, the average tax outlays paid by banks relative to assets were 20 percent. With the new asset tax, this ratio is expected to double. In scenario 1 with assumed unchanged NIM and assumed tax outlays as of 2015, bank profitability is expected to decline by 12 percent. When the asset tax is accounted for in Scenario 2, bank profitability is expected to decline by an additional 18 percent, for a total reduction in profitability of 30 percent. While the coefficient on government securities holdings is positive, a ten percent increase in debt holdings jointly with the new tax will reduce profits by 27 percent.

While the analysis of bank profitability captures the direct effect of the tax on returns, there may, however, be other channels through which the asset tax affects banks’ activity. With the new tax in effect for three months as of May 2016, individual banks have already started to engage in activities to reduce their tax base, including by increasing the holdings of government securities. Changes in the structure of loans may also occur. Nonetheless, whether such changes in the structure of loans will offset the initial negative effect of the asset tax is not clear.

2. Aggregate analysis

Credit flow

There is a natural link between credit flows and deposit flows in bank balance sheet. Therefore the VAR framework seems appropriate to capture the dynamics of these flows. At the same time bank characteristics, such as profitability, leverage, cost to income ratio, and cost components also influence the credit flows. To examine how bank profitability and tax outlays relate to credit supply, these indicators were included as exogenous variables in the VAR model.

The impact of the bank asset tax on credit was estimated using two different approaches, with the difference relating to the choice of exogenous variables:

  • Approach 1: exogenous variables include lagged real GDP growth and bank profitability.

  • Approach 2: exogenous variables include real GDP growth, lagged average banking sector NIM, and tax outlays relative to pretax income.

In Approach 1, quarterly data on return on average assets (ROAA) from the Bankscope database were used. When forecasting four quarters in 2016 in the baseline scenario, ROAA was adjusted downward by 12 percent year-to-year to reflect the estimates of the baseline projections of ROAA from the bank-level analysis. Similarly, in the impact scenario, ROAA in each quarter was reduced by 30 percent relative to the same quarter in 2015 as suggested by the bank-level analysis.

In Approach 2, NIM and tax outlays were obtained from the Polish Financial Supervision Authority (KNF) database, available online on the KNF website. Data for 2016Q1 already reflects recent changes in the tax outlays. For the rest of the quarters of 2016, it was assumed that tax outlays will be adjusted in the same proportion as in Q1 of 2016 relative to Q1 in 2015. No change in the net interest margin was assumed relative to the observed levels in the first quarter of 2016. Acknowledging that tax outlays only started in February, it is likely that the estimate of the impact is underestimated.

The VAR is estimated in log levels of credit and deposits. A Dickey-Fuller test on log credit rejects the null of a unit root, whereas the test for order of integration in log of deposits was inconclusive.2 In selecting the lag structure of the VAR, two criteria had to be met. First, the Akaike Information Criteria in the selected model has to closely match the global min of AIC (which in both cases is modeled with up to 5 lags). Second, the coefficient of interest in each of the Approaches had to be as close as possible to their true coefficient estimates, as shown in the OLS regression of log of credits on lags of credit and the indicator of the interest (ROAA as in Approach 1, and tax to pretax income as in Approach 2). In the end, the lag structure of the VAR model fits best the true relationship between flows of credit, flows of deposits, and bank performance indicators.

Using staff projections for quarterly real GDP growth, it is then possible to estimate the impact on credit. With respect to Approach 1, the results of the effect of the asset tax through reduced bank profitability on credit are shown in column 1 of Table 2. With respect to Approach 2, as expected, tax outlays are negatively associated with credit as shown in column 3.

Table 2.

Poland: Effect of Bank Asset Tax on Aggregate Credit Levels

article image

While the exercise analyzed the change in total loans, there may be changes in the structure of the loan portfolio, which this aggregate analysis does not capture. However, the empirical evidence from the first three months of the new asset tax suggests a reduction in corporate and mortgage loans—two large components of loans—and an increase in consumer lending—a smaller component of total loans. Overall, the changes in the structure of loans are not detrimental to the direction of credit growth in 2016 as a result of the bank asset tax.

One important caveat of the analysis is short data coverage and the use of levels rather than differences in the VAR setup. It is clear from the credit equation in both VAR models that the series is highly persistent. When instead credit growth and deposit growth were considered in the VAR setup, the effect of bank profitability and tax was even more prominent than suggested in the proposed model. In the future, however, as more reliable data becomes available the proposed model could be revisited to account for the statistical properties of the time series.

Real GDP growth

Linking the effect of the bank asset tax to the real economy is not straightforward. One natural channel through which the tax affects the real economy is via credit growth. Thus, by exploring the relationship between credit growth and GDP growth it is possible to approximate the impact of the tax on the real economy.

An OLS setup seems natural and the first step in establishing the direct link between credit growth and GDP growth rates. In this exercise, lagged credit growth rates are added to an autoregressive process of the GDP growth rate. The cumulative effect of the growth rate of credit over two consequent quarters is significantly and positively associated with GDP growth (Table 3).3

Table 3.

Poland: Effect of Credit Growth Rate Change in GDP Growth Rate

article image
1

Prepared by Yorbol Yakhshilikov.

2

While the new tax applies also to insurance companies, for the purposes of this annex, only the impact through the tax on banks and other depository corporations is assessed.

3

Returns on average assets (ROAA) is defined as net income relative to average total assets.

4

State-owned banks (relevant for BGK with legal status of state enterprise) and banks under recovery proceedings, in receivership, in liquidation, or those who filed for bankruptcy are excluded from the tax.

5

For the purpose of this annex, the set of banks and other depository institutions are referred to as banks.

6

According to Narodowy Bank Polski, the bank tax, combined with other costs, could increase the share of banks with negative net earnings from 2 to 22 percent of banking sector assets and reduce net earnings of the commercial banking sector by 60 percent (Narodowy Bank Polski, 2016).

7

Oxford Economics estimates that the impact of the asset tax could reduce 2016 growth by 0.2 percentage points (Bloomberg, April 20, 2016).

8

Israel levies an FAT at its standard VAT rate of 17 percent. Iceland levies a 5.5 percent tax on financial institutions’ wages and a 6 percent tax on their profits above a floor of ISK 1 billion. The profit base of both taxes is normal rather than economic profits. Denmark levies a 10.5 percent tax on wages.

1

While assumption of no change in NIM may appear strong, empirical data from individual banks over the last two years show slowly moving changes in the NIM.

2

Inconclusive results on non-stationary tests may be due to the short sample. However, when the difference in logs of credits and logs of deposits are considered instead, the effect of the asset tax is even larger.

3

In a VAR setup of GDP, investment, and consumption growth, with credit growth as an exogenous variable, the cumulative effect of credit growth over two quarters is of the same magnitude as produced in the OLS regression. In the VAR, credit growth is also positively associated with investment and consumption growth. Considering the effect of suppressed credit growth on investment and consumption, the total effect on the real economy as shown in the OLS setup thus may be underestimated.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Republic of Poland: Selected Issues
Author: International Monetary Fund. European Dept.
Volume/Issue:
Volume 2016: Issue 211
Publisher:
International Monetary Fund
ISBN:
9781475525632
ISSN:
1934-7685
Pages:
51
DOI:
https://doi.org/10.5089/9781475525632.002

Selected Countries: Financial Soundness Indicators

Poland: Factors Contributing to the Recent Drop in Bank Profitability

Poland: New Factors Contributing to the Decline in Bank Profitability

Poland: Profitability of Banks: Impact Analysis

Poland: Effect of Asset Tax on the Real Economy