Labor tax evasion is a major policy issue that is especially salient in transition and post-transition countries. In this brief, we use firm-level administrative data, tax authorities’ audit data and machine learning techniques to detect firms likely to be involved in labor tax evasion in Latvia. First, we show that this approach could complement tax authorities’ regular practices, increasing audit success rate by up to 35%. Second, we estimate that about 30% of firms operating in Latvia between 2013 and 2020 are likely to underreport the wage of (some of) their employees, with a slightly negative trend.
Tax evasion is a major policy issue that is especially salient in transition and post-transition countries. In particular, “envelop wage”, i.e., an unofficial part of the wage paid in cash, is a widespread phenomenon in Eastern Europe (European Commission, 2020). Putnins and Sauka (2021) estimate that the share of unreported wages in Latvia amounts to more than 20%. Fighting labor tax evasion is a key objective of tax authorities, which face two main challenges. The first is to make the best use of their resources. Audits are costly, so the choice of firms to audit is crucial. The second challenge is to track the evolution of the prevalence of labor tax evasion. For this purpose, most of the existing literature relies on survey data.
In our forthcoming paper (Gavoille and Zasova, 2022), we propose a novel methodology aiming at detecting tax-evading firms, using administrative firm-level data, tax authorities’ audit data and machine learning techniques.
This study provides two main contributions. First, this approach can help tax authorities to decide which firms to audit. Our results indicate that the audit success rate could increase by up to 20 percentage points, resulting in a 35% increase. Second, our methodology allows us to estimate the share of firms likely to be involved in labor tax evasion. To our knowledge, this paper is the first to provide such estimates, which are however of primary importance in guiding anti-tax evasion policy. We estimate that over the 2013-2020 period, about 30% of firms operating in Latvia are underreporting (at least some of) their workers’ wages.
The general idea of our approach is to train an algorithm to classify firms as either compliant or tax-evading based on observed firm characteristics. Tax evasion, like any financial manipulation, results in artifacts in the balance sheet. These artifacts may be invisible to the human eye, but machine learning algorithms can detect these systematic patterns. Such methods have been applied to corporate fraud detection (see for instance Cecchini et al. 2010, Ravisankar et al. 2011, West and Bhattacharya 2016).
The machine learning approach requires a subsample of firms for which we know the “true” firm behavior (i.e., tax-evading or compliant) in order to train the algorithm. For this purpose, we propose to use a dataset on tax audits provided by the Latvian State Revenue Service (SRS), which contains information about all personal income tax (PIT) and social security contributions (SSC) audits carried out by SRS during the period 2013-2020, including the outcome of the audit. The dataset also contains a set of firm characteristics and financial indicators, covering both audited and non-audited firms operating in Latvia (e.g., turnover, assets, profit). Assuming that auditors are highly likely to detect misconduct (e.g., wage underreporting) if present, audit outcomes provide information about a firm’s tax compliance. Firms sanctioned with a penalty for, say, personal income tax fraud are involved in tax evasion, whereas audited-but-not-sanctioned firms can be assumed compliant. The algorithm learns how to disentangle the two types of firms based on the information contained in their balance sheets. Practically, we randomly split the sample of audited firms into two parts, the training and the testing subsamples. In short, we use the former to train the algorithm, and then evaluate its performance on the latter, i.e., on data that has not been used during the training stage. If showing satisfying performance on the training sample, we can then apply it to the whole universe of firms and obtain an estimate of the share of tax-evading firms.
In this study, we successively implement four algorithms that differ in the way they learn from the data: (1) Random Forest, (2) Gradient Boosting, (3) Neural Networks, and (4) Logit (for a review of machine learning methods, see Athey and Imbens, 2019). These four data mining techniques have previously been used in the literature on corporate fraud detection (see Ravisankar et al. 2011 for a survey). Each of these four algorithms has specific strengths and weaknesses, motivating the implementation and comparison of several approaches.
Table 1 provides the out-of-sample performance of the four different algorithms. In other words, it shows how precise the algorithm is at classifying firms based on data that has not been included during the training stage. Accuracy is the percentage of firms correctly classified (i.e., the model prediction is consistent with the observed audit’s outcome). In our sample, about 44% of audited firms are required to pay extra personal income tax and social security contributions. This implies that a naive approach predicting all firms to be evading would be 44% accurate. Similarly, a classification predicting all firms to be tax compliant would be correct in 56% of the cases. This latter number can be used as a benchmark to evaluate the performance of the algorithms. ROC-AUC (standing for Area Under the Curve – Receiver Operating Characteristics) is another widespread classification performance measure. It provides a measure of separability, i.e., how well is the model able to distinguish between the two types. This measure is bounded between 0 and 1, the closer to 1 the better the performance. A score above 0.8 can be considered largely satisfying.
Table 1. Performance measures
Random Forest is the algorithm providing the best out-of-sample performance, with more than 75% of the observations in the testing set correctly classified. Random Forest is also the best performing model according to the ROC-AUC measure, with performance slightly better than Gradient Boosting.
Our results imply that a naive benchmark prediction is outperformed by almost 20 percentage points by Random Forest and Gradient Boosting in terms of accuracy. It is important to emphasize that this improvement in performance is achieved using a relatively limited set of firm-level observable characteristics that we obtained from SRS (which is limited compared to what SRS has access to), and that mainly come from firms’ balance sheets. This highlights the potential gain of using data-driven approaches for the selection of firms to audit in addition to the regular practices used by the fiscal authorities. It also suggests a promising path for further improvements, as in addition to this set of readily available information the SRS is likely to possess more detailed limited-access firm-level data.
Share of Tax-Evading Firms Over Time and Across NACE Sectors
We can now apply these algorithms to the whole universe of firms (i.e., to classify non-audited firms). Figure 1 shows the share of firms classified as tax-evading over the years 2014 to 2019 for our two preferred algorithms – Gradient Boosting and Random Forest. Random Forest (the best performing algorithm) predicts that 30-35% of firms are involved in tax evasion, Gradient Boosting predicts a slightly higher share (around 40%). Both algorithms, especially Random Forest, suggest a slight reduction in the share of tax-evading firms since 2014.
Figure 1. Share of tax-evading firms over time
The identified reduction, however, does not necessarily imply that the overall share of unreported wages has declined. In fact, existing survey-based evidence (Putnins and Sauka, 2021) indicate that the size of the shadow economy as a share of GDP remained roughly constant over the 2013-2019 period, and that there was no reduction in the contribution of the “envelope wages”. With our method, we are estimating the share of firms likely to be involved in labor tax evasion. Unlike the survey approach, our methodology does not allow the measurement of tax-evasion intensity. In other words, the share of non-tax compliant firms may have decreased, but the size of the envelope may have increased in firms involved in this scheme.
Next, we disaggregate the share of tax-evading firms by the NACE sector. Figure 2 displays the results obtained with Random Forest, our best performing algorithm.
Figure 2. Share of tax-evading firms by NACE, based on Random Forest
First, the sector where tax evasion is the most prevalent is the accommodation/food industry, where the predicted share of tax-evading firms is 70-80%. Second, our results indicate that the overall decrease in the share of firms likely to evade is not uniform. It is mostly driven by the accommodation/food and manufacturing sectors. Other sectors remain nearly flat. This highlights the fact that labor tax evasion varies both in levels and in changes across sectors.
We show that machine learning techniques can be successfully applied to administrative firm-level data to detect firms that are likely to be involved in (labor) tax evasion. Machine learning techniques can be used to improve the selection of firms to audit in order to maximize the probability to detect tax-evading firms, in addition to the regular practices already used by SRS. Our preferred algorithms – Random Forest and Gradient Boosting – outperform the naive benchmark classification by almost 20 percentage points, which is a substantial improvement. Once implemented, the use of these tools can improve the audit effectiveness at virtually no extra cost.
Our findings also suggest a promising path for further improvements in the application of such methods. The improvement in predictive power achieved by our proposed algorithm is attained by using a limited set of variables readily available from the firms’ balance sheets. Given that SRS is likely to have access to more detailed firm-level information that cannot be provided to third parties, there is clear room for improving the performance of the algorithms by using such limited-access data.
Acknowledgement: The authors gratefully acknowledge funding from the Latvian State Research Programme “Reducing the Shadow Economy to Ensure Sustainable Development of the Latvian State”, Project “Researching the Shadow Economy in Latvia (RE:SHADE)”; project No VPP-FM-2020/1-0005.
- Athey, Susan, and Guido Imbens. 2019. “Machine Learning Methods That Economists Should Know About.” Annual Review of Economics 11: 685–725.
- Cecchini, Mark, and Haldun Aytug, and Gary J. Koehler, and Praveen Pathak, 2010. “Detecting management fraud in public companies“. Management Science 56, 1146-1160.
- European Commission, 2020. “Undeclared Work in the European Union. Special Eurobarometer 498” (Report)
- Gavoille, Nicolas and Anna Zasova, 2022. “Estimating labor tax evasion using tax audits and machine learning”, SSE Riga/BICEPS Research papers, forthcoming.
- Putnins, Talis, and Arnis Sauka, 2021. “Shadow Economy Index for the Baltic Countries 2009–2020” (Report), SSE Riga
- Ravisankar, Pediredla, and Vadlamani Ravi, and Gundumalla Raghava Rao, and Indranil Bose, 2011. “Detection of financial statement fraud and feature selection using data mining techniques“. Decision Support Systems, 50(2), 491-500.
- West, Jarrod, and Maumita Bhattacharya, 2016. “Intelligent financial fraud detection: a comprehensive review“. Computers & security, 57, 47-66
Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.
In August 2017, the Latvian parliament adopted a major tax reform package that will come into force in January 2018. This reform was a long-awaited step from the Latvian authorities to make the personal income tax more progressive. Some of the elements of the adopted reform, e.g. the changes in the basic tax allowance are estimated to help reducing the tax wedge on low wages and help addressing the problem of high income inequality. At the same time, the way the newly introduced progressive tax rate is designed will effectively lead to a reduction in the tax burden on labor and will hardly introduce any progressivity to the system.
In recent years, reducing income inequality has become one of the top priorities of the Latvian government. Income inequality in Latvia is higher than in most other EU and OECD countries, and the need to address this issue has been repeatedly emphasized by the Latvian officials, the European Commission, the World Bank and OECD.
The main reason for high income-inequality is a low degree of income redistribution ensured by the tax-benefit system. The personal income tax (PIT) has been flat since the mid-nineties. While the non-taxable income allowance introduces some progressivity to the system, the Latvian tax system is characterized by a very high tax burden on low wages, compared to other EU and OECD countries.
Since the beginning of 2017, the government has worked on an extensive tax reform package that was passed in the parliament in August and will become effective as of January 2018.
Two years ago, we wrote about the tax reform of 2016. In this brief, we estimate the effect of the 2018 reform on the tax burden on labour and income inequality. We will only consider changes in direct taxes on personal income – the changes in enterprise income tax and excise tax are outside the scope of our analysis. Parts of our estimations are done using the tax-benefit microsimulation model EUROMOD (for more details about the EUROMOD modelling approach, see Sutherland and Figari, 2013) and EU-SILC 2015 data.
Tax reform 2018
We focus our analysis on four elements of the reform that are expected to affect income inequality and that are described below. In our simulations, however, we take into account all changes in the PIT rules.
First, the flat PIT rate of 23% will be replaced by a progressive rate with three brackets: 20% (applied to annual income not exceeding 20,000 EUR), 23% (for annual income above 20,000 EUR and below 55,000 EUR) and 31.4% (applied to income exceeding 55,000 EUR per year).
Second, the maximum possible PIT allowance will be increased and the structure of the PIT allowance will be made more progressive. Latvia has a differentiated allowance since 2016, which means that individuals with lower incomes are eligible for a higher tax allowance. Figure 1 shows the changes in the non-taxable allowance that will be introduced by the reform. Another important change is that the differentiated allowance will be applied to the taxable income in the course of the year. The current system foresees that, during a calendar year, all wages are taxed applying the lowest possible allowance (60 EUR per month in 2017), but workers eligible for a higher allowance have to claim the overpaid tax in the beginning of the next year.
Figure 1. Basic PIT allowance before (2017) and after (2018-2020) the reform, EUR
Third, the rate of social insurance contributions will be increased by 1 percentage point. Social insurance contributions are capped and the cap will be increased from 48,600 EUR per year to 55,000 EUR per year, i.e. to the same income threshold that divides the top PIT bracket.
Finally, the reform will modify the solidarity tax – a tax, which was introduced in Latvia in 2016 and which is paid by top income earners. When this tax was initially introduced, one of its objectives was to eliminate the regressivity from the tax system caused by the cap on social insurance contributions. Hence, the rate of the solidarity tax was set at the same level as the rate of social insurance contributions and was effectively replacing social insurance contributions above the cap. The reform foresees that part of the revenues from the solidarity tax (10.5 percentage points) will be used to finance the top PIT rate. This element of the reform implies that after January 2018 those falling into the top PIT bracket will, in fact, not face a higher PIT rate than those falling into the second income bracket – the introduction of the top rate will be offset by the restructuring of the solidarity tax.
There are four main findings. First, the reform will reduce the tax wedge on labor income, whereas the tax wedge on low wages will remain high by international standards. Second, most of the PIT taxable income earners (93.5%) will fall into the bottom income bracket. Hence the reform will effectively reduce the tax burden, while the effect on progressivity is very limited. Third, the (small) increase in tax progressivity is ensured mainly by changes in the tax allowance, while the effect of changes in the tax rate on progressivity is negligible: Even those few PIT payers that fall into the top tax bracket will not experience any increase in the tax burden due to a compensating change in the solidarity tax. Finally, it is mainly the households in the middle of the income distribution that will gain from the reform.
Effect on tax wedge
We start with a simple comparison of the average labor tax wedge in Latvia and other OECD countries for different wage levels before and after the reform. The tax wedge measures the share of total labor costs that is taxed away in the form of taxes or social contributions payable on employees’ income.
Table 1. Average tax wedge for single wage earners without dependents in Latvia and other OECD countries, before and after the reform
67% of average worker’s wage
100% of average worker’s wage
167% of average worker’s wage
|OECD average in 2016, % (a)||32.3||36.0||40.4|
|Latvia 2016, % (a)||41.8||42.6||43.3|
|Latvia’s rank in 2016* (a)||6||11||16|
|Latvia 2018, % (b)||39.4||42.3||42.6|
|Latvia 2019, % (b)||39.1||42.1||42.6|
|Latvia 2020, %(b)||39.0||41.9||42.8|
Source: (a) OECD and (b) authors’ calculations. Note: * Ranking across 35 OECD countries. Higher ranking implies higher tax wedge relative to other countries.
Table 1 shows that the tax wedge on low wages (67% of an average worker’s wage) in Latvia is pretty high. In 2016, it was the 6th highest across OECD countries, while the tax wedge on high incomes (167% of the wage) is much closer to the OECD average.
While the reform will slightly reduce the tax wedge for low wage earners (from 41.8% to 39.0% in 2020), it will still remain high by OECD standards. Despite an increase in PIT rate for high-income earners, the reform will also lower the tax wedge for those who earn 167% of the average wage. Why? The explanation comes from the income thresholds for the tax brackets. The income of those earning 167% of the average wage is estimated to fully fall into the first tax bracket in 2018–2019 and only slightly exceed the income bracket for the second PIT rate by 2020. This means that most of the incomes of people earning 167% of the average wage will be taxed at the rate of 20%, which is lower than the current flat rate of 23%. Moreover, in 2020, only a small share of their income will be taxed at 23% – the same rate that these individuals would have had faced in the absence of the reform. Hence, we observe a reduction in the tax wedge for high-income earners.
Generally, only a very small share of taxpayers will fall into the middle and the top income brackets. According to our estimations, as many as 93.5% of all PIT taxable income earners will fall into the lowest income bracket, and only about 6.5% will fall into the second income bracket and about 0.5% will face the top PIT rate.
Apart from the progressive PIT schedule, the reform envisages important changes in the solidarity tax. As explained above, part of the revenues from the solidarity tax will be used to finance the top PIT rate. Therefore, even those (very few) taxpayers whose income will exceed the threshold for the top PIT rate, will not experience any increase in the tax burden because of the compensating change in the solidarity tax. Therefore, the reform will effectively reduce the tax burden on labour with very little effect on progressivity.
While lowering the tax burden is generally welcome, the motivation for applying the top rate to such a small group of taxpayers is not clear. For example, in their recent in-depth analysis of the Latvian tax system, the World Bank (World Bank, 2016) came up with a tax reform proposal that envisaged a considerably lower threshold for the top PIT rate, which, according to our estimations, would cover about 12% of the taxpayers. Given the limited budget resources and an especially high tax wedge on low wages, a more targeted reduction in the tax burden would be preferable. Similar concerns about insufficient reduction in the tax burden on low-income earners are expressed in the latest OECD economic survey of Latvia (OECD, 2017).
Effect on income distribution
Below we present the results from the tax-benefit microsimulation model EUROMOD. Figure 2 shows the simulated change in equivalized disposable income by income deciles compared to the baseline “no-reform” scenario in 2018-2020.
Figure 2. Change in equivalized disposable income by income deciles caused by the reform compared to “no-reform” scenario, %
The first thing to note is that these are mainly households in the middle of the income distribution who will gain from the reform – their income will increase due to both the increase in non-taxable allowance and the introduction of the progressive rate.
The gain in the bottom of the income distribution is smaller for several reasons. First, the proportion of non-employed individuals (unemployed and non-active) is larger in the bottom deciles. Second, individuals with low wages are less likely to gain from the reduction in the tax rate and the increase in the basic allowance, since they might already have most of their income untaxed due to the currently effective basic allowance. The same applies to pensioners who have a higher basic allowance than the employed individuals and who are mainly concentrated in the bottom of income distribution.
Our results suggest that the wealthiest households will also see their incomes grow as a result of the reform (by about 1% in 10th decile). The growth is ensured by the fact that annual income below 20,000 EUR will be taxed at a reduced rate of 20%, and, taking into account that even in the top decile only about half of the individuals get income from employment that exceeds 20,000 EUR per year, the gain from the tax reduction is considerable even in the top decile. A reduction in the tax allowance for high-income earners will have a negative effect on wealthy individuals’ income, but this will be more than compensated by the above positive effect of the change in the tax rate. Hence, the net effect on the incomes in the top deciles is estimated to be positive.
Finally, Table 2 summarizes the effect of the reform on the income distribution, measured by the Gini coefficient on equivalized disposable income. On the whole, the reform is estimated to slightly reduce income inequality – in 2020, the Gini coefficient is expected to be 0.6 points lower than it would have been in the absence of the reform. This reduction is mainly driven by the changes in the non-taxable allowance, while the three PIT rates are estimated to have an increasing impact on income inequality.
Table 2. Gini coefficient on equivalized disposable income in the reform and “no-reform” scenario
Source: authors’ calculations using EUROMOD-LV model
The 2018 tax reform was a long-awaited step from the Latvian authorities on the way to a more progressive tax system. The planned changes in the basic tax allowance are estimated to help reducing the tax wedge on low wages and help addressing the problem of high income-inequality.
At the same time, the second major aspect of the reform, the introduction of a progressive PIT rate, raises more questions than answers. The progressive rate, the way it is designed, will effectively lead to an across-the-board reduction of the tax burden on labor and will hardly help to reach the proclaimed objective of taxing incomes progressively. Given the limited budgetary resources and given that taxes on low wages will remain high compared to other countries even after the reform, a more targeted reduction of the taxes on low-income earners would have been a more preferred option.
- OECD, 2017. “OECD Economic Surveys: Latvia 2017”, OECD Publishing, Paris. http://dx.doi.org/10.1787/eco_surveys-lva-2017-en
- Sutherland, H. and Figari, F., 2013. “EUROMOD: the European Union tax-benefit microsimulation model”, International Journal of Microsimulation, 1(6), 4-26.
- World Bank, 2016. “Latvia Tax Review”, available at http://fm.gov.lv/files/nodoklupolitika/Latvia%20Tax%20Review%20Draft%20231216%20D.pdf
2016 was the first full calendar year of the new Polish government elected to power in October 2015. The year marked a number of major changes legislated in the area of socio-economic policy some of which have already been implemented and others that will take effect in 2017. In this policy brief, we analyse the distributional consequences of changes in the direct tax and benefit system, and discuss the long-term implications of these policies in combination with the policy to reduce the statutory retirement age.
The Law and Justice party (Prawo i Sprawiedliwość, PiS) won an absolute majority of seats in both houses of the Polish Parliament in the parliamentary elections of October 2015. Earlier that year, Andrzej Duda of PiS was elected President of the Polish Republic. In both cases, the electoral victories came on the wave of pledges of significant financial support to families with children and to low-income households, especially pensioners. The new president pledged to cut back the pension age to the levels prior to the 2012 reform, which introduced a gradual increase from 60 and 65 to 67 for both women and men, and to nearly triple the income tax allowance. Following Duda’s victory in May 2015, PiS reiterated these pledges in the parliamentary election campaign and added the promise to increase the total level of financial support for families with children by over 140% through a nearly universal benefit called “Family 500+” and to hike the minimum wage by over 8%.
Despite a rather tight budget situation, the government went ahead with the “Family 500+” and successfully rolled it out in April 2016 (Myck et al., 2016a). The new instrument directs support of 500 PLN per child per month (110 EUR) to all second and subsequent children in the family in the age group between 0 and 17. Benefits for the first child in the family in this age group are granted conditional on overcoming an income threshold of 800 PLN (180 EUR) per person per month. Since April 2016, over 2.7 million families have received the benefit and 60% of them received the means tested support (if they have more than one child this is paid out in combination with the universal benefit).
The second key electoral pledge – to increase the tax allowance from 700 to 1,850 EUR at an estimated cost of 4.8 billion EUR – has so far been postponed (CenEA, 2015a). Increases in the allowance became a major policy issue in October 2015 when the Constitutional Tribunal ruled that maintaining its level below minimum subsistence, as it was at the time, was unconstitutional. To satisfy the Tribunal’s ruling, the allowance would have to increase to ca. 1,500 EUR at a cost of nearly 15 billion PLN (3.4 billion EUR, and about 0.8% of GDP, CenEA 2015b). Instead of a simple increase in the allowance, the government decided to implement a digressive tax allowance for 2017. This raised the value to the required minimum subsistence level for the lowest income tax payers, but since it is rapidly withdrawn as taxable income rises, the allowance will be unchanged to a large majority of taxpayers and will cost the public purse only 0.2 billion EUR (CenEA, 2016). This policy will be more than paid for by the fiscal drag given the decision to freeze all other parameters of the tax system, which will cost the taxpayers 0.5 billion EUR (Myck et al., 2016b).
The policies that directly affect household budgets will in total amount to about 5.5 billion EUR in 2017 (1.3% of GDP and 6.2% of the planned central budget expenditures) and will include also an increase in the minimum pension to benefit about 1.5 million pensioners. The cost of the “Family 500+” reform makes up the large majority of this value (5.4 billion EUR). Households from the lower income decile groups will benefit the most from this reform package, with their monthly disposable income increasing on average by 15.1% (ca. 60 EUR). High-income households from the top income decile will see their income grow on average by only 0.5% (see Figure 1). Overall, nearly all of the gains will go to families with children, with single parents gaining on average about 95 EUR and married couples with children about 84 EUR per month. Other types of families will, on average, see negligible changes in their household disposable incomes (see Figure 2). Thus, the implemented package clearly has a very progressive nature and redistributes significant resources to families with children.
Figure 1. Distributional consequences of changes in direct tax and benefit measures implemented between 2016-2017
The pension age and public finances in the years to come
The most recent major reform, legislated at the end of 2016 and which will come into effect in October 2017, represents an implementation of yet another costly electoral pledge. This policy has overturned gradual increases in the statutory retirement age, initiated by the previous government in 2012. Despite the very rapid ageing of the Polish population, the new government decided to return to the pre-2012 retirement ages of 60 and 65 for women and men, respectively. This comes at a time when, according to EUROSTAT (Eurostat, 2014), the old-age dependency ratio in Poland, i.e. the proportion of the 65+ population to the working-age population, will grow from the current 24% to 27% in 2020 and to 40% in 2040. With the defined contribution pension system, the shorter working lives resulting from this change will be reflected in significantly reduced benefits (Figure 2). For example, pension benefits of men retiring in 2020 will on average be 13.5% lower than the pre-reform value. For women that retire in 2040, the pension benefits will on average fall by 15.2%, which corresponds to a 43% lower benefit than the pre-reform value, and with consequences of the reform becoming more severe over time. The reform will also be very costly to the government budget. In 2017, it is expected to cost 1.3 billion EUR and its full effect will kick in after 2021, when the cost of the reform will exceed 3.9 billion EUR per year (Figure 2).
Figure 2. Reducing the statutory retirement age and its implications on pension benefits and public finances
Since coming to power in October 2015, the PiS government has implemented a majority of its costly electoral pledges. Direct changes in taxes and benefits will cost 5.5 billion EUR in 2017 and benefit primarily those in the lower end of the income distribution and in particular families with children. The reduced statutory retirement age will add an extra 1.3 billion EUR in 2017 and as much as 3.9 billion EUR four years later. The very generous “Family 500+” programme has significantly reduced child poverty and may have important positive long-term effects in terms of health and education for today’s beneficiaries. However, its fertility implications are still uncertain and the programme is expected to reduce the employment rate among mothers. While the government maintains that its financing is secured, it is becoming clear that maintaining the policy will not be possible without higher taxes.
The government came to power claiming that the implementation of this programme will be based on reducing tax fraud and that only a small fraction will be financed from tax increases. While it seemed likely at the time when these declarations were made, the expected major shift in the reduction of tax fraud has yet not materialised. The government have withdrawn from the pledge of reducing the VAT and from assisting those with mortgages denominated in Swiss Francs, while its income tax allowance reform was nearly thirty times less expensive compared to that announced in its electoral programme.
With a very tight budget for 2017 based on relatively optimistic assumptions, the key factors determining further realisations of the generous programme will be the rate of economic growth and related dynamics on the labour market. Developments of the labour market will also be essential for the longer-term economic success of the implemented reform package. This relates both to the future level of participation of women and to the success of extend working lives of people who will soon reach the new reduced retirement age.
- CenEA (2015a) Konsekwencje prezydenckiej propozycji podwyższenia kwoty wolnej od podatku (Consequences of the presidential proposal to raise the incoem tax allowance), CenEA press release, 3 December 2015.
- CenEA (2015b) Co z kwotą wolną od podatku po wyroku Trybunału Konstytucyjnego? (what will happen to the income tax allowance after the decision of the Constitutional Tribunal?), CenEA press release, 13 November 2015.
- CenEA (2016) Zmiany w kwocie wolnej od podatku za 800 mln rocznie (Changes in the income tax allowance at the cost of 800m per year), CenEA press release, 29 November 2016.
- EUROSTAT (2014) Eurostat – Population projections EUROPOP2013, access 21 December 2016.
- Myck, M., Kundera, M., Najsztub, M., Oczkowska, M. (2016a) 25 miliardów złotych dla rodzin z dziećmi: projekt Rodzina 500+ i możliwości modyfikacji systemu wsparcia. (25bn for families with children: plans for the Family 500+ reform and other options to modify the system of support.), CenEA Commentaries, 18 January 2016.
- Myck, M., Kundera, M., Najsztub, M., Oczkowska, M., 2016b, Zamrożony PIT i utrzymane wyższe stawki VAT – jak brak zmian w podatkach wpłynie na budżety gospodarstw domowych? (Frozen PIT and higher VAT – how lack of changes in taxees will affect househod budgets?), CenEA Commentaries, 05 October 2016.
- Council of Ministers (2016) Position of the Council of Ministers on the presidential bill proposal, Warsaw, 25 July 2016.
The 2016 budget includes measures aimed at increasing the progressivity of the Latvian income tax system. In this brief we report some exercise on the impact of these measures using the Latvian EUROMOD tax-benefit microsimulation model. We show that by their design, the reforms are aimed at a reduction in income inequality and an increase in the progressivity of the tax system. However, there are risks that the behavioural response of the tax payers will subvert the intended impact of the reforms.
Ever since it was introduced in 1994 the Latvian personal income tax has been applied at a flat rate, albeit varying over time, mitigated only by a small untaxed personal allowance. Partly as a result of this, the Latvian tax-benefit system redistributes less original income than most other EU countries. Is this all about to change? The 2016 budget currently being debated in the Parliament contains two proposals aimed at introducing more progressivity in the personal income tax. These are the introduction of a “solidarity tax” aimed at high earners and the introduction of an earnings differentiated non-taxable allowance. The stated aims of these measures are to reduce inequality and help low wage-earners.
Description of the Reforms
The solidarity tax foresees that income above 48,600 EUR per year will be taxed at a rate of 10.5% (employee’s part), plus 23.59% (employer’s part). The new tax will affect a very small share of wage earners. According to Finance ministry’s estimate, this tax will affect 4.7 thousand persons, whose income in 2015 exceeded this threshold, or 0.59% of all employed individuals (Finance Ministry, 2015).
Differentiated Non-Taxable Personal Allowance
The differentiated non-taxable personal allowance will be introduced gradually between 2016 and 2020. The basic idea is to make the allowance dependent on income: individuals receiving income below a certain threshold are eligible for the maximum possible allowance, then the allowance gradually declines with income until it is zero. The system will be introduced gradually in the sense that the minimum allowance will not reach zero until 2020 – it will be gradually reduced from 85 EUR in 2016 to 0 EUR in 2020.
The way the system will be implemented foresees that during a fiscal year, all individuals will be taxed applying the minimum non-taxable allowance (e.g., 85 EUR in 2016). At the beginning of the next year, people eligible for a higher tax allowance will have the opportunity to apply for a tax refund, by making an income declaration, and to get the overpaid tax back.
Simulations of Reforms: Inequality
Below we present simulation results from EUROMOD, which is an EU-wide tax-benefit microsimulation model (for more details see Jara and Leventi, 2014). The results show the first-round effect of the simulated policies, i.e., they show the pure effect of the proposed reforms abstracting from any behavioural responses that these reforms might induce. We simulate the effect of five reform scenarios: two scenarios of differentiated non-taxable allowance (one scenario reflects the system that is planned to be introduced in 2016, the second scenario represents the system that is planned to be introduced in 2020), one scenario that simulates introduction of the solidarity tax, and two scenarios that combine the solidarity tax with the new non-taxable allowances. We compare these reforms with the baseline system, which describes the tax-benefit rules that are in place in 2015.
It is important to note that we assume in the simulations that everyone who is eligible for a tax refund under the new non-taxable allowance rules does in fact apply for the refund, which means that we estimate the maximum possible effect from the introduction of the higher tax allowances.
Table 1 summarizes the effect of the proposed reforms on income inequality as measured by the Gini coefficient. All the proposed reforms reduce income inequality, but the solidarity tax achieves higher equality by reducing incomes in the top decile. The non-taxable allowance mainly affects people in the middle of the income distribution, as the bottom deciles contain proportionally fewer employed individuals, while in the top deciles the allowance, which is set in absolute terms, makes a smaller share of the income – hence, a weaker effect. Pensioners, who mainly belong to the lower deciles of the income distribution, do not gain from a higher allowance, because of a special taxation regime for pensions that already provides for a higher personal allowance. All major benefits (unemployment benefit, social assistance, child-related benefits) are not subject to personal income tax, hence benefit recipients also do not gain from the proposed changes (see Figure 1).
Table 1. Gini Coefficient Associated with the Reforms
|Baseline||ST*||2016 allowance||2020 allowance||ST + 2016 allowance||ST + 2020 allowance|
Source: authors’ calculations using EUROMOD
Note: ST – solidarity tax
Figure 1. Deviation of Equivalised Disposable Income from the Baseline Scenario, %
Figure 1 also shows that the losers from the solidarity tax are in the highest decile, though it should be borne in mind that enterprises are also losers because they now have to pay part of the solidarity tax. The solidarity tax generates no direct gainers.
Impact on Progressivity
The progressivity of a tax or system is typically measured by the Kakwani index. The Kakwani index (Kakwani, 1977) can vary between −1 and 1 and the larger the index, the more progressive is the tax. A positive index indicates that the tax is progressive and a negative index indicates it is regressive. Table 2 shows the calculated Kakwani index for all major direct taxes (which include personal income tax, social contributions and the newly introduced solidarity tax) and separately for personal income tax (PIT) for each of the postulated scenarios. The results suggest that all of the proposed reforms increase the progressivity of the tax system.
Table 2. The Kakwani Index for the Six Scenarios
|Baseline||ST*||2016 allowance||2020 allowance||ST + 2016 allowance||ST + 2020 allowance|
|All income taxes*||0.034||0.040||0.048||0.058||0.054||0.064|
Source: authors’ calculations using EUROMOD
Note: ST – solidarity tax; income taxes include personal income tax, social contributions and the newly introduced solidarity tax
Qualifications and Risks
The above results capture the so-called first round impact of the tax changes. In practice people will react to the changed incentives by changing behaviour and thereby changing the impacts. For example, the higher net reward for working in low wage jobs may increase the supply of workers willing to work in such jobs thereby possibly having a bigger positive effect on the incomes of low income households than implied by the simulations.
Perhaps more significant is the potential effect of the solidarity tax on the behaviour of high earners and of the enterprises that employ them. This effect is captured by the concept of the elasticity of taxable income – defined as the change in taxable income in response to a change in the marginal tax rate. The taxable income elasticity concept takes into account all the behavioural aspects of the taxpayer in response to a change in the tax rate. As well as labour supply responses it includes other responses e.g. switching the form in which income is received as well as simple tax evasion (Saez et al., 2012). It is the switching of the form in which income is received, away from wage income towards other less-taxed forms of income that can be expected here. Thus according to an internal Latvian Employers Confederation employer survey, if the solidarity tax is implemented one third of employers will consider using legal tax optimization tools such as dividends or the microenterprise tax to avoid paying the tax. Here, employers are important as well as employees, because employers will pay the larger share of the tax. If this happens on a significant scale (high elasticity of taxable income) then the intention of the solidarity tax will be subverted.
There are also risks with the differentiated personal allowance. If the burden of annual reporting of income is too high then many may simply not do it and suffer the loss of income or find a way of recouping through shadow earnings.
The Latvian authorities should be applauded for grasping the nettle of progressive taxation but perhaps only with one hand for the way they have chosen to do it. Thus, the solidarity tax creates an incentive for both employers and employees to find ways of avoiding it and find they surely will. A tax accountant once said of the 80% supertax applied to high earnings in pre-Thatcher UK that it was a ‘voluntary tax’. This is also the likely fate of Latvia’s solidarity tax.
The differentiated personal allowance will clearly benefit low earners, if they claim it. In fact it will also benefit people earning well over the average wage. But will the low earners claim? Very few people in Latvia have ever filed an income declaration and we fear that many low earners will not do so now.
Thus at the top end progressivity is likely to be largely avoided and at the bottom end may not be fully claimed.
- Finance Ministry (2015). “Solidaritātes nodokli maksās tikai personas ar algu virs 48 600 eiro gadā,” available at http://www.fm.gov.lv/lv/aktualitates/jaunumi/nodokli/51253-solidaritates-nodokli-maksas-tikai-personas-ar-algu-virs-48-600-eiro-gada
- Kakwani, Nanak C. (1977). “Measurement of Tax Progressivity: An International Comparison”. Economic Journal 87 (345): 71–80
- Jara, X. and Leventi, C. (2014). “Baseline results from the EU27 EUROMOD (2009-2013),” EUROMOD Working Papers EM18/14, EUROMOD at the Institute for Social and Economic Research.
- Saez, E., J. Slemrod, and S. H. Giertz, (2012). “The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review.” Journal of Economic Literature, 50(1): 3-50