Tax Me, But Spend Wisely? Sources of Public Finance and. Government Accountability

Tax Me, But Spend Wisely? Sources of Public Finance and Government Accountability Lucie Gadenne June 2015 Abstract Existing evidence suggests that extra grant revenues lead to little improvements in public
of 18
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Tax Me, But Spend Wisely? Sources of Public Finance and Government Accountability Lucie Gadenne June 2015 Abstract Existing evidence suggests that extra grant revenues lead to little improvements in public services in developing countries - but would governments spend tax revenues differently? This paper considers a program that invests in the tax capacity of Brazilian municipalities. Using variations in the timing of program uptake I find that it raises local tax revenues and that the increase in taxes is used to improve both the quantity and quality of municipal education infrastructure. In contrast increases in grants over which municipalities have the same discretion as over taxes have no impact on any measure of local public infrastructure. 1 Introduction The idea that increases in tax revenues go hand in hand with more accountable and efficient public spending is at the heart of interpretations of the emergence of representative governments in the West (North and Weingast, 1989, Lindert, 2003). Whether increasing the capacity to tax of governments in today s developing countries would have a similar effect is an open question. It is also an important one as more public investments in infrastructure are necessary to further the economic development of these countries (Duflo, 2011). University College London and Institute for Fiscal Studies, 30 Gordon Street, WC1H 0AX London, United Kingdom. I would like to thank Luc Behaghel, Tim Besley, Richard Blundell, Clement de Chaisemartin, Raj Chetty, Denis Cogneau, Claudio Ferraz, Antonio Guarino, Walker Hanlon, Wojciech Kopczuk, Karen Macours, Magne Mogstad, Thomas Piketty, Imran Rasul, Monica Singhal, Joel Slemrod, Ernesto Stein, Eric Verhoogen and Ekaterina Zhuravskaya for their helpful comments as well as Fernanda Brollo, Stephan Litschig, Tommaso Nannicini and Yves Zamboni for sharing their data, Marcelo Fernandes and Rogerio Goulart for their help in understanding the PMAT program. I acknowledge financial support from the AXA Research Fund and the CEPREMAP. The large literature looking at the consequences of increases in government revenues in developing countries is disappointing in this respect: it typically finds that they have little impact on public health, education or social infrastructure and are often wasted or diverted. 1 A common trait of studies in this literature however is that they consider variations in non-tax revenues, possibly because variations in tax revenues that are unrelated to other determinants of public spending are particularly hard to come by. This paper first asks whether increases in governments capacity to tax have a positive impact on the provision of public infrastructure in the context of Brazilian municipalities. To do this I study a program that helps municipalities increase their tax revenues by subsidizing investments in local tax administrations. I consider whether the program increased local tax revenues and whether the extra revenue generated were spent on improving local public services. Participation to the tax-capacity program is voluntary but the particular timing of its implementation enables me to estimate its causal impact on outcomes: local governments decide when to apply to the program but the date at which they start it is determined by constraints faced by the supplier of the program. This makes it possible to separately identify the impact of the program from a potential selection effect. I then consider whether local governments spend these tax revenues differently from non-tax (transfer) revenues. Variations in non-tax revenues come from a rule determining how much federal transfers municipalities receive, as used by Litschig and Morrison (2013) and Brollo et al. (2013). The rule specifies that the transfer revenues municipalities receive increase discontinuously with local population size at 14 cutoffs, so identification of the impact of transfer revenues comes from municipalities that cross the threshold over time. I compare how governments spend increases in tax and transfer revenues using a 14 years panel dataset on municipal revenues and expenditure outcomes, primarily the quality and quantity of locally funded public education infrastructure. Brazilian local governments are a good context in which to ask whether governments spend tax revenues differently from non-tax revenues for several reasons. First, municipalities control a significant share of public revenues (roughly one-fifth) and are responsible for key public expenditures. Their main spending responsibility, and the main expenditure outcome I con- 1 For example, Reinikka and Svensson (2005) show that schools in Uganda receive only a small share of funds allocated to them by the central government, Olken (2007) estimates that more than 20% of grants that local governments in Indonesia receive to finance road projects are diverted, and Svensson (2000) finds some evidence that aid increases corruption in politically divided countries. 2 sider, is education, an area in which Brazil s performance is generally considered disappointing compared to countries at similar levels of development (Ferraz et al., 2012). Municipalities are in charge of primary education and shoulder much of the blame for this; there is both scope and need for more local investments in education. Second, local governments have the same discretion over the allocation of the transfer revenues considered here as over their own tax revenues, so there are no legal or administrative reasons for the two being spent differently. Third, there is evidence that Brazilian local governments do not use increases in non-tax revenues to improve local infrastructure but instead waste or divert it (Caselli and Michaels, 2013, Ferraz and Monteiro, 2010, Brollo et al., 2013). 2 Asking whether tax revenues are similarly wasted or diverted is particularly relevant in this context. I find that the program is successful in raising local tax revenues: a 1 Real investment in tax capacity leads to a annual increase in tax revenues of roughly 1 Real per year after 5 years. Moreover, the increase in tax revenues generated by the program leads to a 5 to 6% increase in the quantity of municipal education infrastructure and an improvement in an index of the quality of the infrastructure of one-tenth of a standard deviation. An increase in transfer revenues of the same size has no impact on education infrastructure. I then consider alternative uses of public revenue. Neither tax nor transfer revenues have a significantly positive impact on municipal health infrastructure but the difference between the two estimates goes in the same direction as for education. Some evidence on what transfer revenues are spent on is found in Brollo et al. (2013) who show that they lead to an increase in corruption. In contrast I find evidence suggesting that tax revenues have no impact on corruption. To interpret these results as evidence that governments spend revenues from different sources differently one must rely on stronger assumptions than those required to interpret the estimates of the impacts of both tax and transfer revenues as causal. These estimates are obtained on different groups of municipalities so we must assume that municipalities affected by both policies do not have different marginal propensities to spend on education, health and corruption out of all types of public revenues. Whilst this assumption cannot be tested the particular design of the transfer allocation rule - it creates 14 different points in the distribution of municipalities at which the local impact of transfer revenues can be estimated - allows me to consider whether the marginal propensity to spend transfer revenues varies with observable municipal characteristics. 2 An important exception is Litschig and Morrison (2013) who find that these same grants also lead to better education outcomes in the 1980s. I discuss their results in detail below. 3 I find no evidence of such variation. In particular I can rule out that municipalities which look extremely similar to those that enrolled in the tax capacity program spend their transfer revenues differently. Several models of public resources allocation could explain this result. I find some evidence that the difference between how tax and transfer revenues are spent is smaller in municipalities where there is a local radio station potentially informing citizens about public budgets. This is in line with principal-agent models of public finance in which asymmetries of information allow politicians to capture more rents (Besley and Smart, 2007) if we assume that citizens are better informed about increases in taxes than increases in transfers. I discuss and test other mechanisms that could lead to tax revenues being spent differently from transfer revenues and show that the results are not due to different characteristics of the tax and transfer variations studied here that are not necessarily related to their source (predictability, size and sign of the variations in revenues). This paper contributes to the literature on public finance in developing countries in two ways. First, by evaluating the impact of a tax capacity program I present the first estimates of the returns to investment in tax capacity. Second, it is to the best of my knowledge the first to consider the impact of tax revenues on government expenditure outcomes. The idea that the growth of states capacity to tax is an important covariate of economic development (Besley and Persson, 2009) motivates a growing literature that studies the determinants of tax compliance in developing countries. 3 My results suggest that part of the correlation between tax capacity and economic development can be interpreted as causal as I show that governments that tax more will also invest more in human capital infrastructure. The results are closely related to the literature which considers whether the way governments are financed affects their behavior. Fisman and Gatti (2002) establish a positive relationship between the proportion of US states revenues derived from federal transfers and the number of convictions of public employees for abuse of public office. Similarly Zhuravskaya (2000) provides evidence that outcomes affected by public policy improve when Russian cities keep more of their tax revenues. 4 I build on these previous findings by using variations in tax and non-tax revenue that stem from clearly identified sources and considering variations in publicly provided infrastructure that are directly controlled by governments. There is also 3 See for example de Paula and Scheinkman (2010), Olken and Singhal (2011), Carrillo et al. (2011), Kumler et al. (2013), Pomeranz (2013), Best et al. (2013), Naritomi (2013), Khan et al. (2014), Cagé and Gadenne (2014). 4 See also Jin et al. (2005) and Fan et al. (2009). 4 a large literature devoted to explaining the fly-paper effect, the fact that a dollar received by a community in the form of a grant to its government results in greater public spending than a dollar increase in community (private) income. 5 The results presented here suggest that increases in private income could nevertheless improve public infrastructure more than increases in grant income if they lead to higher tax revenues. This paper speaks more generally to the larger literature on the political economy of public good provision (see Banerjee et al. (2013) and Olken and Pande (2012) for recent reviews). I focus on the impact on public good provision of one institutional characteristic - government s capacity to tax - which has so far not been studied. 6 Finally, these findings also contribute to debates on how to finance development. The idea that aid revenues may not be spent as well as tax revenues has long been discussed by policy practitioners and researchers alike (OECD, 2010a, Besley and Persson, 2011, Deaton, 2013) but technical aid on revenue-raising management has always been the poor cousin of official development aid (OECD, 2010b). This paper shows that a resource mobilization program in place in Brazil for nearly two decades has been successful in providing long term sources of funds to local governments. It suggests that technical help in tax capacity building may lead to an increase in government resources which is more conducive to public investments in human capital than traditional budget-support development aid. Relatedly this paper speaks to debates regarding the optimal form of decentralization in developing countries (see Gadenne and Singhal (2014) for a review) by considering whether revenue decentralization - increasing local government s capacity to tax - affects public delivery outcomes for a given level of administrative and expenditure decentralization. The paper is organized as follows. Section 2 presents the context of study, the tax and transfer policies of interest and the data used. Section 3 provides a conceptual framework to formalize the hypotheses of interest and section 4 describes the empirical strategy I use to test them. Section 5 presents the results, section 6 discusses and tests possible explanations for the results. 5 See for example Knight (2002), Singhal (2008), Dahlby (2011) for recent work on the subject. 6 The results are also consistent with the literature on the natural resource curse which finds that governments revenues from the exploitation of natural resources are typically are wasted or diverted (Van der Ploeg, 2011). One explanation for this empirical regularity is that resource-rich countries have little need to levy taxes and therefore respond to their citizens demands. 5 2 Context and data 2.1 Local expenditure responsibilities The Brazilian constitution devolves substantial expenditure and revenue raising responsibilities to the country s more than 5,000 local governments. Mayors and local councils, elected every 4 years, are in charge of allocating one-fifth of all public spending. This paper considers different types of local public expenditure variables which are all inputs in the production of human capital. I focus on inputs not human capital outcomes primarily because of data limitations: there is no measure of outcomes at the local level measured in a consistent way over the years 1998 to Inputs also react faster to local policy choices than outcomes; this choice maximizes the probability that we will see an impact of local revenues on outcomes. The main measure of public expenditure outcomes I consider is municipal education infrastructure. Education is the largest local budget item (a third on average) and local governments are in charge of primary schools: they provide infrastructure, school lunches and transportation and hire and pay teachers. Physical school infrastructure is the type of local education input that is the most likely to be under-funded: municipalities receive federal grants earmarked for expenditures on school staff, school lunches or school transport but not for physical teaching infrastructure. 8 There is ample anecdotal evidence that the supply of municipal education infrastructure has not kept up with the increase in demand over the past two decades. 9 I therefore focus on physical school infrastructure as the type of input that is the most likely to be affected by changes in non-earmarked revenues, but also discuss results for teachers. I use panel data on the quantity and quality of municipal education infrastructure from the annual school census conducted by the Ministry of Education. I consider the number of classrooms in use in municipal schools per thousand school-age inhabitants to measure the quantity of municipal education infrastructure available at the lowest level of disaggregation possible. I combine the eight variables related to the quality of the infrastructure that are measured consistently over the period (number of municipal schools with computers, with internet, with a sports facility, a library, television/video equipment and connected to the sewage and electricity 7 Data on school dropouts is available annually only until Data on other education outcomes is only available in Census years, ie once per decade. 8 60% of the largest education grant, FUNDEB, must fund teacher s salaries. The PNAE grant funds school lunches, the PNATE grant school transportation. 9 A recent PISA study for example argues that lack of infrastructure is the main reason for one of the major challenges of primary education in Brazil (OECD, 2011). It further shows that municipalities that successfully improved local education outcomes often did so by investing in new school infrastructure. 6 systems) using principal component analysis to construct a quality index. I turn to two other expenditure outcomes to complement the results on education: health infrastructure and corruption. Health is the second largest municipal budget item comprising just under a quarter of local expenditures on average. Local governments share responsibility for most of the infrastructure of primary and preventive health units through the Family Health Program with state governments, so health infrastructure could also be affected by changes in local revenues. Data on the number of municipal health units come from a census of health facilities conducted in 1999, 2002, 2005 and Health and education together account for 60% of municipalities budget. Data on alternative uses of funds (such as debt reduction, local police, municipality sewage light and transport systems) is not available. Information on proxies for municipal corruption levels is available since the start of a federal anti-corruption program in Since then every six months local governments are randomly chosen through a public lottery to be audited by staff of the independent audit agency Controladio-Geral da União (CGU). They audit the use of earmarked grants received by municipalities by collecting administrative documents, interviewing citizens and conducting random checks in municipal agencies. Ferraz and Finan (2011a) estimate using the audit reports that 8% of audited revenues were diverted in the period Several teams have coded the reports for different time periods and samples of municipalities; I consider both the indexes compiled by Brollo et al. (2013) for the 925 municipalities with less than 50,000 inhabitants audited over the period and those constructed by Litschig and Zamboni (2012) for the 862 municipalities audited between 2003 and The corruption dataset is a repeated cross-section of municipalities Local public revenues The tax policy A. The tax capacity (PMAT) program Brazilian local governments are in charge of collecting and setting the rates of two main local taxes, a service tax and an urban property tax. Local tax revenues represent 13% of total tax 10 Allocating a date to the audit data is complicated. Auditors are typically supposed to audit the use of federal grants over the last two-three years but sometimes report irregularities that occurred five years ago. Moreover the date at which the irregularity occurred is often not specified in the reports - possibly because it can be hard to pin down. In my main specification I say that an irregularity measure corresponds to the year of the lottery if the audit took place in June of that year or later, I also consider what happens when an irregularity is allocated to a date one or two years prior to the lottery as a robustness check. 7 revenues on average, roughly 2% of GDP. Anecdotal evidence suggests local administrations have little capacity to enforce tax payments. Municipal staff have outdated tax registers, little institutional memory and weak methods to accurately assess tax liabilities; high costs of understanding and paying taxes combined with low penalties of getting caught push many citizens into non-compliance. Some local officials have publicly admitted to tolerating a situation of permanent tax amnesty where tax arr
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!