CALCULATING THE OPTIMAL SMALL BUSINESS EXEMPTION THRESHOLD FOR A U.S. VAT Edith Brashares, Matthew Knittel, Gerald Silverstein, and Alexander Yuskavage Most countries with a VAT have a tax exemption for small businesses that have taxable sales below a certain threshold. The exemption reduces both the compliance costs for small businesses and the administrative costs for the government. We use a static analysis of the U.S. economy to determine the optimal level of taxable sales for this threshold and analyze the potential behavioral response by businesses. For a 10 percent VAT rate, we find that the optimal level for the threshold in the United States is $200,000, and that behavioral responses may be significant close to the threshold. Keywords:

VAT, small business, threshold

JEL codes:

H21, H25

November 5, 2013 Revised: May 30, 2014 Published in June 2014 National Tax Journal

_____________________________________________________________________________ Edith Brashares: Office of Tax Analysis, U.S. Department of Treasury, Washington, D.C. [email protected] Matthew Knittel: Director, Independent Fiscal Office, Harrisburg, PA [email protected] Gerald Silverstein: Contractor, Office of Technical Assistance, U.S. Department of Treasury, Washington, D.C. [email protected] Alexander Yuskavage: Office of Tax Analysis, U.S. Department of Treasury, Washington, D.C. [email protected]

Electronic copy available at: http://ssrn.com/abstract=2443979

I. INTRODUCTION The level of the small business exemption threshold is one of the most important decisions that policymakers must make when adopting a value-added tax (VAT). Businesses that fall below the small business exemption threshold (hereafter referred to as the “threshold”) are not required to collect VAT, drastically affecting the burden of the tax. In this paper we use a database of tax returns developed by Knittel et al. (2011) and the theoretical framework developed by Michael Keen and Jack Mintz (2004) to determine the optimal threshold for a VAT in the United States. We believe this is the first published measurement of the optimal threshold for the United States based on individual firm data. The primary argument for having a threshold is that collection costs (the private costs of collecting and remitting the tax and the public costs of administering the tax) can be reduced without giving up significant tax revenue. Small businesses, defined as those businesses under the threshold, are typically not required to issue VAT invoices and file a VAT return. This significantly reduces the number of returns filed and thus lowers the administrative costs of the VAT. In addition, business compliance costs are generally considered to be regressive with respect to tax revenue, which means that the firms which must comply with VAT recordkeeping are those most able to handle the additional costs. Finally, without a threshold many small businesses receive refunds or pay little VAT so that exempting small businesses may even raise net revenue. For example, businesses temporarily in a loss position or startup businesses with low taxable sales but significant purchases may have negative value-added and be owed a refund. The theoretical framework developed by Keen and Mintz characterizes the optimal threshold by accounting for the trade-off between tax revenues and collection costs. While

2 Electronic copy available at: http://ssrn.com/abstract=2443979

estimates of the optimal threshold in other countries can rely on data for current VAT filers, the United States does not have a VAT, so we instead estimate the VAT base for each business from income tax filings. We assume that all business types may potentially be required to file a VAT return. We use tax data to estimate taxable sales, purchases, compliance costs, and administrative costs for each business. Exports are imputed from the input-output tables. This allows us to empirically estimate, using a static model of the U.S. economy, the optimal level of the threshold in the United States. We also use these data to characterize the extent to which we expect firms to adjust their sales to fall below the threshold and to voluntarily register to pay VAT when they are exempt. Our major finding is that the optimal threshold for a 10 percent VAT in the United States is approximately $200,000, which is the 90th percentile of the size distribution of businesses. Approximately 43 million businesses are below the threshold, and of those, 12.6 million have negative value added.1 The net value added below the threshold is -$145 billion, so in the absence of a significant response by firms, setting the threshold at $200,000 would increase net revenues for the government. The total reduction in administrative and compliance costs resulting from a $200,000 threshold is $2.6 billion and $25.5 billion respectively. We also find that approximately 0.9 million businesses have an incentive to reduce their taxable sales to fall under the threshold, while at least eight million businesses below the threshold have an incentive to voluntarily register for the VAT.

1

Negative value-added occurs when creditable purchases exceed taxable sales. This most commonly occurs when businesses are temporarily in a loss position, are primarily exporters, have significant capital purchases, are a startup company, and overstate purchases or understate sales. While some of these are transitory phenomenon, negative value-added is not uncommon and would usually generate a refund of taxes.

3

II. BRIEF REVIEW OF HOW A VAT WORKS For the standard destination-based credit-invoice VAT, a business calculates its tax liability as its taxable sales times the tax rate, less its input tax credits, which can be calculated as purchases of inputs (including purchases of capital goods but excluding labor costs such as wages and salaries) times the tax rate.2 The VAT that is paid on the purchase of inputs to the production of exported goods and services is also credited. The business remits the net amount of tax collected on its sales less the input tax credits. As described in the literature (U.S. Treasury 1984; McLure, 1987; and Congressional Budget office, 1992), the credit-invoice VAT has the advantage that multiple tax rates, including a zero rate, are easier to handle correctly than with the subtraction method. Almost all countries use a credit-invoice VAT so it is likely that, in practice, the United States would also adopt a credit-invoice VAT. However, because we cannot distinguish sales to registered businesses from sales to consumers and unregistered businesses, in this paper we measure the VAT base for each business as taxable sales less purchases, including those from exempt small businesses. For the credit-invoice method VAT, there are two ways in which the value added for goods and services can be excluded from the tax base: zero rating and exemption.3 Zero rating completely removes tax from the good or service by taxing the sales at a zero rate while still allowing a credit for the tax paid on inputs. Exports are zero rated under a destination-based credit-invoice VAT. Exemption is similar to zero-rating in that the sales price of the good or service has a zero VAT rate. However, some VAT is implicitly paid on exempted goods and services since producers cannot claim a credit for the VAT they have paid on their inputs. A key

2

We use the terms “taxable sales” or “sales” since these are more familiar terms for U.S. readers. This is identical to “supply” in VAT terminology. 3 For examples and more discussion see McLure (1987) and Congressional Budget Office (1992).

4

difference for collection cost purposes between zero rating and exemption is that exempt businesses do not have to file VAT returns while zero-rated businesses do have to file. In addition, businesses selling both taxed and exempt sales must apportion their input tax credits while businesses selling both taxed and zero rated sales do not need to apportion credits. The impact of zero rating and exemption on tax revenues depends upon whether the sale is to another business as an intermediate purchase or to a consumer as a final sale. When a zero rated good or service is sold to another business and thus taxed at a subsequent stage of the production process, zero rating will not impact VAT revenues. In contrast, when an exempt good or service is sold to another business, the exemption will increase VAT revenue as the loss of input tax credits results in the collection of taxes on both all prior value added and all prior taxes at the time that the good or service is sold to the next business. This multiple application of tax is known as cascading. When the sale of the good or service is to the final consumer or nonregistered business, zero rating will reduce tax revenues to zero, whereas exemption will simply reduce the effective tax rate on the final sale. The small business threshold provides an exemption for those sales made by firms below the threshold. High collection costs relative to the tax revenues collected cause most countries to exempt small businesses from the VAT. Under the credit-invoice method, small businesses with taxable sales below a threshold are typically not required to file a return. Prices for goods sold by such small businesses incorporate the VAT only to the extent that VAT is included in the price of inputs purchased by untaxed businesses from VAT-paying businesses. If the exempt business sells to a taxable business, then the purchasing business does not receive input credits for VAT paid at the prior stage of production. Given this effect, if there is optional registration, small businesses selling predominantly to taxable businesses are more likely to voluntarily register for

5

the VAT. While the tax base for a VAT is often modeled as total domestic consumption, the base may in fact be different.4 Hard-to-tax goods and services, such as financial intermediation where payment is the difference between the interest received and paid, are often exempted. Similarly, goods and services with public good characteristics provided by governments and non-profit organizations are often zero rated or exempted. Often, the entire sector producing such goods is zero rated or exempted. However, concern about a level playing field can result in the taxation of certain government and non-profit activities that compete with the private sector. Finally, taxing the value of owner occupied residential housing services would bring many taxpayers into the VAT system that would not otherwise register and collect the VAT. In order to avoid that result many countries tax only the sale of new owner-occupied housing and renovations. Our modeling approach allows relatively broad tax base adjustments. We do not include governments and non-profit organizations as taxpayers. We include financial services where there are separate charges, such as bank deposit fees and property and casualty insurance, while excluding most margin-based financial services, such as deposit and lending activities and life insurance. Not including government and non-profit organizations as taxpayers and excluding bank margins and life insurance from the tax base amounts to treating such sales as exempt from the VAT. The service flows from tenant-occupied dwellings are taxed in our model regardless of ownership while the existing stock of owner occupied housing falls entirely outside our tax base. However, most new owner occupied housing, renovations and repairs are taxed in our model as a result of the inclusion of the construction industry in our database.

4

See Pike (2012) for a discussion of typical VAT exemptions including exemptions for small businesses.

6

III. VAT THRESHOLDS IN OTHER COUNTRIES Most of the recent proposals for a VAT in the United States do not discuss a small business exemption, let alone the more complex treatment of small businesses in many countries.5 In particular, the small business registration threshold may differ from the threshold for collection of VAT. Smaller businesses that pay VAT may have options to pay using a simplified system, to use cash accounting, and to file and pay less frequently (Organization for Economic Co-operation and Development (OECD), 2009). The simplified schemes tend to base VAT remittance on sales alone, and may have different tax rates and thresholds that vary by business sector. Furthermore, countries vary in terms of the minimum registration period. Approximately one-half of the OECD countries require businesses to register for at least one year to five years (OECD, 2008). This limits businesses from registering just to receive refunds, such as for the payment period when they purchase equipment.6 [Table 1 about here] As shown in Table 1, we can divide the OECD member countries into three groups based on thresholds.7  The six countries with the highest thresholds all have thresholds that exceed $63,000,  The 18 countries have thresholds that vary from $2,000 to $45,000, and  Five countries have no general threshold. 8 5

For example, the Debt Reduction Task Force (2010) proposal includes a credit-invoice VAT called the deficit reduction sales tax while U.S. Representative Ryan’s (2010) original plan included a subtraction VAT, but neither discusses a small business exemption threshold. 6 Refunds are payments from the government to businesses with negative value added and are discussed in more detail in Section VI.C.2. 7 Conversion to U.S. dollars is done using the OECD’s purchasing power parities (PPP) rates in order for the numbers to match those in (OECD, 2008). Unlike currency exchange rates that reflect short term events, the PPP reflects longer run relationships. The thresholds presented in the table are not necessarily the most recent thresholds for each country, but rather the threshold as of January 2007. This is done for consistency since our U.S. data are also from 2007.

7

Even if a business is below the threshold, most countries allow businesses the option of registering and collecting the VAT. This enables small businesses that primarily sell to other businesses to pay VAT so that there is no tax cascading. Surprisingly, there does not appear to be a clear correlation between the threshold and the percentage of voluntary registrants. An OECD study (2009) noted that the percentage of voluntary registrants did not increase with the threshold for a sample of countries.9 In addition to these general thresholds, other thresholds may apply for certain sectors. In particular, the taxable sales threshold for service suppliers in France ($30,963), Ireland ($31,537), and Greece ($5,734) is lower than for non-service providers. Similarly, the non-profit and charitable sector has higher thresholds in Australia ($70,922), Switzerland ($88,325), Canada ($41,667), and Norway ($15,748). Special thresholds apply for the sale of artwork (Denmark), lawyers, writers, and artists (France), and certain small retailers (Portugal) (OECD, 2011). Note that this special treatment is in addition to reduced rates or exemption for specific goods and services. [Figure 1 about here] However, the VAT thresholds in Table 1 by themselves do not provide much information about how many businesses are exempted or the revenue forgone by having a specific threshold. Estimating the effect of the threshold requires further information on the distribution of businesses, which is often difficult to obtain. Based on information from the United Kingdom (UK) Office of National Statistics (2010), Figure 1 shows that there were 1.9 million VAT

8

These thresholds cannot be directly compared to the U.S. threshold calculated in this paper. For instance, the statutory VAT rate in these countries ranges from 5 percent to 25 percent, while we will consider only a 10 percent rate. 9 These include the United Kingdom (36 percent), Slovak Republic (26 percent), and Australia (39 percent).

8

payers in 2010 when the threshold was generally $104,345 (£67,000). In addition, there are approximately 2.4 million unregistered enterprises that consist of self-employed individuals working alone or in partnership.10 Thus, in total, there are approximately 4.5 million potential VAT paying businesses in the UK if an expansive definition of a business is adopted. Of those 4.5 million businesses approximately 2.4 million were unregistered in the lowest taxable sales category of less than $77,859 (£50,000) per year in Figure 1 (United Kingdom Department for Business Innovation & Skills, 2011). As described in the next section, information on the distribution of businesses by taxable sales provides a framework for evaluating a particular threshold as optimal. IV. DETERMINING THE OPTIMAL SMALL BUSINESS THRESHOLD Keen and Mintz recognize that the collection of VAT revenues is not costless since the government must bear administrative costs and businesses must bear compliance costs. These collection costs tend to be regressive, with costs (relative to sales) being higher for smaller businesses. In addition, collection costs often exceed the benefits of VAT collection from very small businesses. Intuitively, the VAT threshold should be set where the incremental benefit of tax revenue just offsets incremental collection costs. In their simple model, Keen and Mintz assume that collecting VAT revenues from a given business costs authorities A in administrative costs. Similarly, each business faces compliance costs of Γ. Both these costs, A and Γ, are assumed to be fixed and independent of taxable sales. They assign δ >1 to be the marginal cost of public funds (MCPF); this assumption states that a dollar in the hands of the government has greater social value than a dollar in the 10

Because most countries do not make this data publicly available, it is difficult to determine how specific this pattern is to the United Kingdom. However, the graphs in Appendix D kindly provided to us by Rauhanen and Venetoklis present the distribution of businesses that register for the VAT as well as those that do not surrounding the VAT threshold of 8,500 €.

9

hands of private individuals. Potential VAT revenues depend on the tax rate, τ, as well as ν(z), the ratio of value added to sales at a given level of sales z. After setting up the policymaker’s problem to choose the threshold z* such that it maximizes social welfare net of collections costs, the first-order conditions can be rearranged to generate the decision rule (1)

∗

∗

.

The optimal threshold is when the additional administrative and compliance costs in the numerator balance the social value of the additional VAT revenue gained.11 To illustrate use of this rule, Keen and Mintz use the estimates by Cnossen (1994) of $100 per business in administrative costs and $500 per business in compliance costs. Assuming a tax rate of 15 percent, a constant ratio of value added to sales of 35 percent, and a marginal cost of public funds of 1.3, the rule generates an optimal threshold of $40,000. Keen and Mintz extend their results in two ways. First, they calculate optimal VAT thresholds in the presence of an existing sales tax. Second, they model the firm’s decision of whether or not to shrink in order to avoid the VAT.12 While Keen and Mintz simulate results for their model using a Cobb-Douglas production function calibrated to Canadian data, we instead examine how to apply their rule using actual micro-level data. Although Keen and Mintz predict there will likely be significant behavioral

11

Zee (2005) has an alternative model that focuses on consumption inefficiencies rather than production inefficiencies. As a result, the optimal threshold accounts for the elasticity of substitution in consumption, with the Keen and Mintz model being a special case. Zee further finds that the Keen and Mintz model may overstate the optimal threshold. 12 Kim (2005) extends the Keen and Mintz model to account for tax evasion by VAT businesses.

10

responses by firms, we start by considering a static analysis of VAT thresholds in the United States and then expand the examination to include some behavioral analysis.13 Since we use observational data in our analysis, we do not have a continuous functional form for ν(z). Rather, we have an empirical distribution where the relationship between value added and taxable sales is not straightforward. Even if we identify a candidate business that satisfies the condition for z*, it does not follow that all businesses with a higher sales will also satisfy that condition.14 Therefore we modify the decision rule described by Keen and Mintz to instead calculate, for each level s of taxable sales, the statistic (2)





,

where VAs is the total value added for firms within that level of taxable sales, As is the administrative cost for all firms at that level of sales, and Γs is the compliance cost. We then pick our threshold (Zs) to maximize the sum of these calculations for all businesses with z>Zs. We find the specific value of this threshold, rounded for discursive purposes, by performing a grid search over each possible one-percentile band of taxable sales.15 This leads to a threshold that closely approximates one where marginal tax revenues just exceed the marginal administrative and compliance costs adjusted for the MCPF. This method is quite flexible; for example, it allows us to examine administrative and compliance cost functions that depend on business characteristics, as well as the ability to use actual firm micro-data.

13

As shown in Table 2 on p. 573 of Keen and Mintz (2004), with a 10 percent tax rate the simple rule generated a $21,900 threshold. Allowing for behavioral adjustments increased the threshold to $98,100 or $279,600, depending on the distribution of firms. 14 This is not a merely theoretical concern. For almost any level of taxable sales, we have both firms that satisfy the threshold decision rule and firms that violate the threshold decision rule. 15 Although it is possible for this approach to produce multiple optimal thresholds, in practice our distributions were relatively smooth, making the threshold choice unambiguous.

11

Using this approach, the determinants of the threshold still affect the optimal value as we would expect. For a fixed distribution of value added and taxable sales, raising either the administrative costs or compliance costs will raise the threshold. Raising the tax rate lowers the threshold as does increasing the relative value of public funds. V. DATA FOR A U.S. ANALYSIS A. Income Tax Data Our analysis of the optimal VAT threshold in the United States relies primarily on income tax data for tax year 2007.16 We identify the business entities liable for tax as well as the taxable sales and value added tax base for these businesses from the information appearing on income tax forms.17 We include entities that file one or more of the following business tax returns: Form 1040 Schedules C (nonfarm sole proprietor), E-Part I (rental real estate income), and F (farm sole proprietor), Form 4835 (farm rental income and expenses), Form 1065 (partnership), Form 1120 (C corporation), and Form 1120S (S corporation).18 We treat each of these as a separate business liable for VAT taxes.19 This expansive definition means we treat “labor suppliers” as separate businesses liable for VAT if they report “non-employee compensation” from the Form 1099-MISC on a business tax return as gross receipts.20

16

Note this means our estimates implicitly include the effect of non-compliance with respect to the existing income tax. Non-compliance may differ under the VAT. 17 We obtain data from the samples drawn by Statistics Division of IRS from the population of filers for each of the legal forms of business included. 18 We consider a filed Schedule E a business only if it has entries for real estate income in Part I. Other parts of Schedule E contain income passed through from flow through entities that we have elsewhere treated as a business. 19 We treat multiple types of business entities reported by an individual as separate businesses. If an individual Form 1040 has more than one schedule attached to it, we just treat the two schedules with the largest turnovers as separate businesses. This treatment differs than that used by Knittel et al. (2011). We also assume the corporate VAT return would be filed at the same level of consolidation as under the current income tax. 20 Knittel et al. (2011) note that excluding independent contractors from the definition of a small business results in a significant reduction in the number of small businesses. Some of these filers are technically misclassified employees while others are correctly treated as independent contractors.

12

After choosing the entities to include in our analysis, we estimate taxable sales, purchases and exports. We define taxable sales as the sum of gross receipts, gross rents, gross royalties, fee income for depository institutions, and proceeds from the sale of used business property. We use Form 4797 for information on the gross sale price of business property. We define purchases for this exercise as the items for which a business would receive VAT input credits. There is quite a bit of variation in the detail provided on purchases among the various tax forms. For most businesses, purchases include repairs and maintenance, bad debts, rents, advertising, other deductions, purchases of business property, and cost of goods sold less the embedded labor. Information on purchases of business property is from Form 4562 and is included instead of depreciation.21 VAT input credits accrue in the period that the purchase is made, whereas for income taxes, expenses are typically amortized to more closely match the timing of the recognition of the associated income. Other than the replacement of depreciation with purchases of business property, we do not correct the tax data for timing differences. Corrections of other timing differences would be difficult and these other timing differences have a smaller impact on total purchases.22 B. The Estimation of Exports VAT paid on inputs for exported goods and services are credited to the exporter under a destination-based VAT. Rather than estimating input credits, we arrive at the value added tax base for each business by simply reducing the estimated sales by an estimate of exports. There is

21

For a detailed description of the types of issues in using tax data to estimate purchases and sales of capital, see Kitchen and Knittel (2011). 22 For example, for the property and casualty insurance sector, we use earned premiums and losses incurred in our estimates, which matches income tax timing. For a VAT we would want premiums written and losses paid, but these are not readily available from tax returns.

13

no information about exports on the income tax return so we estimate exports based on information from the Input-Output Table of the United States for 2007 (“I-O Table”) (U.S. Department of Commerce, Bureau of Economic Analysis (BEA), 2010). Using information from the I-O Table’s Use and Make Tables we create a weighted export-to-gross output ratio for each industry in the I-O Table. Industry is reported on each tax return so we are able to estimate exports for each business by applying the appropriate I-O industries export to gross output ratio to our measure of taxable sales for each business in that industry.23 The treatment of wholesale and retail trade in the I-O Table poses a difficulty for the estimation of exports. Exports of commodities purchased for resale by the trade industries are not shown in the trade industries in the I-O Table but are instead shown exported from the producing industry. However, tax returns for the trade industry do include goods purchased for resale in both sales and cost of goods sold. We estimate exports for the trade industries by assuming that the difference between total exports in the I-O Table and exports estimated from tax returns for all industries other than trade are attributable to trade.24 Note the I-O Table essentially classifies establishments into industries, but we apply I-O Table ratios to companies. Corporations in particular are likely to have multiple establishments. In order to correct this industrial classification problem, we apply an establishment to company adjustment matrix supplied by BEA to both gross output and exports in order to adjust the export ratios obtained from the I-O Table to a company basis before applying them to corporate businesses. 23

Since our estimate of exports per business is imputed based on sales and relative size in each major industry, we do not capture all of the variability between firms within each industry and size class. This will in particular affect our estimate of the number of businesses voluntarily registering. 24 There are two limitations of our method for estimating exports. First, businesses with zero taxable sales are treated as having no exports since exports are a percent of sales. Second, zero rating refunds all of the value added taxes paid by an exporter on inputs but not on the exporter’s margin. Our estimate of the refund includes the exporter’s margin and thus is a slight overestimate of the refund resulting from zero rating exports.

14

We adjusted the export percentages to differ between small and medium enterprises (“SMEs”) and large enterprises based on I-O Use Tables provided to us by the United States International Trade Commission (“USITC”).25 We use this information to adjust our more detailed I-O Table results. In order to use the USITC I-O Tables, we must impute the number of employees for each of our businesses since this information is not available on the tax return. In particular, we must distinguish between large businesses with 500 or more employees and SMEs with fewer than 500 employees. We estimate the number of employees by dividing the labor compensation reported on the relevant tax schedule by total annual compensation per employee by industry.26 C. Description of the Database Since there is almost no information reported on business income tax returns about the types of goods and services produced by the business, our database generally does not allow exclusions from the VAT base by type of product. Instead, we exclude from the tax base certain kinds of entities that typically produce sales that are not taxed under a VAT. For this analysis we exclude governments, non-profit organizations (even if they have commercial operations that would typically be taxed), many financial service providers (such as life insurance companies), and the output associated with owner-occupied housing.27

25

Appendix H of the U.S. International Trade Commission (2010) report describes the I-O Table they developed. William Deese and Zhi Wang kindly provided these files to us. Note that there is a loss of precision in our export ratios, as the USITC I-O tables collapse our 61 industries into just 19. 26 The total annual compensation per employee by industry is from the 2007 Occupational Employment Statistics Survey of the Bureau of Labor Statistics (2008). Since this survey does not cover agricultural employees, information from the Department of Agriculture (2007) is used. The resulting employee estimates are consistent with the Small Business Administration data on employment by industry and entity type. 27 Income tax data we have available for consolidated life insurance does not allow us to separate life and non-life activity so we exclude these businesses. Life consolidation with property/casualty insurance operations is typical. However, we do include Real Estate Investment Trusts (REITs) and effectively connected income from foreign corporations.

15

Under a credit-invoice VAT, input credits are allowed only for purchases from a taxable supplier. However, our data do not allow us to identify exempt purchases or exempt sales, including exempt purchases from businesses below the threshold. Therefore we treat all purchases as if they carried input credits. In this way, for purposes of measuring the VAT base for each business, we measure the base as a consumption subtraction VAT. To the extent that we allow credits for exempt goods, we underestimate VAT revenue. We are not able to identify exempt purchases from governments and non-profit businesses so we cannot measure the effect of allowing credits for these purchases on our estimate of the small business threshold. We do not apportion purchases of financial institutions between exempt and taxable sales and so underestimate the value-added base of such institutions. However, we do exclude financial flows, such as interest income and expense, from non-financial firms and thus do account for the impact of the financial services exemption on the tax base of such firms. A robustness analysis presented later in this paper suggests that our inability to identify exempt purchases from small businesses has a minimal effect on our calculation of the threshold. [Table 2 about here] Table 2 shows the number of businesses, taxable sales, purchases, and value added grouped by legal form and by industry. In this table, purchases are defined to include purchased goods and services that carry input tax credits, purchased business property and exports. The largest amount of taxable sales and purchases is in the manufacturing and trade industries. Note that the largest count of businesses is in the real estate industry. This is because we place all of the 11 million Schedule E filers in the real estate industry. [Table 3 about here]

16

The businesses that are liable for a VAT are ranked by size of taxable sales and shown by decile in Table 3. This table shows, within a decile, the distribution among legal forms of both the number of businesses and total sales. Schedule C and E filers together account for 75 percent of all business. The lower deciles are mostly populated by Schedule C, E and F filers and partnerships, while the upper deciles are mostly populated by C and S corporations and partnerships. Schedule C, E and F filers dominate the count of businesses, but C and S corporations and partnerships account for most of the taxable sales; together they account for 93 percent of sales. The top decile includes businesses with more than $228,000 in taxable sales and accounts for approximately 95 percent of total sales. [Figure 2 and Figure 3 about here] As shown in Tables 2 and 3, a key characteristic of businesses in the United States that would be subject to a VAT is that a small number of businesses account for a large portion of the total sales. Figures 2 through 4 illustrate the degree to which some firms contribute disproportionately to both total sales and total value added. Figure 2 shows the number of businesses by level of taxable sales. There are about 1.6 million businesses (or a little more than 3 percent of all businesses) with sales of more than $1 million. At the other end of the distribution, there are approximately 36 million businesses (or 76 percent of all businesses) with taxable sales of less than $50,000. Figure 3 shows both the total sales and value added by taxable sales range. Panel A shows that businesses with more than $1 million in taxable sales account for approximately 89 percent of total sales, while businesses with less than $50,000 in taxable sales account for about 1.6 percent of total sales. Panel B shows that businesses with more than $1 million in taxable sales account for approximately 91 percent of the value added, whereas

17

businesses with less than $50,000 in taxable sales in aggregate actually have negative value added. If administrative and compliance costs are ignored and the threshold is based solely on the taxable sales level where tax revenue is positive, Panel B shows that the threshold would be approximately $20,000.28 This is similar to the level chosen in many countries. Choosing $20,000 as the threshold in the United States would eliminate 29 million businesses from filing VAT returns and decrease refunds by approximately $30 billion because of the predominance of businesses with negative value added. [Figure 4 about here] While Figure 3 shows the net value added in each size group, Figure 4 separates each of these net amounts into both positive and negative components. The positive portion for each size group is all firms with positive value added in that group, and the negative portion is all firms with negative value added in that group. Panel A shows the number of businesses in negative and positive value added position in each size group. There are about 12 million businesses (or almost 25 percent of all businesses) with less than $50,000 in taxable sales that have negative value added. While the largest firms contributed much of the negative value added, there are only around 200,000 firms above $1 million in taxable sales that had negative value added. Panel B separates the positive value added from the negative value added for each group. While the largest businesses contribute most of the positive value added, they also contribute more negative value added than the smallest businesses. The concepts we use for taxable sales, business property, exports and value added are similar to National Income and Product Account (NIPA) concepts so it is natural to compare 28

While revenue in the range of taxable sales of “$5000-$25,000” is negative, the value added at $20,000 and above of sales is positive.

18

measures. Adjustments must be made to the NIPA measures to account for the sectors that are excluded from our estimates as well as for other identifiable conceptual differences. Once these adjustments are made, the measures are generally comparable. Gross output from the 2007 I-O Table is approximately $26 trillion, while sales estimated from business tax returns is approximately $24 trillion. Business property in the NIPA is approximately $1.6 trillion as compared to our $1.7 trillion. Exports are the same $1.5 trillion for both measures because we targeted the NIPA measure. Finally, value added from the NIPA is $8.5 trillion compared to our measure of $6.6 trillion and labor compensation in the NIPA is approximately $6.5 trillion compared to our measure of $3.9 trillion. This difference is attributable to many conceptual, definitional and statistical issues. For example, our tax data reflect non-compliance, while data are corrected for non-compliance before inclusion in the NIPA. Our tax data is pre-audit while the NIPA corrects original tax submissions for post-audit adjustments.29 BEA uses the Quarterly Census for Employment and Wages from the BLS to measure labor compensation, whereas we use compensation as it appears on the tax form. (U.S. Department of Commerce, Bureau of Economic Analysis, 2012) Finally, the adjustments we make to NIPA data, such as sectors that are excluded from our estimates and other identifiable conceptual differences, are limited by source data availability. D. Compliance Costs, Administrative Costs and the Marginal Cost of Public Funds Obviously, there are no data for the United States on the compliance costs of a VAT, so we apply parameters estimated from overseas VAT collection experiences to firm-level data of 29

The BEA (2013) publishes tables that reconcile corporate profits appearing in the NIPA with net income appearing on the corporate tax return as well as nonfarm proprietors income appearing in the NIPA with nonfarm proprietors income appearing on the tax return. See BEA, NIPA, Tables 7.16 and 7.14, http://www.bea.gov/iTable/iTable.cfm?ReqID=9&step=1#reqid=9&step=3&isuri=1&903=291 and 293.

19

U. S. businesses. Since compliance costs are regressive with respect to sales and may depend on firm characteristics, we use a compliance cost function that allows for costs to vary by taxable sales, industry, and whether a paid preparer is utilized. We use Swedish survey data of business compliance costs because Sweden has no VAT small business threshold and, like the United States, Sweden has both personal and corporate income taxes. Therefore, our estimate of compliance costs for a VAT based on Swedish data represents the additional compliance costs for the VAT given that there is an existing income tax.30 For simplicity, we assume that businesses either register and file VAT returns, or do not register and do not file.31 Under this method, aggregate VAT compliance costs would be approximately $40 billion per year if there is no small business threshold.32 Small businesses with taxable sales less than $50,000 have average compliance costs per business of approximately $450. While these businesses make up 76 percent of potential taxpayers, they account for only 45 percent of compliance costs and 4 percent of positive value-added.33 On the other hand, businesses with taxable sales of $1 million or greater have average compliance costs per business of $3,900. These businesses account for 17 percent of compliance costs while making up only 3 percent of potential taxpayers and 79 percent of positive value added. While administrative costs are usually expressed in aggregate as a percentage of GDP, we need to be able to assign administrative costs on a per-business basis. To that end, we estimate a 30

Please see Appendix A for more details. We thank George Contos and John Guyton of the Internal Revenue Service’s Research, Analysis, and Statistics who assisted us in developing the compliance cost estimates. They estimated the coefficients for a VAT compliance cost function using Swedish VAT compliance cost survey data that we apply at the individual business level. We also thank Mats Andersson of Sweden’s National Tax Board for sharing data with us on VAT compliance costs. 31 Registration and return filing may be separate activities in practice. In a small number of countries, businesses below the threshold do not need to collect VAT but must still register (OECD, 2011). 32 These VAT compliance costs are approximately 22 percent of estimated total compliance costs for existing taxes (President’s Advisory Panel on Federal Tax Reform, 2005). The comparable percentage for the United Kingdom is 20 percent (KPMG, 2006). While there are many differences between the United States and United Kingdom tax systems, this similarity is reassuring. 33 On a net basis, this group has an overall negative value added of $295 billion.

20

firm’s administrative costs as a proportional function of that firm’s compliance costs. By using a proportional relationship, we allow for economies of scale in tax administration of large companies similar to the economies of scale in compliance. This proportional relationship enables us to vary administrative costs by business sales, industry, and the use of paid tax preparer similar to the way we vary compliance costs. We assume that the VAT would have a broad base and that reporting requirements would be comparable with existing U.S. taxes. We estimate the specific relationship by taking the ratio of the overall IRS budget to the total cost of tax compliance in the United States, and arrive at a ratio of 10 percent.34 This results in estimated aggregate administrative costs of approximately $4 billion per year if there is no small business exemption level. Because the marginal cost of public funds depends both on the country and tax being analyzed, the ideal MCPF for our use would be one that is measured for a VAT in the United States. However, this is not available. Estimates of the MCPF for other taxes in the United States and for other countries have a wide range.35 Following Keen and Mintz (2004), we choose an estimate of 1.3.36 Finally, we fix the tax rate, τ, to 10 percent and assume that a single rate would apply to all industries and products covered by the tax. The estimates of VAT collections shown in this paper should not be confused with revenue estimates. Revenue estimates account for noncompliance and other behavioral adjustments, including the assumption of a fixed GDP (Nester, 1987).

34

See Appendix A for further discussion. For example, see Table 3 in Kleven and Kreiner (2006). 36 See Table 1 of Jones (2010) for a summary of MCPF estimates. 35

21

VI. THE OPTIMAL U.S. THRESHOLD AND BUSINESS RESPONSE A. The Optimal U.S. Threshold Total sales and total value added are both concentrated at the top of the size distribution of businesses. At the bottom of the size distribution, there are many businesses with low or negative value added. A threshold set at a high level of taxable sales would significantly reduce the number of businesses that would be required to register for a VAT, while having little impact on VAT revenues.37 In addition, tax authorities tend to spend significant resources examining firms that receive refunds compared to firms that are net taxpayers given problems with missing trader fraud schemes.38 Eliminating the very large number of businesses with little or negative value added at the lower end of the size distribution allows administrative resources to be focused on the upper end of the distribution where the payoff to careful examination is likely to be much higher. [Figure 5 and Table 4 about here] Figure 5 shows that the estimated threshold of $200,000 appears at the 90th percentile of the distribution of U.S. businesses.39 The figure focuses on the 60th through 95th percentiles of the size distribution of businesses by taxable sales. The bars show our test statistic from (2), the revenue collected net of administrative and compliance costs adjusted for the MCPF. The threshold is approximately where net revenues exceed zero. Approximately 43 million businesses are below the threshold, and of those, 12.6 million have negative value added. The net 37

If firms below the threshold with negative value added do not find it worthwhile to register for the VAT, a high threshold could even raise VAT revenues. We discuss voluntary registration in Section VI.C.2. 38 Missing trader and carousel fraud occurs when a firm registers for the VAT, buy goods VAT free from another country, sells them with a VAT inclusive price, and then disappears without paying the VAT due. This sequence can be repeated generating large VAT refunds. 39 Our threshold estimate is not very sensitive to the choice of administrative and compliance costs. We calculated the optimal threshold using alternative per business estimates of administrative costs of $100 and compliance costs of $1,000, and find the threshold would be approximately $170,000. These collections costs are based on Cnossen (1994) for a U.S. VAT.

22

value added below the threshold is -$145 billion so that at the 10 percent tax rate approximately $14 billion in refunds are saved by the government.40 The total reduction in administrative and compliance costs respectively are $2.6 and $25.5 billion. Table 4 shows the distribution of the 89 percent of businesses below the $200,000 threshold by legal form and industry.41 The columns showing percentages indicate the percent of the adjacent column that is below the threshold. As expected based on preceding figures and tables, most Schedule E, F, C filers and partnerships are below the threshold. Because we cannot distinguish sales to registered businesses from sales to consumers and unregistered businesses, we measure VAT liabilities for each business as taxable sales less purchases, including those from exempt small businesses, multiplied by the tax rate. In reality the portion of the inputs purchased from exempt small businesses would be ineligible for an input tax credit. The effect that this has on the location of the threshold depends primarily on the overall magnitude of these non-creditable inputs and the location of the firms that purchase them in the size distribution of all firms. We estimate the aggregate quantity of inputs ineligible for a credit by calculating that 4.3 percent of all output is produced by businesses below the threshold. We then assume that 4.3 percent of all purchases and investment would be ineligible for an input credit. If these inputs are distributed throughout the entire economy proportionally to use of inputs by each firm, then the VAT base for each firm falls slightly, in theory raising the threshold. However, the vast majority of inputs are purchased by businesses far above the threshold, and in practice the threshold does not move outside of the one-percentile band containing the $200,000 threshold. 40

We assume that the revenue for the entire value added by an exempt business is lost. However, it is possible some of that value added may have been taxed if the exempt business purchased intermediate inputs from a VAT registered business. 41 For convenience, Table 4 also includes the distributions for businesses above the threshold.

23

While this is not the only method we could use to distribute the non-creditable inputs, other approaches have little or no effect on the location of the threshold. Firms already above the threshold that use non-creditable inputs have no effect on the location of the threshold by construction. Firms below the threshold that use non-creditable inputs increase their VAT base and can potentially lower the threshold. However, since businesses above the threshold prefer inputs that are creditable, any attempt to lower the threshold is likely to result in the newly-taxed firms substituting away from exempt purchases and towards creditable purchases. This reduces their VAT base and moves the threshold back up to the original level.42 We do not believe that purchases of the small amount of non-creditable inputs available from exempt small businesses are likely to significantly affect the location of the threshold. B. Cross-National Comparison of Thresholds The $200,000 threshold is larger than the thresholds for other countries that appear in Table 1. There are several reasons for optimal thresholds to differ across countries. First, the characteristics of the VAT differ among countries (OECD, 2010). The tax rate in particular has a profound impact on the threshold. Our $200,000 threshold is measured for a VAT with a single 10 percent rate, a broad base, and a single uniform threshold. A tax rate of 5 percent would result in a threshold of $600,000 in the United States, while a tax rate of 20 percent which would be more typical of the top rate for most countries in the sample would result in a threshold of $90,000. Column 3 of Table 1 shows the standard VAT rate for each country.

42

It is certainly possible to derive lower thresholds if certain firms are restricted from significantly changing their use of non-creditable inputs. However, we believe that a more realistic assumption is that firms above and below the threshold will generally arrive at an efficient allocation of creditable and non-creditable inputs. While businesses above the threshold may not be able to totally avoid purchases of non-creditable inputs, the volume of turnover and inputs below the $200,000 threshold suggest that most businesses below the threshold could sell to purchasers also below the threshold.

24

Second, two countries implementing the same VAT may still have different thresholds if the underlying economies are different. Even when the VAT rate and definition of the base are the same in two countries, the optimal threshold would differ if either the size or industry distribution of firms differs. For instance, in the United States the optimal threshold for the Agricultural industry by itself is almost $2 million, while the threshold for professional services by itself is a little over $100,000.43 Therefore if the economy were weighted more heavily in either of these two areas, the overall optimal threshold could change dramatically. Furthermore, specific industries may differ across countries according to their level of value added relative to turnover. In countries where value added is high relative to turnover, the threshold will typically be lower than ours. Third, there is a difference between the optimal threshold that we measure and the “appropriate” threshold that would take into consideration political concerns. If two identical countries implement identical VATs, they may still pick different thresholds if their political goals are different. For instance, businesses below the threshold that sell to consumers may have a competitive advantage leading some to argue for a lower threshold than the one at which we arrive. This paper assumes that the goal is to maximize the real value of net tax collections, which incorporates the collection costs as well as the value of public funds but does not take into consideration any other political goals. As mentioned in Section V.C. if we were just concerned with net VAT revenues being positive, and collection costs were not considered, the threshold would be around $20,000. This would place it firmly within the bottom range of thresholds in Table 1.

43

See Appendix B for table of U.S. thresholds by business form and industry.

25

C. The Response of Businesses to the Threshold When a VAT includes a small business threshold, some businesses above the threshold may choose to produce less so that their taxable sales is below the threshold and they do not have to register and collect the VAT. Other businesses that are below the threshold may choose to voluntarily register and pay the VAT in order to gain certain benefits. Either of these responses can result from the rational maximization of after-tax profits. 44 In this section we describe the various incentives that businesses face to avoid or register for the VAT and how we model the response of businesses to these incentives. We also present estimates of the effects that these responses would have on VAT registration if all firms that could benefit from a threshold response do in fact respond. While we do not consider such a complete response to be likely, we model it as such in the spirit of providing an upper bound. 1. The Response of Firms Above the Threshold [Figure 6 about here] Actual collection of VAT revenues may be smaller than desired due to avoidance of the VAT by intentional reductions in taxable sales. Taxing authorities frequently search for legislation that would reduce the incentives for such behavior. For example, Figure 6, taken from Rauhanen and Venetoklis (2011), shows the distribution of firms by taxable sales in Finland that are not registered for the VAT in 2003.45 At the level of taxable sales where the registration threshold occurs, $8,412 (8,500 €) per year, there is a bunching of firms with a steep fall off in

44

Note that the threshold calculated in this paper only reflects the static distribution of businesses. This threshold may no longer be optimal once behavioral responses are taken into account. 45 Appendix D includes a more recent distribution of businesses in Finland provided to us by Rauhanen and Venetoklis separated by those that register and those that do not. Interestingly, despite applying graduated rates just above the 8,500 € threshold, there is still a slight bunching of firms below the threshold.

26

the sales classes just below the threshold. This is consistent with the behavior of businesses that avoid registration for the VAT by decreasing or under-reporting their taxable sales. We assume that a firm will reduce taxable sales to fall below the threshold when the cost of the VAT, relative to the firm’s profit without the VAT, is larger than the reduction in taxable sales needed to fall below the threshold, relative to the business’s sales without the VAT.46 Stated slightly differently, a business will reduce its taxable sales to fall below the threshold when it finds that the savings in net VAT paid exceeds the lost profits from the reduction in sales. It is also possible that businesses above the threshold would split into multiple smaller businesses to effectively reduce their taxable sales. We assume that the United States, like other countries, would institute consolidation rules to prevent this behavior. Although the rules may not entirely eliminate splitting, we do not measure that behavior in this paper.47 Appendix C presents a more detailed derivation of our decision rules which are based on businesses wanting to maximize after-tax profits. Below we provide the intuition for our results.48 Firms more likely to move below the threshold share several characteristics. First, the closer a business is to the threshold the more likely it is to engage in this behavior because the lost revenue is small relative to the cost of the VAT. Second, businesses with significant export sales are less likely to reduce their sales. As long as exports are zero rated, the business can claim

46

Because we assume firms have linear production functions, relative reductions in taxable sales correspond directly to relative reductions in profit. 47 For example, Onji (2009) looked at tax-motivated splitting with the introduction of the Japanese VAT in 1989. Japanese tax law anticipated and tried to prevent such “artificial” splitting, for example by basing eligibility on the amount of sales consolidated with the divesting entity. However, Onji found the distribution of corporations changed with the imposition of the VAT such that there was a clustering of firms just below the threshold. 48 Appendix C presents a more detailed derivation of our decision rules, which are based on businesses wanting to maximize after-tax profits.

27

a refund on the VAT paid on inputs.49 Third, firms are more likely to move below the threshold if they are labor intensive. Firms that are more labor intensive have proportionately less taxable input than comparable firms with little or no labor input, and thus are eligible for fewer input credits. Because they receive fewer input credits, their net cost of complying with the VAT is higher, and they have greater incentive to avoid it. A fourth characteristic, which we unfortunately do not observe in the tax data, is the degree to which the business primarily sells to consumers. A retailer with taxable sales below the threshold would not have to pay VAT on the value added of the last (retail) stage. The tax “discount” may be retained by the retail business or passed on to the final consumer. However, a business that primarily sells to other businesses has reduced incentive to avoid the VAT. If there is a single market price for intermediate goods, a registered purchaser would prefer to buy inputs from another registered business since it will not receive input credits from an unregistered business. Thus businesses that sell to other businesses would be less likely to reduce sales to fall below the threshold. [Table 5 about here] After identifying firms which would benefit from decreasing taxable sales to avoid the VAT, we place them into one of three categories: those which are unprofitable regardless of whether or not they are subject to the VAT, those which are unprofitable only when subject to a VAT, and those which are profitable under a VAT. Table 5 shows estimates of the number of businesses that would potentially adjust sales to fall below the threshold, classified by their percentile rank in the distribution of businesses and the reason for their behavior.

49

Since our estimate of exports per business is imputed based on sales and relative size in each major industry, we do not capture the likely variability. This will in particular affect our estimate of the number of businesses voluntarily registering.

28

The first category covers businesses that have negative net income even when they are not required to collect the VAT and are referred to as the “In Loss” category.50 Many unprofitable firms also have negative value added, meaning they would receive a net payment from the government. However, not all firms in loss have negative value added, and some firms have compliance costs that exceed their VAT refund. For these businesses, the reduction of domestic sales to fall below the threshold lowers their total losses. These in loss businesses make up 7 percent of businesses above the threshold and account for approximately $15 billion of the VAT revenue. The second category, referred to as the “High Labor Cost” category, includes the approximately 3 percent of businesses that are profitable when they do not need to collect the VAT, but become unprofitable when they do. This category includes the businesses with high value added stemming from high labor costs. These firms are unlike the firms in a loss position, because if businesses in this group are able to avoid the VAT, they regain their profitability. The third category of businesses, the “Profitable Responders” category, includes those businesses that are profitable whether or not they are subject to the VAT, but nonetheless choose to reduce taxable sales. They choose to reduce sales both because they are relatively close to the threshold and because their VAT burden, both the tax and compliance costs, are relatively high. This category accounts for 8 percent of businesses above the threshold and 2 percent of the VAT. In particular, those close to the threshold that adjust their behavior make up 6 percent of businesses above the threshold but 38 percent of businesses in the 90th percentile above the threshold. These businesses account for less than 1 percent of the VAT burden above the 50

Net income used in our estimates of the response to the small business threshold is the tax return measure of net income before net operating losses and before income taxes. We do not know if these firms are always in a loss position, or if they are only unprofitable for the current year. Because we have only one year of data, we model them as if this is their permanent state.

29

threshold but 57 percent of the 90th percentile above the threshold. Moving away from the threshold, the percentage of profitable responders rapidly decreases such that by the 94th percentile they are less than 5 percent of businesses and account for less than 8 percent of the VAT. [Figure 7 about here] In all, of the 5.2 million businesses above the threshold, 18 percent could potentially adjust to fall below the threshold. Figure 7 shows the percentage of businesses in each percentile that would potentially reduce domestic sales to fall below the threshold by each of the categories described above. 2. The Response of Firms Below the Threshold There are also businesses below the threshold and otherwise untaxed on taxable sales that have an incentive to voluntarily register for the VAT. We divide these businesses into two groups: those with a negative VAT base, and those with growth potential that would make it advantageous to register. The first group includes businesses with a “Negative VAT Base” that have an incentive to register. These businesses have a potential VAT refund that exceeds the compliance costs that they would incur if they were to register. However this category does not include businesses with VAT refunds that are less than their compliance costs. Our second group of businesses, “Profitable Registrants,” will register if doing so provides an advantage to them as a supplier of goods and services. As discussed in more detail in Bird and Gendron (2007), the response of businesses with a positive VAT base depends on how much the marketplace values inputs from taxed firms relative to untaxed firms. Registered businesses would generally prefer to purchase inputs from registered firms because those

30

purchases convey input credits, and thus lower the purchaser’s VAT liability. This implies that a firm below the threshold may be able to increase demand for its products, relative to similar products sold by unregistered businesses, by registering for the VAT. We do not have enough information to predict how much a business’s taxable sales would increase as a result of voluntary registration. However, we can calculate a statistic that indicates how much demand must grow to induce a firm to register. The assumption underlying the statistic is that a business will voluntarily register when it expects registration to result in an increased demand for its products that will increase after-tax profits. These firms must weigh the potential gains in profit against the cost of collecting the VAT and bearing compliance costs. The overall cost to the firm of registering for the VAT will depend primarily on the tax rate, compliance costs, and the degree to which labor is used in the production process. The growth rate in demand resulting from registration must exceed 11 percent before a significant number of businesses would voluntarily register. The similarity between this growth rate and the tax rate is not a coincidence. For a firm with no labor costs, no exports, and small compliance costs, the growth rate in demand must be at least as large as the new tax rate in order to encourage registration. The only firms that respond to a growth rate in demand below the tax rate are exporters with low labor and compliance costs who are rare in our sample. We present the effects of a 12 percent growth in demand to identify likely responders in the following tables, but a sensitivity analysis is included in the Appendix C. [Table 6 and Figure 8 about here] Table 6 shows the number of businesses that will voluntarily register and the impact on VAT revenues classified by both the reason for voluntary registration and percentile rank. Of the

31

approximately 43 million businesses below the threshold, eight million have a negative VAT base and would be candidates to register and receive a refund. This group accounts for $62 billion of VAT refunds. We find that exports play a small role in determining whether or not businesses receive VAT refunds. This may be due to our imputation methods. However, purchases of capital goods are a major driver of refunds. The bottom percentiles of taxable sales, ranging from $0 to $18, includes a large number of businesses eligible for a VAT refund. Approximately 2.9 million of the 4.3 million businesses in this range have a negative VAT base, and they account for VAT refunds of $22 billion. Because of fraud and administrative cost concerns, countries may only provide VAT refunds over a certain threshold and may also have requirements to limit voluntary registration such that our lowest percentiles would likely be excluded from registering.51 We expect that the United States would have similar rules. If so, we are overstating how many businesses could voluntarily register to receive VAT refunds at the very low sales levels. In Table 6, we also show the number of firms, so called “Profitable Registrants,” that would voluntarily register by percentile ranking if, once they voluntarily registered, they expected demand for their products would grow at 12 percent. Under these circumstances, two million businesses would find it advantageous to register. As shown in Figure 8, this group starts at approximately the 60th percentile and increases until the threshold. Approximately 2 million profitable registrants account for $14 billion in VAT revenue as shown in Table 6.

51

Harrison and Krelove (2005) note that Italy, France, and Peru do not process VAT refunds below a specific claim amount. Moreover, countries may have pre-registration checks to verify the fact the business is an ongoing concern and likely to have a positive VAT liability in the future or is a reliable exporter.

32

Overall, we expect that 8 to 10 million businesses below the threshold will voluntarily register. These businesses account for 23 percent of all businesses below the threshold and would lower VAT revenues by approximately $48 billion. 3. VAT Registrations After Business Response to the Threshold [Figure 9 about here] Figure 9 shows the percentage of businesses in each taxable sales percentile that would register for the VAT, both with and without taking into account their likely behavioral response. The solid line represents the percentage of registrants, absent any behavioral responses. The bars show the percentage of each taxable sales percentile that would likely register accounting for the behavioral responses describes above. Most of the voluntary registrations in the smaller taxable sales percentiles are by businesses eligible for VAT refunds. Moving towards the threshold, the proportion of businesses opting to voluntarily register increases and becomes mostly composed of businesses responding to potential demand growth resulting from registration. Above the threshold, businesses respond by lowering taxable sales in order to fall below the threshold. As described previously, this group is made up of those with losses, high labor costs, and for whom the benefits of adjusting are high relative to the costs. VII. CONCLUSIONS In this paper we use a database of tax returns developed by Knittel et al. (2011) and the theoretical framework developed by Keen and Mintz (2004) to determine the static optimal threshold for a VAT in the United States when the VAT rate is 10 percent. We believe this is the first published measure of an optimal threshold for the United States based on individual firm data.

33

The micro-data allow us to describe the types of businesses that would fall below the optimal threshold both by legal form and by industry. These data also allow us to estimate the response of businesses to the threshold. Relying on after-tax profit maximization behavior, we develop methods to determine which firms will voluntarily decrease sales to move below the threshold and which firms will voluntarily register despite being below the threshold. We estimate the optimal threshold for a 10 percent VAT to be $200,000. This threshold would reduce the number of businesses required to register for a VAT by 89 percent or by some 43 million businesses. Most of these businesses are small non-corporate filers, the largest number being Schedule E filers in the Real Estate industry. Absent voluntary registration, the threshold would also increase net VAT revenues by 4 percent (to $353 billion) since many of the firms below the threshold would otherwise receive refunds. We also explore reasons why this threshold differs from those used in other countries. We estimate almost 1 million businesses have an incentive to reduce their taxable sales in order to fall under the threshold, causing VAT revenue to decline by $31 billion. We categorize these businesses according to their profitability with and without the costs of the VAT. Below the threshold, we estimate at least 8 million businesses would likely voluntarily register for the VAT, resulting in VAT revenues declining by approximately $48 billion. While we find that some profitable businesses will register if it results in an increase in demand for their products, most of those that will voluntarily register are businesses with negative VAT bases.

34

ACKNOWLEDGMENTS We would like to thank Katherine Baer, Walter Hellerstein, Victoria Perry, Ralph Rector, George Zodrow, and two anonymous referees for helpful comments on this paper. Further details, including many tables, are available upon request.

DISCLAIMERS The views and opinions expressed are those of the authors and do not necessarily represent official Treasury positions or policies. Data were provided to researchers in the Office of Tax Analysis (OTA) by the Internal Revenue Service (IRS). However, the IRS did not exercise any editorial control over the presentation or findings of the research. The IRS and OTA operate as separate agencies within the Department of the Treasury.

DISCLOSURE No financial arrangements exist with respect to any of the authors that might give rise to a conflict of interest.

35

REFERENCES Bird, Richard M. and Pierre-Pascal Gendron, 2007. The VAT in Developing and Transitional Countries. Cambridge University Press, New York, NY. Bird, Richard M. and Michael Smart, 2009. “The Economic Incidence of Replacing a Sales Tax with a Value-Added Tax: Evidence From the Canadian Experience.” Canadian Public Policy 35 (1), 85-98. Cnossen, Sijbren, 1994. “Administrative and Compliance Cost of the VAT: A Review of the Evidence.” Tax Notes 36 (12), 1609-1626. Contos, George, Ardeshir Eftekharzadeh, Brian Erard, John L. Guyton, and Scott Stilmar, 2009a. “Econometric Simulation of the Income Tax Compliance Process for Small Business.” Proceedings of the 2009 Winter Simulation Conference. Contos, George, John L. Guyton, Patrick Langetieg, and Susan Nelson, 2009b. “Taxpayer Compliance Costs for Small Businesses: Evidence from Corporations, Partnerships, and Sole Proprietorships.” Proceedings of the 102th Annual Conference on Taxation, 2009. Contos, George, John Guyton, Patrick Langetieg, and Melissa Vigil, 2010. “Individual Taxpayer Compliance Burden: The Role of Assisted Methods in Taxpayer Response to Increasing Complexity.” Proceedings of the 2010 IRS Research Conference. Congressional Budget Office, 1992. Effects of Adopting a Value-Added Tax. Congressional Budget Office, Washington, DC. Debt Reduction Task Force, 2010. Restoring America’s Future: Reviving the Economy, Cutting Spending and Debt, and Creating a Simple Pro-Growth Tax System. Bipartisan Policy Center, Washington, DC, www.bipartisanpolicy.org/sites/default/files/BPC%20FINAL%20REPORT%20FOR%20 PRINTER%2002%2028%2011.pdf. Harrison, Graham and Russell Krelove, 2005. “VAT Refunds: A Review of Country Experience.” IMF Working Paper WP/05/218. International Monetary Fund, Washington, DC. Jones, Chris, 2010. “Interpreting and Using Empirical Estimates of the MCF.” ANU Working Papers in Economics & Econometrics No. 519, Australian National University, Canberra, Australia http://cbe.anu.edu.au/media/12695/519.pdf Keen, Michael and Jack Mintz, 2004. “The Optimal Threshold for a Value-added Tax.” Journal of Public Economics 88 (3-4), 559-576. Kim, Sangheon, 2005. “Optimal Threshold for a VAT in the Presence of Tax Evasion.” Unpublished manuscript. Seoul National University, Seoul, Republic of Korea.

36

editorialexpress.com/cgibin/conference/download.cgi?db_name=iipf2004&paper_id=107. Kitchen, John and Matthew Knittel, 2011. “Business Use of Special Provisions for Accelerated Depreciation: Section 179 Expensing and Bonus Depreciation, 2002-2009.” Paper presented at the Annual Conference on Taxation of the National Tax Association, November 17-19, New Orleans, LA. Kleven, Henrik Jacobsen and Clause Thustrup Kreiner, 2006. “The Marginal Cost of Public Funds: Hours of Work Versus Labor Force Participation.” Journal of Public Economics 90 (10-11), 1955-1973. KPMG LLP, 2006. Administrative Burdens: HMRC Measurement Project. KPMG, London, UK, http://www.hmrc.gov.uk/better-regulation/kpmg1.pdf Knittel, Matthew, Susan Nelson, Jason DeBacker, John Kitchen, James Pearce, and Richard Prisinzano, 2011. “Methodology to Identify Small Businesses and Their Owners.” Technical Paper 4, Office of Tax Analysis, Department of the Treasury, Washington, DC, www.treasury.gov/resource-center/tax-policy/tax-analysis/Documents/OTA-T2011-04Small-Business-Methodology-Aug-8-2011.pdf McLure, Charles E., Jr., 1987. The Value-Added Tax: Key to Deficit Reduction? American Enterprise Institute for Public Policy Research, Washington, DC. Nester, Howard W., 1987. “A Guide to Interpreting the Dynamic Elements of Revenue Estimates.” In Steuerle, C. Eugene, and Thomas S. Neubig (eds.) Compendium of Tax Research 1987, 13-41. Office of Tax Analysis, Department of the Treasury, Washington, DC. Onji, Kazuki, 2009. “The Response of Firms to Eligibility Thresholds: Evidence from the Japanese Value-Added Tax.” Journal of Public Economics 93(5-6), 766-775. Organization for Economic Cooperation and Development, 2008. Consumption Tax Trends 2008: VAT/GST and Excise Rates, Trends and Administrative Issues. Consumption Tax Trends. OECD, Paris, France. Organization for Economic Cooperation and Development, 2009. Taxation of SMEs: Key Issues and Policy Considerations. OECD Tax Policy Studies No. 18. OECD, Paris, France. Organization for Economic Cooperation and Development, 2011. Consumption Tax Trends 2010: VAT/GST and Excise Rates, Trends and Administrative Issues. OECD, Paris, France. Pike, Andrew, 2012. “U.S. Taxes Corporate Income at Comparatively Low Rate.” Tax Notes. 134(12), 1533-1547.

37

President’s Advisory Panel on Federal Tax Reform, 2005. Simple, Fair, and Pro-Growth: Proposals to Fix America’s Tax System. U.S. Government Printing Office, Washington, DC. Rauhanen, Timo and Takis Venetoklis, 2008. “Graduated Tax Relief System of VAT in Finland.” Government Institute for Economic Research (VATT), Helsinki, Finland, https://editorialexpress.com/cgibin/conference/download.cgi?db_name=iipf64&paper_id=310. Rauhanen, Timo and Takis Venetoklis, 2011. “Graduated Tax Relief System of VAT in Finland.” Government Institute for Economic Research (VATT), Helsinki, Finland. Ryan, Paul D., 2010. A Roadmap for America’s Future: Version 2.0. U.S. House of Representatives, Committee on the Budget, Washington, DC, www.roadmap.republicans.budget.house.gov/UploadedFiles/Roadmap2Final2.pdf. Sweden Skatteverket, 2006. Compliance Costs of Value-Added Tax in Sweden. Report 2006:3B. www.skatteverket.se/download/18.906b37c10bd295ff4880002550/rapport200603B.pdf United Kingdom Department for Business Innovation & Skills (BIS), 2011. “Business Population Estimates for the UK and the Regions 2010.” Statistical Release, URN 11/92A, BIS, Sheffield, United Kingdom, https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/32514/bpe _2010_-_statistical_release.pdf. United Kingdom Office for National Statistics, 2010. UK Business: Activity, Size and Location – 2010. Office for National Statistics, Newport, United Kingdom, http://www.ons.gov.uk/ons/rel/bus-register/uk-business/2010/index.html. U.S. Department of Commerce, Bureau of Economic Analysis, 2010. Input-Output Table of the United States for 2007. BEA, Washington, DC, www.bea.gov/industry/io_benchmark.htm. U.S. Department of Commerce, Bureau of Economic Analysis, 2012. “Chapter 10: Compensation of Employees,” Concepts and Methods of the U.S. National Income and Product Accounts. Bureau of Economic Analysis, Washington, DC, http://www.bea.gov/national/pdf/methodology/ch10%202012.pdf U.S. Department of Commerce, Bureau of Economic Analysis, 2013. Interactive Data. Tables 7.12 and 7.14. http://www.bea.gov/iTable/iTable.cfm?ReqID=9&step=1#reqid=9&step=3&isuri=1&90 3=293 http://www.bea.gov/iTable/iTable.cfm?ReqID=9&step=1#reqid=9&step=3&isuri=1&90 3=291

38

U.S. Bureau of Labor Statistics, 2008. “Occupational Employment Statistics Survey of the Bureau of Labor Statistics.” Bureau of Labor Statistics, Washington, DC, http://stats.bls.gov/oes/oes_data.htm. U.S. Department of Agriculture, 2007. “Hired Workers Unchanged, Wage Rates Up 4 Percent From a Year Ago.” U.S. Department of Agriculture, National Agricultural Statistics Service, Agricultural Statistics Board Washington, DC. http://usda01.library.cornell.edu/usda/nass/FarmLabo//2000s/2007/FarmLabo-05-182007.pdf. U.S. Department of the Treasury, Internal Revenue Service, 2007. IRS Data Book: 2007. Publication 55B. Department of the Treasury, Internal Revenue Service: Washington, DC. U.S. Department of the Treasury, 1984. Tax Reform for Fairness, Simplicity, and Economic Growth, Volume 3, Value-Added Tax. U.S. Government Printing Office, Washington, DC. U.S. International Trade Commission, 2010. Small and Medium-Sized Enterprises: Characteristics and Performance. Publication No. 4189. US International Trade Commission, Washington, DC. http://www.usitc.gov/publications/332/pub4189.pdf. Zee, Howell H., 2005. “Notes and Communications: Simple Analytics of Setting the Optimal VAT Exemption Threshold.” De Economist 153 (4), 461-471.

39

APPENDIX A COMPLIANCE AND ADMINISTRATIVE COSTS We asked George Contos and John Guyton of the IRS’s Research, Analysis and Statistics to estimate a VAT compliance cost function for us to use with our micro data. Since the United States does not have a VAT but does have an income tax, we discussed what information would need to be collected and retained by a business that is currently not reported for income tax purposes. For example, under a credit-invoice VAT, businesses would need to provide invoices that report taxes for all their sales as well as gather similar data on their purchases. Where possible, information on current compliance costs was used. In particular, similar to the employment tax system, businesses would be expected to transfer VAT collections relatively frequently and file returns at least quarterly. The IRS has developed a model to measure and explain compliance costs of individual taxpayers for the income tax system (see for example, Contos et al. 2009a and Contos et. al. 2009b). Using survey data for Tax Year 2007, the IRS developed a log-linear regression specification in which the natural logarithm of compliance costs is linearly related to a set of explanatory variables available from tax returns. Compliance costs include both the money spent by, as well as the monetized time of, taxpayers.52 This model controls for both the type and the volume of activities performed by taxpayers to meet their tax obligations and includes variables such as modified positive income and whether a paid preparer is used. Using this log-linear specification, Contos and Guyton estimated VAT compliance costs using 2006 survey data from Sweden’s National Tax Board provided by Mats Andersson. Sweden had surveyed businesses in both 1993 and 2006 about compliance costs of the tax 52

The IRS calculates the three-year average wage earned net of Federal and employment taxes, on a taxpayer by taxpayer basis. Details are provided in Contos et al. (2010).

40

system in part to assess the impact of a more complex VAT law, increased computerization, and changes in the economy (Sweden Skatteverket, 2006).53 An advantage of the Swedish data is that there is no small business threshold. Given that compliance costs are regressive, we thought it was important to have data on businesses typically below thresholds. Using the Swedish data, Contos and Guyton estimated coefficients for the weighted regression of the log of compliance costs and adjusted these for estimated variance. Based on simulations, the total monetized compliance costs were adjusted for the residual effects of heteroskedasticity to fit the reported costs. [Table A1 about here] Table A1 shows the coefficients used to estimate the VAT compliance costs for each business in our data set. The three characteristics that determine compliance costs are the taxable sales of the business, the use of a paid preparer for the tax return, and the predominant industry. Compliance costs are higher for businesses with higher sales levels, that use a paid preparer, and that are in the construction industry.54 Administrative costs of the VAT are more problematic to estimate given their variation with the design of the tax as well as the interplay between compliance and administrative costs with tax revenues. As a result we rely on a rough calculation. The IRS budget for 2007 was approximately $10.8 billion (U.S. Department of the Treasury, Internal Revenue Service, 2007). Estimates of the total compliance costs of the Federal income tax system are approximately $140 billion per year (President’s Advisory Panel on Federal Tax Reform, 2005). The ratio of compliance to administrative costs is thus 13. Since the numerator includes only the income tax 53

In Sweden compliance costs increased approximately 40 percent from 2.5 to 3.0 percent of VAT revenues with much of the increase being the result of there being more businesses and complexity in 2005 than in 1993. This is despite some offsetting effect from computer technology making complexity easier to handle. 54 For businesses with zero taxable sales, we approximated log of taxable sales as $1.

41

while the denominator covers all Federal taxes, we round it to 10 and estimate administrative costs as being 10 percent of compliance costs. The advantage of using this approach is that it provides some degree of economies of scale in tax administration given the functional form used to estimate compliance costs. [Table A2 about here] Table A2 shows both compliance and administrative cost estimates by decile for our sample. For comparative purposes we also show the number of businesses by decile and the estimated taxable sales. While total compliance costs are estimated to be approximately $40 billion, they are less skewed than sales per decile. Compliance costs vary from $198 million for the lowest decile to $13.804 billion for the highest. Administrative costs have a similar pattern given they are a fraction of compliance costs. However, on a per business basis, compliance costs vary from $41 per business to $2,883 per business in the lowest and highest deciles. While compliance costs per business rise with taxable sales, as a proportion of taxable sales they range from over 500 percent for the smallest businesses to 0.1 percent for the largest businesses. Our estimates reflect the regressivity of compliance costs noted by others (Cnossen, 1994). While our per business compliance costs average $836 and are similar to the per business average compliance costs in Sweden of $1,070 (9,516 SEK in 2005), our prototype VAT is simpler since it has only one rate and a relatively broad base than the VAT in Sweden.

42

APPENDIX B U.S. Distribution of Businesses and Optimal Thresholds by Legal Form and Industry [Table B1 here]

43

APPENDIX C DERIVATION OF THE DECISION RULES FOR BUSINESS RESPONSE TO THE SMALL BUSINESS THRESHOLD The detailed information we have on businesses allows us to identify the businesses that are likely to respond to the thresholds. The general principle underlying our analysis is that a firm will maximize its profits by comparing after tax profits when it is registered to profits when it does not register. This comparison takes a different form depending upon whether the business is above or below the threshold. 1. Decision Rules for Businesses Above The Threshold For a business above the threshold, profits are (C1)

1

1

1

Where R = Domestic taxable sales revenue E = Export taxable sales revenue C = Purchases other than depreciable property I = Purchases of depreciable property L = Labor expenses = VAT tax rate m = compliance costs Z= VAT taxable sales threshold For a business below the threshold, profits are (C2)

1

1

1

44

We assume here that the equilibrium price for both goods sold and taxable inputs will rise by the full amount of the tax.55 Because of this, firms that do not collect VAT will see increases in both revenue and costs. Firms that do collect VAT will have their revenue and cost increases offset by the remittance, but will still bear the compliance costs. A taxed firm that decreases its taxable sales to reach the threshold, z, is assumed to also scale down all inputs accordingly. If firms scale down both domestic and foreign sales to the same degree, then the profit is given by:56 1

(C3)

1

1

The profit at the threshold will be greater than that earned above the threshold if

, or

when (C4)



1

1

1

1

1



1

Rearranging this, we get (C5)



1 1

1

1

1

So for the profit earned to be higher by decreasing taxable sales to the level below the threshold

(C6) This can be interpreted as











55



See Bird and Smart (2009) for evidence that much or all of the cost of the VAT is passed on to consumers. This scale assumption may be unrealistic. Because the VAT remittance balances out the additional revenue and costs from the increased price level, taxed firms are indifferent between exports and domestic sales. But if the world price is unaffected, then firms below the threshold now prefer to sell domestically at the margin. However, for the purposes of this analysis, we assume that in the short run the firm will be reluctant to disproportionately cut back on exports. 56

45

or











As long as the relative cost of the VAT is greater than the relative distance to the threshold, then the business has an incentive to decrease output to below the threshold and not pay the VAT. Of the businesses that satisfy this condition and have an incentive to lower their taxable sales and no longer be subject to the VAT, we identify three categories of businesses. The first are firms that are unprofitable even without a VAT, and the second are firms with high labor costs that become unprofitable with the VAT. The remaining firms that have an incentive to reduce their taxable sales below the threshold are profitable responders who would be more profitable if they do not collect VAT. 2. Decision Rules for Businesses Below The Threshold As noted above, for businesses below the threshold (C7)

1

1

1

These businesses can potentially register for the VAT and face profits given by (C8)

1

1

1

In modeling voluntary registration, we consider two types of firms that may wish to register. The first are those that expect a VAT refund because they have a negative VAT base. Any firm that pays tax on purchases exceeding the sum of taxes on sales plus compliance costs will increase profits through registration. This would include start-up businesses with major depreciable asset purchases (I) who would register to obtain a refund of the VAT paid on the purchases of capital goods if there are no restrictions on having to register for multiple years.57

57

As noted previously, we only observe a single year’s income tax statements, so we do not know if a firm has high costs on a permanent basis or a temporary one. For instance, large purchases of capital goods can cause a firm to expect a positive return from the VAT in the year of purchase but to owe net VAT in following years.

46

Similarly, exporters are likely to register assuming their exports are zero rated. For them the profit function becomes 1

(C9)

1

As long as the refunded VAT on inputs outweighs compliance costs, registering for the VAT increases their profits. The second type of business, profitable registrants, may register to increase demand for their product. A large purchaser who finds the VAT unavoidable (or unprofitable to avoid) would rather buy inputs from another taxed entity, so that they can claim a refund on inputs. Thus, small suppliers may wish to register in order to cater to these VAT-paying firms. However, there must be a sufficient increase in demand for their product in order to make registration worthwhile. In such a situation, a firm will face a potential profit of (C10)

∗



1

1

1

1

1

1 1

1 1

Where g is the demand growth rate for this firm’s products if it registers. Rearranging, we get (C11)

∗

So that

∗

if

(C12)



This can be interpreted as









Determining the expected increase in demand for a given firm’s products is beyond the scope of this paper. But for a given growth rate g, we can describe the characteristics of the businesses wishing to register. According to our calculations, up to a growth rate in demand of

47

approximately 11 percent, the registration behavior of most businesses under a 10 percent VAT would not be affected.58 [Figure C1 goes about here.] Starting with a growth rate of 12 percent, an additional 250,000 businesses would register. Not surprisingly, these businesses would have a positive VAT liability. In general terms, voluntary registrations grow most quickly for growth rates between 12 percent and 20 percent in our analysis. In this range, each 1 percent increase in demand leads to an additional 3 percent of unregistered firms signing up for the VAT. Above the 20 percent growth rate, the pace of registration slows dramatically. Figure C1 shows the likelihood of registration for growth rates of 12 percent, 15 percent and 18 percent. As discussed by Rauhanen and Venetoklis (2011), as a business below the threshold grows above the threshold, initially its profits are less because it must pay VAT and bear compliance costs. Eventually, its profits are high enough such that it earns profits that enable it to “jump” over the added costs of the VAT. This hurdle increases the higher the tax rate, compliance costs, and labor costs as a proportion of all costs.

58

For a firm with no labor costs, no exports, and small compliance costs, g must be at least the same size as the tax rate in order to encourage registration. The only firms which respond to g below the tax rate are exporters with low labor and compliance costs and are rare in our sample.

48

APPENDIX D Finnish Distribution of Businesses [Figures D1 and D2 here] In personal communications, Rauhanen and Venetoklis suggested a more informative presentation of the data on the effects of a threshold (8,500 €) on business registration is to present two distributions. Figure D1 shows the distribution of VAT registered businesses in Finland in 2011 while Figure D2 shows the distribution of businesses that did not register for the VAT. They note that businesses above the threshold in Figure D2 include businesses that are exempt from the VAT in Finland, often because the organization provides exempt goods and services. This includes hospital and medical care, welfare and social security, school and university education, as well as financial and insurance activities. Moreover, the distributions reflect the effects of a graduated rate structure that was launched in 2004 while Figure 6 is prior to this change. While the reduction in bunching near the threshold has been slowly dissipating, they note a 2010 change allowing small firms to declare VAT less frequently and a simplification of the application for graduated relief should encourage more firms to take advantage of graduated rates (Rauhanen and Venetoklis, 2011).

49

Table 1 Annual Taxable Sales for Small Business Registration and Collection Thresholds and Tax Rate as of January 2007

Country

US $

National Currency

Standard Rate (%)

Allow Voluntary Registration?

Highest Thresholds United Kingdom France Slovak Republic Japan Czech Republic Ireland

93,558 61,000 87,500 76,300 86,705 1,500,000 80,645 10,000,000 69,930 1,000,000 63,073 55,000

17.5 19.6 19.0 5.0 19.0 21.0

Yes Yes Yes Yes Yes Yes

Lower Thresholds Switzerland Australia Austria Hungary New Zealand Canada Poland Germany Greece Luxembourg Portugal Finland Italy Belgium Denmark Norway Iceland Netherlands

44,118 35,461 34,404 30,769 26,316 25,000 20,895 20,069 11,468 11,468 11,468 9,748 8,028 6,399 5,828 5,624 4,762 2,159

7.6 10.0 20.0 20.0 12.5 6.0 22.0 19.0 19.0 15.0 21.0 22.0 20.0 21.0 25.0 25.0 24.5 19.0

Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No

75,000 50,000 30,000 4,000,000 40,000 30,000 39,700 17,500 10,000 10,000 10,000 8,500 7,000 5,580 50,000 50,000 500,000 1,883

Sector Thresholds?

Minimum Registration Period

Yes

Yes

Yes Yes

Yes

Yes

Yes Yes

No Thresholds Korea None None 10.0 NA Mexico None None 15.0 NA Spain None None 16.0 NA Sweden None None 25.0 NA Turkey None None 18.0 NA Yes Notes: Thresholds converted from home currency based on OECD's purchasing power parities. Source: OECD (2008, 2011)

50

None 2 years 1 year 2 years 1 year None

None 1 year 5 years 2 years None 1 year 3 year 5 years 5 years 5 years None None None None None 2 years 2 years None

None None None None None

Table 2 Distribution of Businesses, Taxable Sales, Purchases and Value Added by Legal Form and by Industry, 2007 Total Number of Businesses Thousands %

Taxable Sales $Billions %

Purchases $Billions %

Value Added $Billions %

By Business Form C Corps S Corps Schedule E Schedule F Schedule C Partnerships Total

1,865 3,990 10,767 2,622 25,537 3,096 47,877

4% 8% 22% 5% 53% 6% 100%

12,137 6,028 303 135 1,366 4,302 24,270

50% 25% 1% 1% 6% 18% 100%

10,748 4,687 292 160 1,040 3,956 20,884

51% 22% 1% 1% 5% 19% 100%

1,389 1,341 11 -25 326 346 3,387

41% 40% * -1% 10% 10% 100%

By Industry Agriculture Mining Utilities Construction Manufacturing Wholesale Retail Transport Information Finance Real Estate Prof Service Administrative Education Health Arts Accommodation All Other Total

3,210 226 23 4,038 713 806 3,559 1,470 555 1,384 14,731 4,549 2,810 739 2,650 1,653 844 3,916 47,877

7% * * 8% 1% 2% 7% 3% 1% 3% 31% 10% 6% 2% 6% 3% 2% 8% 100%

299 445 711 1,912 5,566 3,673 3,144 736 885 816 2,246 1,426 564 53 768 159 549 315 24,270

1% 2% 3% 8% 23% 15% 13% 3% 4% 3% 9% 6% 2% * 3% 1% 2% 1% 100%

325 350 649 1,419 5,011 3,735 3,108 616 666 743 2,113 765 264 31 365 115 394 212 20,884

2% 2% 3% 7% 24% 18% 15% 3% 3% 4% 10% 4% 1% * 2% 1% 2% 1% 100%

-26 95 62 493 555 -62 37 120 219 73 133 661 300 22 403 43 155 103 3,387

-1% 3% 2% 15% 16% -2% 1% 4% 6% 2% 4% 20% 9% 1% 12% 1% 5% 3% 100%

Note: Asterisk denotes amounts less than $1 billion or percentages less than 1 percent.

51

Table 3 All Businesses, Number and Taxable Sales by Decile and by Legal Form, 2007

Taxable Sales Range ($) Thousands 0 - 124 125 - 1,489 1,490-3,618 3,619 - 6,851 6,852 - 11,374 11,375 - 18,449 18,450 - 33,837 33,838 - 73,199 73,200 - 227,999 228,000 and up Total Taxable Sales Range ($) 0 - 124 125 - 1,489 1,490-3,618 3,619 - 6,851 6,852 - 11,374 11,375 - 18,449 18,450 - 33,837 33,838 - 73,199 73,200 - 227,999 228,000 and up

C Corps

4,780 4,795 4,788 4,788 4,787 4,788 4,788 4,788 4,788 4,788

5% 1% 1% 1% 1% 1% 2% 3% 7% 19%

47,877 100%

1,865 4%

($Billion) * 4 12 25 43 70 120 239 619 23,138

C Corps 1% 1% 1% 1% 1% 1% 2% 3% 8% 52%

Number of Businesses Percent of Businesses in Decile S Corps Schedule E Schedule F Schedule C 8% 1% 1% 1% 2% 2% 4% 8% 18% 37%

14% 21% 22% 31% 36% 35% 29% 22% 11% 3%

17% 6% 6% 5% 4% 4% 4% 3% 4% 3%

Parts

37% 70% 69% 61% 56% 56% 58% 57% 49% 21%

19% 1% 1% 2% 2% 2% 4% 6% 10% 16%

3,990 10,767 2,622 25,537 8% 22% 5% 53% Total Taxable Sales Percent of Taxable Sales in Decile S Corps Schedule E Schedule F Schedule C

3,096 6% Parts

2% 1% 1% 1% 2% 2% 4% 9% 20% 25%

45% 19% 23% 31% 36% 34% 29% 21% 10% *

3% 6% 6% 5% 4% 4% 4% 3% 4% *

47% 72% 68% 61% 56% 56% 58% 57% 47% 3%

1% 1% 1% 1% 2% 2% 4% 7% 11% 18%

24,270 50% 25% 100% 12,137 6,028 Note: Asterisk denotes percentages less than 1 percent.

1% 303

1% 135

6% 1,366

18% 4,302

Total

52

Table 4 Distribution of Businesses, Taxable Sales, Purchases and Value Added by Legal Form and by Industry For Businesses Below and Above $200,000, 2007

Number of Businesses Thousands %

Below Threshold Taxable Sales Purchases $Billions % $Billions %

Value Added $Billions %

Above Threshold Number of Businesses Taxable Sales Purchases Thousands % $Billions % $Billions %

Value Added $Billions %

By Business Form C Corps S Corps Schedule E Schedule F Schedule C Partnerships Total

922 2,089 10,568 2,469 24,355 2,263 42,666

49% 52% 98% 94% 95% 73% 89%

49 126 194 41 551 80 1,042

* 2% 64% 31% 40% 2% 4%

94 106 234 68 424 260 1,186

1% 2% 80% 42% 41% 7% 6%

-46 20 -40 -26 127 -180 -145

-3% 1% -362% 103% 39% -52% -4%

943 1,901 199 153 1,182 833 5,211

51% 48% 2% 6% 5% 27% 11%

12,088 5,902 109 94 814 4,221 23,228

100% 98% 36% 69% 60% 98% 96%

10,654 4,581 58 93 616 3,696 19,697

99% 98% 20% 58% 59% 93% 94%

1,434 1,321 51 1 199 526 3,531

103% 99% 462% -3% 61% 152% 104%

By Industry Agriculture Mining Utilities Construction Manufacturing Wholesale Retail Transport Information Finance Real Estate Prof Service Administrative Education Health Arts Accommodation All Other Total

2,980 189 18 3,288 487 510 2,921 1,310 494 1,193 13,909 3,996 2,619 720 2,219 1,577 557 3,679 42,666

93% 83% 78% 81% 68% 63% 82% 89% 89% 86% 94% 88% 93% 97% 84% 95% 66% 94% 89%

52 5 * 121 14 19 63 61 10 30 308 117 49 9 52 23 24 82 1,042

17% 1% 1% 6% * 1% 2% 8% 1% 4% 14% 8% 9% 18% 7% 14% 4% 26% 4%

92 14 3 94 19 19 70 45 15 98 445 84 34 8 36 25 27 60 1,186

28% 4% 4% 7% * 1% 2% 7% 2% 13% 21% 11% 13% 26% 10% 21% 7% 28% 6%

-40 -8 -3 27 -5 * -7 16 -5 -67 -137 33 15 1 16 -2 -2 22 -145

155% -9% -4% 6% -1% * -19% 14% -2% -92% -103% 5% 5% 7% 4% -4% -2% 21% -4%

230 37 5 750 226 296 638 160 60 191 822 553 192 19 432 76 287 237 5,211

7% 17% 22% 19% 32% 37% 18% 11% 11% 14% 6% 12% 7% 3% 16% 5% 34% 6% 11%

247 440 711 1,791 5,552 3,655 3,081 675 875 786 1,938 1,309 515 43 716 136 525 234 23,228

83% 99% 100% 94% 100% 99% 98% 92% 99% 96% 86% 92% 91% 82% 93% 86% 96% 74% 96%

233 337 646 1,325 4,992 3,716 3,038 572 651 646 1,668 681 230 23 329 91 367 153 19,697

72% 96% 100% 93% 100% 99% 98% 93% 98% 87% 79% 89% 87% 74% 90% 79% 93% 72% 94%

14 103 65 466 560 -61 44 103 224 140 270 628 285 21 386 45 157 81 3,531

-55% 109% 104% 94% 101% 100% 119% 86% 102% 192% 203% 95% 95% 93% 96% 104% 102% 79% 104%

Note: Asterisk denotes amounts less than $1 billion or percentages less than 1 percent.

53

Table 5 Number of Businesses and VAT Revenue by Taxable Sales Range above $200,000 Threshold, 2007

Taxable Sales Range ($)

Percentile

90 200,000 - 227,999 91 228,000 - 265,844 92 265,845 - 317,726 93 317,727 - 383,505 94 383,506 - 470,554 95 470,555 - 597,747 96 597,748 - 788,750 97 788,751 - 1,120,733 98 1,120,734 - 1,827,228 99 1,827,229 - 3,889,581 100 3,889,582 and up Above Threshold

All 425 479 479 479 479 479 479 479 479 479 479 5,212

0.1

Percentile

Taxable Sales Range ($)

90 200,000 - 227,999 91 228,000 - 265,844 92 265,845 - 317,726 93 317,727 - 383,505 94 383,506 - 470,554 95 470,555 - 597,747 96 597,748 - 788,750 97 788,751 - 1,120,733 98 1,120,734 - 1,827,228 99 1,827,229 - 3,889,581 100 3,889,582 and up Above Threshold

Number of Businesses (Thousands) All Decreasers In Loss High Labor Profitable Position Costs Responders

All 2,419 3,316 3,575 4,816 5,759 7,197 10,099 13,835 20,250 33,396 248,466 353,127

39 31 36 38 38 37 34 32 32 31 32 379

13 13 12 9 15 12 15 16 14 14 12 146

162 90 49 33 23 13 11 6 3 2 1 393

VAT Revenue ($Million) All Decreasers In Loss High Labor Profitable Position Costs Responders 104 77 126 137 272 266 433 484 750 1,078 11,492 15,219

54

98 93 150 119 205 187 302 412 719 1,186 6,715 10,185

All

1,382 922 586 519 435 305 312 221 156 290 251 5,381

213 135 98 79 76 62 61 55 49 47 44 919

All 1,583 1,092 862 776 913 758 1,047 1,118 1,625 2,554 18,458 30,785

Table 6 Number of Businesses and VAT Revenue of Firms that Voluntarily Register by Taxable Sales Range, 2007 0.1

Percentile Taxable Sales Range ($) 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61

0 - 18 19 - 124 125 - 270 271 - 424 425 - 575 576 - 681 682 - 799 800 - 919 920 - 1,031 1,032 - 1,180 1,181 - 1,299 1,300 - 1,489 1,490 - 1,624 1,032 - 1,816 1,817 - 1,999 2,000 - 2,219 2,220 - 2,439 2,440 - 2,667 2,668 - 2,924 2,925 - 3,149 3,150 - 3,419 3,420 - 3,618 3,619 - 3,948 3,949 - 4,199 4,200 - 4,499 4,500 - 4,799 4,800 - 5,045 5,046 - 5,399 5,400 - 5,749 5,750 - 6,002 6,003 - 6,437 6,438 - 6,851 6,852 - 7,199 7,200 - 7,641 7,642 - 8,044 8,045 - 8,462 8,463 - 8,954 8,955 - 9,362 9,363 - 9,831 9,832 - 10,247 10,248 - 10,799 10,800 - 11,374 11,375 - 11,999 12,000 - 12,449 12,450 - 13,059 13,060 - 13,770 13,771 - 14,399 14,400 - 15,078 15,079 - 15,799 15,800 - 16,693 16,694 - 17,569 17,570 - 18,449 18,450 - 19,535

Number of Businesses (Thousands) Likely to Register Negative Profitable VAT Base Registrants All All 4,307 473 484 475 482 481 476 483 476 482 456 501 477 480 368 590 479 478 480 471 483 482 479 431 494 436 553 444 507 485 479 479 394 563 478 480 479 479 479 479 452 506 444 512 481 478 407 551 473 485 479 478 479

2,880 118 120 117 140 79 70 80 88 93 102 89 98 79 57 98 95 94 75 72 82 74 72 68 73 73 81 69 68 67 71 63 55 72 64 57 67 51 72 46 59 56 51 45 51 59 51 54 49 59 58 52 51

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1

55

2,880 118 120 117 140 79 70 80 88 93 102 89 98 79 57 98 95 94 75 72 82 74 72 68 73 73 81 69 68 67 71 63 55 72 64 57 67 51 72 46 59 56 51 45 51 59 51 54 49 59 58 52 52

All -21,932 -218 -204 -237 -218 -267 -218 -195 -221 -260 -225 -300 -336 -239 -165 -354 -206 -382 -219 -330 -250 -116 -182 -166 -257 -211 -314 -176 -194 -406 -245 -156 -107 -96 -238 -201 -161 -5 -276 -157 -266 -117 -214 -122 -17 -83 -58 -24 149 -123 -161 -93 67

VAT Revenue ($Million) Likely to Register Negative Profitable VAT Base Registrants All -21,926 -216 -202 -234 -217 -275 -228 -208 -237 -280 -240 -327 -364 -270 -191 -404 -249 -427 -272 -392 -319 -192 -269 -244 -355 -303 -441 -285 -325 -539 -383 -303 -236 -294 -423 -387 -367 -226 -493 -416 -511 -401 -474 -449 -363 -408 -348 -470 -246 -526 -571 -538 -410

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1

-21,926 -216 -202 -234 -217 -275 -228 -208 -237 -280 -240 -327 -364 -270 -191 -404 -249 -427 -272 -392 -319 -192 -269 -244 -355 -303 -441 -285 -325 -539 -383 -303 -236 -294 -423 -387 -367 -226 -493 -416 -511 -401 -474 -449 -363 -408 -348 -470 -246 -526 -571 -538 -409

Percentile Taxable Sales Range ($) 19,536 - 20,631 62 20,632 - 21,899 63 21,900 - 23,355 64 23,356 - 24,668 65 24,669 - 26,142 66 26,143 - 27,852 67 27,853 - 29,727 68 29,728 - 31,557 69 31,558 - 33,837 70 33,838 - 36,092 71 36,093 - 38,699 72 38,700 - 41,630 73 41,631 - 44,703 74 44,704 - 48,241 75 48,242 - 52,018 76 52,019 - 56,527 77 56,528 - 61,408 78 61,409 - 66,915 79 66,916 - 73,199 80 73,200 - 79,999 81 80,000 - 88,093 82 88,094 - 97,037 83 97,038 - 108,070 84 108,071 - 120,291 85 120,292 - 134,069 86 134,070 - 150,221 87 150,222 - 170,841 88 170,842 - 196,711 89 $196,712 - $200,000 90 Below Threshold

Number of Businesses (Thousands) Likely to Register Negative Profitable VAT Base Registrants All All 479 474 483 479 479 479 478 479 478 479 478 480 479 479 479 479 478 479 479 476 482 479 478 479 479 479 479 479 54 42,665

52 45 51 54 47 49 47 47 57 55 44 40 49 49 48 46 47 58 47 48 47 41 51 45 44 44 48 41 8 8,035

2 4 7 13 17 15 28 29 32 41 43 47 51 51 49 54 64 77 93 106 108 130 129 149 148 153 149 149 15 1,952

56

54 49 57 67 64 64 75 77 89 96 88 87 100 100 97 100 110 135 140 154 155 171 180 193 192 197 197 190 23 9,987

All -106 126 65 40 113 123 207 218 -16 103 210 429 419 14 266 489 400 371 865 418 889 1,444 1,185 1,224 1,832 1,700 1,303 2,649 263 -14,461

VAT Revenue ($Million) Likely to Register Negative Profitable VAT Base Registrants All -567 -378 -442 -542 -497 -520 -464 -518 -747 -725 -654 -506 -572 -1,025 -876 -656 -913 -967 -656 -1,218 -902 -564 -849 -1,148 -740 -1,192 -1,863 -1,038 -149 -61,565

4 8 14 30 43 39 76 80 90 122 138 155 185 191 188 231 308 381 515 656 743 954 1,007 1,269 1,294 1,537 1,617 1,808 199 13,882

-563 -370 -427 -512 -454 -481 -389 -438 -657 -603 -517 -350 -387 -834 -688 -425 -605 -586 -141 -562 -158 390 157 121 555 344 -246 770 51 -47,684

Table A1 Coefficients to Estimate Business Compliance Costs (Dependent Variable Log Predicted Compliance Costs)

Variable Intercept Log supply Dummy for paid preparer Dummy for agriculture, hunting, forestry, and fishing Dummy for construction Dummy for wholesale and retail trade; repair of motor vehicles, motorcycles and personal and household goods; transport, storage, and communication Dummy for financial intermediation and real estate, renting and business activities Dummy for public administration and defense; compulsory social security, other community, social and personal service activities Dummy for no industry code Adjusted R-Square

Estimated Adjusted Coefficient 2.94 0.3105 0.6688 -0.0619 0.1605 0.1393

t-Statistic 9.17 20.59 17.47 0.37 2.25 0.77

-0.0483

-1.12

0.09

0.78

-0.51

00.33

0.35

57

Table A2 All Businesses, Number and Taxable Sales, Compliance and Administrative Costs, 2007

Taxable Sales Range ($) 0 - 124 125 - 1,489 1,490-3,618 3,619 - 6,851 6,852 - 11,374 11,375 - 18,449 18,450 - 33,837 33,838 - 73,199 73,200 - 227,999 228,000 and up

Number of Businesses (Thousands)

Taxable Sales ($Billions)

4,780 4,795 4,788 4,788 4,787 4,788 4,788 4,788 4,788 4,788

Total

47,877 0% Note: Asterisk denotes amounts less than $1 billion.

58

Compliance Costs ($Millions)

Administrative Costs ($Millions)

* 4 12 25 43 70 120 239 619 23,138

198 1,174 1,725 2,197 2,635 3,085 3,734 4,781 6,704 13,804

20 117 172 220 264 308 373 478 670 1,380

24,270 0%

40,037 0%

4,004 0%

Table B1 Distribution of Businesses and Optimal Thresholds By Legal Form and Industry, 2007 Number of Businesses Optimal Threshold (Thousands) ($) By Business Form C Corps S Corps Schedule E Schedule F Schedule C Partnerships

1,865 3,990 10,767 2,622 25,537 3,096

318,000 171,000 150,000 1,827,000 171,000 789,000

By Industry Agriculture Mining Utilities Construction Manufacturing Wholesale Retail Transport Information Finance Real Estate Prof Service Administrative Education Health Arts Accommodation All Other

3,210 226 23 4,038 713 806 3,559 1,470 555 1,384 14,731 4,549 2,810 739 2,650 1,653 844 3,916

1,827,000 384,000 3,890,000 228,000 318,000 789,000 1,121,000 471,000 384,000 108,000 318,000 108,000 150,000 150,000 120,000 228,000 471,000 318,000

59

Figure 1 Number of Registered UK VAT Payers by Taxable Sales, 2010 (Converted from English Pounds using OECD's Purchasing Power Parities. 3,000

Number VAT Payors (Thousands)

2,500

Threshold $103,345 2,000

1,500

2,412

1,000

500

551 454 334 233

0 0 - 77

78 - 155

156 - 388 389 - 778 Sales Range ($Thousands)

60

146

178

779 - 1,556

1,557 & up

Figure 2 Number of Businesses by Taxable Sales, 2007 16

14

Number of Businesses (Millions)

12

10

8

6

4

2

0

Taxable Sales Range ($)

61

Figure 3 Total Taxable Sales and Value Added by Taxable Sales Range, 2007 Panel A: Total Taxable Sales by Taxable Sales 20,000 17,865.6

18,000

Total Taxable Sales ($Millions)

16,000 14,000 12,000 10,000 8,000 6,000 3,762.1

4,000 1,995.1

2,000 0

0.0

0.4

25.5

181.4

174.6

265.6

Taxable Sales Range ($)

Panel B: Total Value Added by Taxable Sales 2,500 2,071.4

Total Value Added ($Millions)

2,000

Value Added Becomes Positive

1,500

995.7

1,000 563.1

500

0

-218.8

-8.6

-54.8

-32.8

20.9

50.6

-500 Taxable Sales Range ($)

62

Figure 4 Number of Businesses and Total Amount of Positive and Negative Value Added by Taxable Sales, 2007 Panel A: Number of Businesses with Positive and Negative Value Added by Taxable Sales

14,000 12,000

11,417

Number of Businesses (Thousands)

10,000 8,000

7,034 5,585

6,000 4,111

4,000 2,000

3,132 1,203

1,025

923

157

0

4,000

598

793

845

2,000 3,242

840

173

40

3,122

3,636

6,000

Positive Value Added

Negative Value Added

Panel B: Positive and Negative Value Added by Taxable Sales

3,500 2,890.4

3,000

Total Value Added ($Millions)

2,500 2,000 1,500

1,217.1

1,000

794.6

500 0

0.0

-218.8

0.2

-8.8

13.1

-67.9

94.6

-127.4

85.6

-64.7

119.7 -69.1

-231.6

-500 -1,000 -1,500 Positive Value Added ($ millions)

Negative Value Added ($ millions)

63

-221.4

-819.0

Figure 5 VAT Tax Revenue Less Collection Costs Adjusted for MCPF, 2007 60th to 95th Percentiles

3,000

Threshold $200,000

2,000

1,000

470.6

383.5

317.7

265.8

64

228.0

Minimum Taxable Sales per Percentile ($Thousands)

196.7

-3,000

170.8

-2,000

150.2

-1,000

134.1

120.3

108.1

97.0

88.1

80.0

73.2

66.9

61.4

56.5

52.0

48.2

44.7

41.6

38.7

36.1

33.8

31.6

29.7

27.9

26.1

24.7

23.4

21.9

20.6

19.5

18.5

0

17.6

Per Business Tax Revenue Less Collection Costs Adjusted for MCPF ($ Millions)

4,000

Figure 6 VAT Registered and Not Registered Businesses in Finland, 2003 4,000

3,500

(Converted from EUR's using OECD's Purchasing Power Parities)

Registration Threshold

3,000

Number of Businesses

2,500

2,000

1,500

1,000

500

0

Maximum Limit Taxable Sales ($)

65

Figure 7 Percent of Percentile of Businesses Above Threshold That Decrease to Fall Below Threshold, 2007 90th Percentile and Above

60

Threshold $200,000

Profitable Responders High Labor Costs

50

In Loss Percent of Percentile

40

30

20

10

3,890

66

1,827

Minimum Taxable Sales ($Thousands) per Percentile

1,121

789

598

471

384

318

266

228

200

0

Figure 8 Percent of Percentile of Businesses Below Threshold that Voluntarily Register, 2007 Up to 90th Percentile

80

Threshold $200,000

Profitable Registrants 70

Negative VAT Base

Percent of Percentile

60

50

40

30

20

10

170.8 134.1 108.1 88.1 73.2 61.4 52.0 44.7 38.7 33.8 29.7 26.1 23.4 20.6 18.5 16.7 15.1 13.8 12.5 11.4 10.2 9.4 8.5 7.6 6.9 6.0 5.4 4.8 4.2 3.6 3.2 2.7 2.2 1.8 1.5 1.2 0.9 0.7 0.4 0.1 0.0

0

Minimum Taxable Sales ($Thousands) per Percentile

67

Figure 9 Percent of Percentile of Businesses Registering for VAT with and Without Behavior Adjustments, 2007 120

Threshold $200,000 100

With Behavior No Behavior

Percentage of Percentile

80

60

40

20

1,827.2 788.8 470.6 317.7 228.0 170.8 134.1 108.1 88.1 73.2 61.4 52.0 44.7 38.7 33.8 29.7 26.1 23.4 20.6 18.5 16.7 15.1 13.8 12.5 11.4 10.2 9.4 8.5 7.6 6.9 6.0 5.4 4.8 4.2 3.6 3.2 2.7 2.2 1.8 1.5 1.2 0.9 0.7 0.4 0.1 0.0

0

Minimum Taxable Sales ($Thousands) per Percentile

68

Figure C1 Percent of Percentile of Businesses Voluntarily Register with Various Growth Rates, 2007 Up to 90th Percentile 80%

Threshold $200,000

18% Growth Profitable Registrants 70%

15% Growth Profitable Registrants 60%

Percent of Percentile

12% Growth Profitable Registrants 50%

40%

30%

20%

10%

0% 170.8 134.1 108.1 88.1 73.2 61.4 52.0 44.7 38.7 33.8 29.7 26.1 23.4 20.6 18.5 16.7 15.1 13.8 12.5 11.4 10.2 9.4 8.5 7.6 6.9 6.0 5.4 4.8 4.2 3.6 3.2 2.7 2.2 1.8 1.5 1.2 0.9 0.7 0.4 0.1 0.0 Minimum Taxable Sales ($Thousands) per Percentile

69

Figure D1 VAT Registered Businesses in Finland, 2011 14,000

12,000

Number of Businesses

10,000

8,000

6,000

4,000

2,000

0

Maximum Limit Taxable Sales (Eur)

Personal communications with Timo Rauhanen and Takis Venetoklis, Government Institute for Economic Research (VATT), Helsinki, Finland.

70

Figure D-2 Businesses Not Registered for the VAT in Finland, 2011

6,000

5,000

Number of Businesses

4,000

3,000

2,000

1,000

0

Maximum Limit Taxable Sales (Eur)

Personal communications with Timo Rauhanen and Takis Venetoklis, Government Institute for Economic Research (VATT), Helsinki, Finland.

71