Does Lowering Dividend Tax Rates Increase Dividends Repatriated? Evidence of Intrafirm Cross-Border Dividend Repatriation Policies by German Multinational Enterprises

Abstract

This paper explores the impact dividend taxes exert on the dividends repatriated from foreign affiliates to their German parent company. Based on an augmented Lintner model of firms' dividend payout decisions, the paper focusses on cross-border intra-firm dividend payments of wholly-owned foreign affiliates in the manufacturing sector. Firm-level data from the Microdatabase Direct Investment (MiDi) of the Deutsche Bundesbank is used. Results firstly signal the validity of the original Lintner model for cross-border intra-firm dividend payments of German affiliates abroad, although the target payout ratio and the degree of dividend smoothing drops substantially once timeinvariant unobserved heterogeneity is controlled for. Secondly, results from an augmented Lintner model imply that increases in dividend taxes indeed have a statistically significant negative impact on the expected value of dividends repatriated: Evaluated at the overall mean dividend payment a one percentage point increase in the dividend tax rate would decrease dividends repatriated by about 3.5 percent. Evaluated at the mean of positive dividend payments a semi-elasticity of -1.6 is derived.Keywords: Dividend Policy, Taxes, Lintner Model, Multinational Enterprise JEL codes: G35, H25 Non technical summaryThe aim of this study is to analyse whether dividend taxes exert a statistically and economically relevant impact on the expected value of dividends repatriated from foreign affiliates to their German parent company over the 1999-2005 period. The paper contributes to the literature as evidence on the impact dividend taxes have on cross-border intra-firm dividend payments is rather scarce. This is especially the case for countries applying the exemption system in international taxation, like Germany. The analysis is based on data provided by the Deutsche Bundesbank's Microdatabase Direct Investment (MiDi) database. Specifically, data on German foreign affiliates in the manufacturing sector which are directly and whollyowned by a German parent company is used for the analysis. Host countries included are the EU member countries (as of 2007) and the most advanced Non-EU OECD countries. In total, information contained in yearly data on 587 German affiliates abroad over 7 years is explored. The analysis is based on the widely known Lintner model of dividend smoothing, which is frequently seen as a relevant description of firms' dividend payout behavior. The Lintner model is a partial adjustment model suggesting that firms have a target dividend payment, which is a fraction of its current earnings. The model leads to an empirical specification relating dividends paid to lagged dividends and current earnings of the firm. However, the basic Lintner model has been established for dividend payments to public shareholders, but not the intra-firm case and moreover it does not include taxes as determinants of dividend payments. Thus, two issues prior to exploring the tax effect within the Lintner framework arise: (i) it has to be established that the...

Do multinational firms respond to personal dividend income tax rates?

“…Based on the Lintner model, Bellak and Leibrecht (2010) find a negative effect of taxes on dividend repatriations of German parent companies from foreign affiliates. Furthermore, the authors introduce a solution for the "initial conditions problem," i.e., while dividend payments depend on past dividend payments, typically, the first payment is unobserved.…”

Consequences of the New UK Tax Exemption System: Evidence from Micro-level Data

“… Notes . Since dividend payments are not directly observed in the data, we follow Bellak and Leibrecht () and calculate DIV i ,2009 as the difference between available shareholder funds for distribution (equity) after current profits in the 2008 financial statement (Amadeus codes: SHFD i ,2008 + PL i ,2008 ) and available shareholder funds for distribution before current profits in the 2009 financial statement ( SHFD i ,2009 ). Where we observe negative values, DIV i ,2009 is set to zero.…”

“…Dividends are not directly observable in the data. We, therefore, approximate dividend payments from the liability side of the balance sheets (Bellak and Leibrecht, , for this approach). Any reduction in shareholder funds available for distribution should be due to dividend payments.…”

“…Although distributed dividends are not directly observed in the data, we can calculate them from balance sheet information by comparing shareholder funds available for distribution (a firm's equity including accumulated retained earnings) over time. This approach follows Bellak and Leibrecht () who show by comparing dividends calculated from balance sheet data with dividend payments directly observed that this method leads to reliable measures of dividend payments. The latter data are not available for a sample of European multinationals as needed for the purpose of our study.…”

“…Notes. Since dividend payments are not directly observed in the data, we follow Bellak and Leibrecht (2010) and calculate DIV i,2009 as the difference between available shareholder funds for distribution (equity) after current profits in the 2008 financial statement (Amadeus codes: SHFD i,2008 + PL i,2008 ) and available shareholder funds for distribution before current profits in the 2009 financial statement (SHFD i,2009 While an increase in the flow of repatriated dividends seems to be a natural prediction as the tax system (tax credit versus tax exemption) of the ultimate owner affects repatriation policy of firms directly, it is not clear for which purposes hitherto deferred foreign-earned income is used in the ultimate owner country in case of repatriation. Although, e.g.…”