Making Only America Great? Non-U.S. Market Reactions to U.S. Tax Reform

We study the foreign externalities of the recent U.S. tax reform, commonly known as the Tax Cuts and Jobs Act (TCJA). Specifically, we examine foreign firms’ stock returns around key tax reform events. We find significant heterogeneity in market responses by country, industry, and firm. Chinese firms experience large negative returns; especially steel, business equipment, and chemical manufacturers; while the rest of the world experiences positive returns. Firms operating in more differentiated product markets experience positive returns, while firms in financial distress experience negative returns, consistent with the TCJA having competitive repercussions. We also find that firms experiencing decreases in effective tax rates following tax reform experience positive returns. Overall, our results suggest that the TCJA had varied, yet systematic effects on foreign firms’ shareholders’ wealth and the global competitive landscape.


Introduction
We study the foreign externalities of the recent U.S. tax reform, commonly known as the Tax Cuts and Jobs Act (TCJA). The TCJA included a corporate tax rate cut from 35 percent to 21 percent. Proponents of the TCJA claimed this tax rate cut was necessary to make U.S. firms more competitive because U.S. statutory corporate tax rates were among the highest in the world.
Opponents, on the other hand, pointed out that American firms pay far less than the statutory tax rate and that U.S. corporate tax revenues are low as a percentage of GDP. Opponents also doubted that the U.S. tax system put firms at a competitive disadvantage. 1 To date, research assessing the "winners" and "losers" of the TCJA has focused on the U.S. stock market's response to tax reform developments (Blanchard et al., 2018;Wagner et al., 2018a;Wagner et al., 2018b). However, tax cuts can have externalities (Donohoe et al. 2018;Kim et al. 2018). In a competitive global market with scarce resources, tax reform that strengthens U.S. firms could harm foreign firms via increased competition. Alternatively, tax reform could help foreign firms by lowering taxes on their U.S. operations or by increasing U.S. prosperity and in turn increasing demand for foreign goods. Finally, positive and negative foreign externalities could offset and result in U.S. tax reform having no net effect on foreign firms. As there is no theoretical consensus on the dominant effect of U.S. tax reform on foreign firms, we look for empirical evidence using foreign firms' equity prices.
We examine short-window stock returns for foreign firms on days of heightened attention to major legislative tax reform developments. We begin by identifying 17 legislative developments leading to the eventual passage of the TCJA. We then use Google Trends as an objective way to 1 The corporate tax rate reduction is not the only feature of tax reform. Proponents also argued other features of tax reform would increase U.S. competitiveness; including new international tax provisions, the elimination of the alternative minimum tax (AMT), and changes to bonus depreciation.
2 identify the most significant developments. Google Trends measures relative volume in search traffic for a given search term. Examining the Google Trends index for "tax reform" among U.S. searchers, we find significant increases in public interest on six of these events, which we use as our event dates. 2 After identifying event dates, we confirm U.S. stock prices increased on these event dates to validate our event date selection process. On average, U.S. firms experienced positive, significant returns of 60 basis points per event window. Summed across all six windows, this equals a 3.6 percent total return; suggesting the market responded significantly to news on those days. We also validate our event dates using cross-sectional tests consistent with prior literature (Wagner et al. 2018b). 3 We then examine the foreign externalities of U.S. tax reform by examining event returns for all foreign firms. On average, we find that foreign firms' stock prices increased by 10 basis points during each tax reform event. Summed across all six events, this represents a total return of 60 basis points. Overall, the net estimated economic impact of U.S. tax reform on all foreign firms is not economically large.
We then examine event returns by country. Foreign firms in most sample countries (i.e., 33 of our 38) experienced positive average stock returns across our event windows. This is consistent with evidence from the European Union (Overesch and Pflitsch 2019). Chinese firms, however, experienced a 4.9 percent total event-period stock market decrease-the largest decline of any country. This decline corresponds to a total market value decrease of about $237 billion for level), and find that these firms experienced more negative returns to tax reform, but the difference is only marginally significant. Third, we find that foreign firms that had an ex post reduction in their effective tax rate (our proxy for exposure to tax cuts in U.S. operations) had positive returns, consistent with markets pricing the value of U.S. tax savings from U.S. tax reform for foreign multinationals. Fourth, we find that firms in financial distress experienced lower returns, consistent with literature on competition that finds that the most vulnerable firms in an industry are the targets of predation. Finally, we find larger firms experienced higher returns, suggesting that our results may be driven by firms with greater exposure to global markets.
This study broadens our understanding of how corporate tax rates affect economic outcomes by exploring how foreign firms reacted to news about the progression of U.S. tax reform.
While prior literature has documented U.S. firms' reactions to the Trump election and the passage of the TCJA (Wagner et al. 2017;Wagner et al. 2018;Blanchard et al. 2018;Koutney and Mills 2018), the literature is largely silent on tax reform's foreign spillover effects. We contribute by documenting significant heterogeneity in foreign firms' response to the TCJA and exploring potential reasons for this heterogeneity. These findings should be useful in ongoing ex post analysis of the effects of TCJA-especially as certain political candidates are currently promoting additional tax reform as part of their legislative agendas and as foreign trade tensions continue.

TCJA and Foreign Stock Returns
For decades, the U.S. had one of the highest statutory corporate tax rates in the developed world. Concrete progress on the first major tax reform since 1986 began with a Ways and Means Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006 6 cleanly attributable to tax reform. 6 We start by examining 17 major developments leading to the passage of the TCJA. We list these events in Table 1. We then use Google Trends to narrow down these 17 events and identify the most informative events. We use the Google Trends index among U.S. searchers for the term "tax reform" as a measure of attention paid to the TCJA, and we assume attention is directly related to changes in the probability of the bill's eventual passage. 7 We consider local peaks in the Google Trends index that correspond with actual legislative events to be event dates. all of which correspond with actual legislative developments in the TCJA (see Table 1). 9 In addition to search data from Google Trends, we obtain financial and market data from Compustat Global for non-U.S. firms. We require each country to have at least 50 firms to be in the sample and omit financial firms. Unless otherwise noted, we use financial data from fiscal year end 2016 to ensure that our measures are not contaminated by the TCJA itself. We measure U.S. export intensity using the OECD's Structural Analysis Database (STAN). Our full sample includes 6 In the online appendix, we examine Election Day returns and find that China actually had positive Election Day returns, while Mexico had negative returns. Additionally, we find very few news articles from China around the time of the election discussing how U.S. tax reform will affect Chinese firms. 7 We examined several alternatives to the simple term "tax reform," and all yield similar patterns, but less search traffic. The (unofficial) title for tax reform was not established as the "Tax Cuts & Jobs Act" until it was introduced by the House on November 2, 2017. 8 The Google Trends index varies from 0 to 100, where 0 indicates little search relative to other search activity for other topics on that day and 100 represents the day with the highest search volume for the search term over the sample period. 9 Another potential source for identifying TCJA dates is PredictIt, a market-based prediction website that allows users to vote on the probability of different events, including tax-related events. In 2017, there was a market for the U.S. corporate tax rate being decreased by December 31, 2017. However, the market was shallow and most large swings do not coincide with any discernable news. Hence, we determined the PredictIt market was not deep enough to produce an informative price. We further discuss and provide empirical facts about the PredictIt market for tax reform in our online appendix. 7 19,410 foreign firms, representing about $22 trillion (USD) in market capitalization.
Tables 2 and 3 report descriptive statistics. In Table 2, we list our sample countries, the number of sample firms in each country, the total market value of sample firms in each country, and each country's total U.S. trade balance. In Table 3, we report variable distributions. The mean firm has about $1.1 billion in market capitalization, a 0.78 book-to-market ratio, and a 0.82 debtto-market equity ratio. Our sample size is limited when we require U.S. export intensity or future decreases in effective tax rates.

Domestic, Foreign, and by Country Stock Returns on Tax Reform Event Dates
In Table 4, we use event time and report value-weighted returns for each of the five days surrounding the event dates listen in Table 1. We report returns for all foreign firms and U.S. firms, then individually by foreign country. In columns (1) -(5), we report the average daily returns from trading days t-2 to t+2. 10 We focus on two day-event returns (i.e., the sum of days t and t+1) and report the average event return for each group in column (6). 11 In column (7), we show the sum of event window returns across all six events and call this the total event return. We use daily unadjusted returns throughout our analysis, which implicitly assumes that the expected daily return for any given stock on any given day is zero. We do not include a market adjustment because the systematic component of foreign firms' stock returns to U.S. tax reform is part of the effect we 10 There is a two-day overlap in two of our windows (i.e., returns for days t-2 and t-1 of 12/20 are also returns for t+1 and t+2 of 12/15). In Figure 2 we omit day t-2 and returns for 12/20 from the total and average daily summaries. We report returns by country for the individual event dates in the online appendix. 11 We use two-day windows in part because differences in time zones prevent news on calendar day t in the U.S. from impacting foreign returns in some countries until t+1. Because Asian firms respond to U.S. stock market news day with a one day lag due to time differences, we use day t+1 across our six event dates to evaluate the economic impact in Asia. We further discuss time zones in the online appendix.
8 seek to capture. 12 To validate our event dates, we first examine U.S. firms' average event period returns. In Table 4, we find that the average return on event day 0 is 43 basis points and is 17 basis points on event day t+1. Summed together, this indicates that the average event window return was 60 basis points and the total event window return across the six events was approximately 3.6 percent. This number does not imply the total U.S. stock market growth due to anticipation of U.S. tax reform was 3.6 percent; rather it is the return across 12 days when the U.S. market was reacting to new information about tax reform. We validate our event dates using cross-sectional analysis similar to Wagner et al. (2018b), who test cross-sectional variation in an expanded event window to test cross-sectional variation in the stock price response to tax reform for U.S. firms. In results reported in the online appendix, we find similar cross-sectional results in the U.S. as those reported in Wagner et al. (2018b). We also find that their full-window results, computed over a two-month period, are largely driven by our event dates. Overall, our U.S. returns analysis helps validate that markets reacted to tax reform news during our event dates, and validates our use of Google Trends to identify dates.
In Table 4, we also report the value-weighted return of all foreign-listed firms in our sample during the five days surrounding our events. The average event return on days t and t+1 are 6 and 4 basis points, and the total event return across all six events is 60 basis points. Thus, our initial results only show a muted foreign response to U.S. tax reform for all foreign countries.
Although we find little aggregate foreign response to key TCJA developments, the results in the remainder of Table 4 indicate there is large heterogeneity in event returns across countries. 12 Kothari and Warner (2008) review the econometrics of event studies and conclude that risk adjustment in short window studies is "typically unimportant," and that "the error in calculating abnormal performance due to errors in adjusting for risk in a short-horizon tests is likely to be small." We find that, on average, firms in only five of 38 countries had negative event returns (China, Singapore, Philippines, Denmark, and Malaysia). The other 33 countries had positive returns, suggesting that investors expected many firms to enjoy positive (or neutral) externalities from U.S. tax reform. The evidence does not suggest that investors believed these foreign firms would be harmed in absolute terms by increased U.S. competitiveness.
However, China's average (total) return on day t+1 is -82 basis points (-4.9 percent). We note that a reversal in the Chinese market is apparent from day t+1 to t+2. It is not clear the source of this reversal, especially given it is not isolated to specific industries or to specific event dates.
However, while the cause of this reversal is unknown, the total return from t+1 to t+2 remains negative even accounting for the reversal, and is still -1.55 percent. 13 In Figure  The large negative reaction on day t+1 is also consistent with Chinese investors responding to news of U.S. returns when the markets opened the next day. The figure also highlights the reversal 13 In subsequent industry tests and tests examining examine returns by event date (see online appendix), we do not find evidence that the reversal is concentrated in a single industry or event date. While we cannot explain the cause of the reversal, this helps alleviate concerns that the reversal is due to an isolated, unrelated event. 14 We include a presentation of these returns aggregated over time in the online appendix.

Returns by Industry
We next examine specific foreign industries. In Table 5, we report value-weighted event window returns by Fama-French 17 industry for both U.S. and foreign firms. In Panel A of Table   5, we compare U.S. firms to non-Chinese foreign firms. The results show that non-Chinese foreign industry returns are highly correlated with contemporaneous returns in U.S. industries (Spearman correlation = 0.629). When examining returns for non-Chinese firms, we find that all foreign industries experienced positive returns to U.S. tax reform news. The industry-level detail suggests that the positive effects of U.S. tax reform were spread across most industries, similar to the distribution of U.S. returns.
The results in Table 5, Panel B show a much different story when comparing U.S. returns to Chinese returns. While the U.S. steel industry had a total positive return of approximately 10 percent, Chinese steel firms had a negative 10 percent return. In nearly every industry, Chinese firms had negative returns. The diverging pattern of U.S. and Chinese returns suggests the possibility that TCJA had competitive ramification for public U.S. and Chinese firms.

Cross-sectional Regression Analyses
The country-level and industry-level returns in Tables 4 and 5 help answer the question: "Who are the foreign winners and losers of U.S. tax reform?" However, they do little to shed light on why certain foreign firms benefit from U.S. tax reform while others do not. While it is Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006 impossible to precisely explain all the ways U.S. tax reform affects foreign firms, we use crosssectional tests to examine several potential channels. Specifically, we estimate the following regression: (1) We estimate equation (1) with unadjusted daily returns (RET) from September 1, 2017 through December 25, 2017 (see Figure 1). We do not use abnormal returns because we do not wish to discard systematic effects, of which tax reform was one. TAX REFORM is an indicator variable equal to one for dates within our two-day event windows (i.e., [0, +1]) surrounding our six events. X is one of five cross-sectional variables: GROSS MARGIN, U.S. EXPORTS, ETR DECREASE, DISTRESS, and SIZE. We also include controls for size, book-to-market ratio, leverage, return on equity, and country fixed effects. We use heteroscedasticity-robust standard errors that are clustered by both firm and date. Regression estimates are weighted by firm market value of equity so that our results are not unduly influenced by small firms. Our estimates of equation (1) are presented in Table 6.
In our first cross-sectional test, we investigate if competition moderates foreign firms' reaction to U.S. tax reform news. We follow prior literature by using industry-adjusted gross margin as a measure of a firm's competitive position, which assumes that firms with higher margins have some advantage that allows them to collect quasi-rents (Li et al. 2013;Kubick et al. 2015). 15 The results in column (1) of Table 6 show that the coefficient on TAX REFORM * GROSS MARGIN is positive and significant, suggesting foreign firms facing lower within-industry competition experienced more positive returns.
15 Consistent with this notion, Kim, Nessa, and Wilson (2018) find that U.S. domestic firms with low product differentiation are most impacted by foreign tax cuts.
12 In our second cross-sectional test, we examine whether foreign firms that export products to the U.S. react negatively to U.S. tax reform news. Our proxy for U.S. export dependence is a foreign industry's U.S. export intensity (U.S. EXPORTS), which is the level of U.S. exports as a percentage of total exports, both measured annually at the country-industry level. Column (2) in Table 6 presents the estimates of equation (1) using U.S. EXPORTS for X. We find a negative coefficient on TAX REFORM * U.S. EXPORTS, although it is only marginally statistically significant (i.e., 10 percent one-tailed). Overall, we observe limited evidence consistent with investors expecting foreign firms with high U.S. exports to face increased U.S. competition following U.S. tax reform.
In our third cross-sectional test, we investigate whether the preferential treatment of a foreign firms' U.S. income moderates foreign firms' response to U.S. tax reform news. Because we are unable to identify the extent of a foreign firm's U.S. operations, we proxy for preferential taxation of U.S. income indirectly with an indictor variable that captures whether a foreign firm's effective tax rate (ETR) decreased following U.S. tax reform (ETR DECREASE). We assume that an ETR decrease is more likely for foreign firms with substantial operations in the U.S.
Specifically, we take the difference in ETR from the most recent quarter available as of the date we obtained the data (August 31, 2018), and the ETR from the same quarter in 2016. Table 6, column 3 presents the results of estimating equation (1) using ETR DECREASE for X. We find that the coefficient on the interaction of TAX REFORM and ETR DECREASE is positive and significant, but economically small (i.e., foreign firms whose ETRs increased after tax reform had 0.08 percent higher event returns).
Fourth, we examine whether financial distress moderates foreign firms' response to U.S.
tax reform. In a competitive environment, constrained firms are most likely to be preyed upon by Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006 competing firms (Bernard 2016). Therefore, any harm related to increased U.S. competitiveness from tax reform should be concentrated among financially distressed firms. In Table 6, Column 4, we present the results of estimating equation (1) using DISTRESS for X. DISTRESS is an indicator for firms in the distress zone of the Altman-Z score (Altman 1968). We find a negative and significant coefficient on the interaction between TAX REFORM and DISTRESS. This estimate is consistent with the theory that financially distressed foreign firms are expected to be most harmed by U.S. tax reform-presumably by the actions of strengthened U.S. competitors.
Finally, we find larger firms experienced higher returns, as the coefficient on TAX REFORM * SIZE is positive and significant. While size is correlated with many different firm characteristics, larger firms are more likely to be multinationals than are smaller firms. This finding is consistent with our results being driven by firms with greater exposure to global markets.

Other Analyses
It is possible that the progression of tax policy simply provided a general signal about the political prospects (and the prospects of future policies) of Donald Trump and the Republican Party, rather than specific news about future tax law. If this was the case, then foreign firms' post-Election Day returns in 2016 should be highly correlated with their returns around tax reform news in 2017. Analysis in the online appendix does not corroborate this concern. The country with the most negative Election Day returns was Mexico. China's Election Day returns are unremarkable-25th in order of smallest to largest. This analysis suggests that our tax reform date returns are not merely capturing increasing power to move an agenda by Donald Trump and the Republican Party.
We tabulate and discuss these findings further in our online appendix.
Relatedly, we examine Google Trends data on other policy topics to ensure news of these policies is not conflated with tax reform news. Specifically, we search for the terms "tariff," "deregulation," and "immigration." We graph search term frequencies in the online appendix, and find that public interest in these other policies do not appear to coincide with interest in tax reform, and that they peaked well after tax reform had passed.
In the online appendix, we also perform additional tests to better understand the stock market reaction in China. Specifically, we examine several China-specific mechanisms that may have been more closely related to the large, negative Chinese return. We examine the price of U.S.
Treasury notes around tax reform news and find no significant change in the price of long-term composite rate for U.S. Treasury securities (which is also consistent with holders of U.S. Treasuries not believing tax reform substantially increased U.S. default risk). We also investigate, but do not find, changes in the dollar-yuan exchange rate around our event dates. Next, we examine Chinese returns by event day and consistently find negative returns, mitigating concerns that our Chinese result is due to a spurious event that is not related to tax reform. Finally, we examine the headlines of the South China Daily News, a major Chinese-focused newspaper based in Hong Kong, to search for major confounding concurrent events. 16 Examples of these headlines are in the online appendix. We find no major confounding events.

Conclusion
We examine how shareholders of foreign firms respond to the events leading up to the TCJA. Specifically, we examine foreign firms' short-window stock returns around six events that drew heightened attention to tax reform developments. We find significant heterogeneity in the worldwide response. While most of the world experienced positive returns, suggesting positive spillover effects, Chinese firms experienced large negative returns in response to the news.
We also examine several different channels through which these return results might occur.
Cross-sectional tests reveal that foreign firms facing little competition and firms best positioned to compete against increasingly profitable U.S. firms exhibit more positive market reactions than other foreign firms. Firms with lower ETRs following the TCJA experience slightly more positive returns. Finally, the general trend in positive foreign returns is concentrated in firms with greater global market exposure.
One limitation of our study is that it only studies expectations of externalities. Similarly, our use of market returns introduces noise from other events and even the possibility of confounding events, although we do our best to rule out these possibilities. Regardless, our results suggest that U.S. tax reform changed the global competitive landscape in systematic, but nonuniform ways. They also lay the groundwork for future research on the actual effects of U.S. tax reform.

Figure 1 Google Trends Index for "Tax Reform" Over Time
In Figure 1, we plot the Google Trends index for "tax reform" over the last four months of 2017 when U.S. policymakers were actively working on federal tax reform. The Index (y-axis) varies from 0 to 100, where 100 represents the highest search activity for a specific time period. The local peaks correspond to periods of relatively high search activity regarding "tax reform," and comprises our events of interest. The dots correspond to dates in 2017 and Index values, respectively. Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006

Figure 2 U.S. and Foreign Stock Returns around High Google Search Activity for "Tax Reform"
In Figure 2, we plot value-weighted unadjusted stock returns by event day (x-axis). Returns are aggregated by event day across the six events listed in Table 1 (e.g., the return on day 0 represents the sum of day 0 value-weighted returns across all six event dates). We examine four groups of firms that are not mutually exclusive: U.S. firms, all foreign firms, foreign firms excluding Chinese firms, and Chinese firms. Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006  Table 1 lists key dates in the passage of the TCJA. "✔" indicates a specific date that is identified as a high attention event for tax reform by Google Trends, and therefore is included in our sample (see Figure 1).   Table 2 lists our sample countries, the number of sample firms in each country, the total market value of sample firms in each country, and each country's total U.S. trade balance (a negative balance means the sample country exports more to the U.S. than it imports from the U.S.). Financial values are measured as of the end of 2016.     Table 4 reports value-weighted unadjusted stock returns around the six events listed in Table 1. The first two rows report returns for foreign and domestic firms, respectively. The first five columns report average returns surrounding tax reform by event day. We use event days [0, +1] as our event returns and present the average event return in the sixth column, Avg. Ret. The last column, Total Ret, is the sum of the event returns for all six events. We also report these returns by country. * indicates the total event return is statistically significance at the 5 percent level.    Table 5 compares total event returns between U.S. and foreign firms. Panel A contrasts stock returns between the U.S. and non-Chinese foreign firms. Panel B contrasts returns between the U.S. and China. We present returns for all Fama-French 17 industries except financial institutions. For each location, the total event return is calculated by aggregating the value-weighted two-day event returns across the six events listed in Table 1. In both panels, Column 3 reports the difference in U.S. and foreign total event returns by industry. Industries are listed in ascending order by the size of the U.S. and foreign return difference. Each panel also reports the correlation between U.S. and foreign returns at the industry level. * indicates statistical significance at the 5 percent level. (1) (2)    (1). The sample covers September 1, 2017 to December 31, 2017. TAX REFORM is an indicator variable set equal to one for days in each of our six event windows (i.e., days t and t+1 on 9/27/2017, 11/2/2017, 11/16/2017, 12/2/2017, 12/15/2017, 12/20/2017). Regression estimates are weighted by each firms' market value of equity. Our variable of interest is the interaction of TAX REFORM and X, where X is one of five variables. GROSS MARGIN is a firm-level industry-adjusted measure of margins. U.S. EXPORTS is the value of exports destined for the U.S. divided by total exports (constructed at the foreign country-industry level), ETR DECREASE is an indicator variable coded to equal one if the firm's quarterly ETR decreased between 2016 and 2018, and DISTRESS is an indicator variable for financial distress consistent with Altman (1968), and SIZE is the natural log of MVE. BOOK-TO-MARKET is book value of equity divided by MVE, LEVERAGE is long-term debt (including the current portion) scaled by MVE, and RETURN ON EQUITY is net income before extraordinary items scaled by MVE. We include country fixed effects and cluster standard errors by firm and date. *, **, and *** denote two-tailed statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

A1. Tariffs, Deregulation, and Immigration
We examine the possibility that foreign markets reacted to non-tax U.S. legislative developments over our event dates. This is particularly relevant, since in the months following tax reform, new tariffs were implemented, particularly on certain Chinese goods. To examine whether these tariffs were discussed in concert with tax reform, we examine the Google Trends index for "Tariff" over the time period leading up to tax reform. We also examine Google Trends indexes for "Deregulation," and "Immigration," both of which foreign markets could also have reacted to.
We graph the additional three indexes described above in Figure A1. Panel A examines daily Google Trends indexes from September to December in 2017. Panel B examines weekly indexes over an extended time period. The figures show no discernable peak in searches for tariffs, deregulation, or immigration during our key tax reform dates, or even throughout that time period.
Indeed, throughout the time period leading up to tax reform, there is little search activity on tariffs or deregulation, while interest in immigration appears to be stable. However, after tax reform was passed, tariffs became a policy priority of President Trump. Similarly, immigration garnered more attention in 2018. As can be seen in Panel B, it is clear that the period when the most attention was paid to tariffs and immigration occurred after the passage of tax reform. Further, relatively little attention was paid to deregulation either before or after passage of the TCJA. This evidence suggests that other policy news did not occur concurrently with tax reform news.

A2.1. Validation of Event Dates
Our study is closest to Wagner et al. (2018b), who examine how U.S. firms respond to legislative developments leading to the TCJA. While a key difference in our study is our focus on foreign firms, rather than U.S. firms, we also differ in our choice of event dates. Wagner et al.
(2018b) consider the entire window between November 2 and December 22 in 2017, as well as interim and single-day periods, to examine cross-sectional variation in the stock price response to tax reform for U.S. firms. Our approach uses Google Trends to capture the most important legislative developments. While tax reform was considered over several months, long-window tests in our setting would increase the probability that other events contaminate the event period return in foreign markets. Since our population of interest is all foreign firms, the contamination of event window returns is a non-trivial concern.
To validate our approach, we estimate a cross-sectional model of U net interest to 30% of EBIT . To evaluate these predictions using our event dates, we estimating the following model: RET is the daily return for firm i on each of our six event dates (t In Table A1, Column (1) In Column (2) of (2018b) also conduct analysis on three specific event dates within their window, two of which are our event dates (November 2 and 16), the other of which is two days after our date (December 4). Table A2. Cross-Sectional Variation in U.S. Event Window Returns Table A2 presents cross-sectional tests similar to Wagner et al. (2018b). RET is the daily unadjusted return from CRSP. The remaining variables are drawn from Compustat (Compustat mnemonics in capitals in parentheses). CASH ETR is cash taxes paid in percent of current year pretax income, adjusted for special items (100*(TXPD/(PI-SPI))), FOREIGN INCOME is the absolute value of foreign income in percent of total assets (100*(ABS(PIFO)/(AT))). CAPEX is capital expenditures in percent of assets (100*CAPX/AT), INTEREST is interest deductibility curtailed (a binary indicator variable equal to one for firms where interest expense exceeds interest income plus 30% of EBIT, that is, XINT>IDIT+0.3*EBIT). Controls follow Wagner et al. (2018b), and include the log of market capitalization, one-year revenue growth (100*(SALE-SALEt-1)/SALEt-1), and profitability (100*pretax income / assets = 100*(PI/AT)). Industry indicators follow the Fama-French 30 industry classification. *, **, and *** denote two-tailed statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. (1)

A2.2. PredictIt as an Alternative Method to Identify Event Dates
Our analysis uses Google Trends to identify the most important dates with regards to tax reform. 18 We assume that the act of searching for information reveals what news is informative to the searcher. Hence, we assume that the public is interested in newsworthy events and not interested in non-events. Therefore, we assume that an actual legislative event (e.g., formal action, vote, committee discussion, etc.) that is combined with intense public interest signals a meaningful event that advanced legislation and is not simply a political formality. This assumption is corroborated by positive U.S. returns in our event windows, in addition to cross-sectional tests discussed above suggesting these returns are concentrated within firms most likely to be affected by tax reform.
Another potential source for identifying TCJA dates is PredictIt, a market-based prediction website that allows users to bet on the probability of different events, including tax-related events.
PredictIt, and other market-based prediction tools, have been used in academic research before. Relevant to our study, there was a market for the U.S. corporate tax rate being decreased by December 31, 2017. We note, however, that this market was not liquid. To see this, we compare the tax cut market with a market that has been successfully used in research. Beard et al. (2017) examine the probability of President Trump being elected.  (89), the average (median) volume was 4,090.78 (1,133.5). Indeed, the minimum volume for the Trump market was larger than the mean volume for the tax cut market. Total Trump market volume was 3,562,500 shares, while total tax cut market was 368,170 (46.7% of which volume occurred on the final four days of the market).
Further, while the Trump market appears to have been very liquid, intraday price fluctuations somewhat reasonable, etc., PredictIt found that the odds of Donald Trump being elected president were 22% the day before the election, so even a liquid market can be very wrong. Figure A2, Panel A, graphs three day returns in the PredictIt market for a corporate tax rate cut happening by the end of the year. The black bars represent any of the legislative events in Table   1. Most of the spikes in PredictIt prices on this market do not coincide with the actual tax reform process. It appears that these dates merely represent speculators trying to profit from a relatively shallow prediction market, as very few of the actual high-return days align with actual legislation event dates. Further, due to the specific way the prediction market contract was written (passage by end of the year, 2017), a potential delay of passage into 2018 would have also affected the prediction market. For example, there was some belief after Senate and House passage that the bill may not be signed until January, which caused the prediction market price to drop shortly before Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006 the bill was signed. Figure A2, Panel B, graphs the actual price (high, low, and close) to trade, and volume, again showing the erratic nature of the prediction market, both across days, and intraday.
As a result of this noise in the prediction market data, we do not use PredictIt data.  Table 1, and the listed dates are the highest 5 returns (and the only noticeable peaks) in PredictIt returns. Panel B graphs the high, low, and close share price for the same PredictIt market (graphed on the left axis), and share volume (graphed on the right axis). Data are available here: https://www.predictit.org/markets/detail/2726/Will-the-corporate-tax-rate-be-cutby-the-end-of-2017#data. Prior to the TCJA, the most recent large change to the corporate tax code was the Tax Reform Act of 1986 (TRA86). TRA86 dramatically altered the U.S. tax code, placing the U.S.
corporate tax rate at a competitive 34 percent (down from 46 percent), which at the time represented one of the lowest corporate tax rates among developed economies. TRA86 was the result of more than a year of policymakers, practitioners, and academics working together to create a tax reform package that achieved bipartisan support (Slemrod 2018). In the decades that followed TRA86, many countries lowered their corporate tax rate, so that eventually, the U.S. had one of the highest statutory tax rates of any developed economy. The high tax rate allegedly encouraged outbound income shifting, corporate inversions, outsourcing of jobs, and at least according to many pundits, resulted in the U.S. having a generally less competitive business environment.
However, for many years, despite the general consensus among both Democrats and Republicans that the corporate tax code needed changing and that the corporate tax rate should be lowered, fundamental tax reform was "frequently in the air, [but] rarely to be spotted on the ground" (Shaviro 2013). While there had been bipartisan support for a lower corporate tax rate (among other features, such as a switch to a more territorial tax system-(Slemrod 2018)), many of the details of tax reform did not have bipartisan support and tax reform had not progressed in decades. 19 Legislative inaction on corporate tax reform changed with the election of Donald Trump 19 While our focus is on the change to the corporate tax rate, the TCJA included multiple tax changes. In fact, various concepts for international taxation reform were proposed and abandoned . Over time, legislators changed the proposal multiple times (Hills 2019). However, throughout the debate, legislators maintained corporate tax rate cuts as the key feature of U.S. corporate tax reform. For most of the process, the target tax rate was 20 percent, although the final compromise was a corporate tax rate of 21 percent. One limitation of our study is that we cannot isolate the effects of the rate change from other features of the TCJA, which changed throughout the legislative process.
as President on November 9, 2016, along with a Republican-controlled House of Representatives (hereafter "House") and Senate (Wagner et al. 2018b Unlike the TRA86 process, which took more than a year, and involved input from many 20 The market seemed to appreciate this fact, as returns following the Trump election are consistent with the market expecting fundamental tax reform of some type (Wagner et al. 2018a). However, different iterations of tax reform were considered subsequent to the election. Primarily, the border adjustment tax was heavily considered and eventually dropped months prior to the events leading to the passage of the TCJA  examine the U.S. market response to the border adjustment tax). We consider foreign market returns to the Trump election later in the online appendix. 21 The official name of the act was "The Act to provide for reconciliation pursuant to titles II and V of the concurrent resolution on the budget for fiscal year 2018," which, while less glamorous, is far more descriptive. President Trump wanted the name of the bill to be the "Cut, Cut, Cut" bill (Lee 2017) . different groups, the TCJA happened quickly with little outside input. While potentially suboptimal from a policy design perspective, from a research design standpoint, this setting allows us to examine a small set of well-defined dates on which the probability of fundamental tax reform increased in a discrete, rather than continuous, manner. Several pieces of evidence suggest this explanation is not plausible. First, while the longterm composite rate for Treasuries with a life of greater than 10 years did increase around our set of tax reform events, the increase is not large. In Figure A3, Panel A shows that, on average, the increase was merely 2 basis points. 22 Additionally, none of the major credit rating agencies (Moody's, Fitch, or S&P) downgraded their ratings of U.S. government data following tax reform.

A3. Alternative Explanations of Chinese
While both of these facts have implications for our Chinese result, they also suggest that traders in the market for U.S. Treasury bills did not believe that tax reform significantly increased the risk that the U.S. defaulted on its debt.
Moreover, Labonte and Nagel (2013) report that China and Japan both hold more than $1 trillion of U.S. federal debt (21.9 percent and 19.9 percent of all federal debt collectively

A3.2. Reversal of Stock Returns in China
While we observe a large stock return in China on event day t+1 in our main empirical analysis, we also observe a large reversal on day t+2. Because of this large systematic reversal, we provide a range of estimates for Chinese event window returns rather than a single point. In this section, we tabulate more detail about Chinese returns to ensure that the large reversal is not an artifact of an event or data error. In Panel B of Table 5, reported in the paper, we see the reversal across all industries (i.e., every Chinese industry had a negative return on day t+1 and positive return on day t+2). Below we tabulate Chinese returns by event date. As we find when examining Chinese market reactions by industry, we find a reversal from t+1 to t+2 in every event window.
As such, we do not find evidence of a systematic concentration of the reversal around any single event.    Table A4 reports value-weighted unadjusted stock returns by country around the presidential election in 2016 (i.e., Election Day is day 0). The first five columns report average portfolio returns for the five days surrounding tax reform event days. The last column, Total Ret, is the sum of t+1 and t+2.

A5.1. Time Zone Adjustments
We do not adjust our event dates by time zone to try to center the event on day t. Such an effort would be difficult because (1) tax reform news came out at different times of day, (2) different markets operate in different time zones, and (3) markets hours vary. Thus, any simplification would come at the cost of adding noise into the event dates. For robustness, however, we repeat our analysis and assume that North and South American markets were open and information was available on the actual event date. We assume other markets (e.g., European and Asian markets) do not receive the news until the following day. We acknowledge this simplification is imperfect. Figure A4 makes this adjustment to our return windows. The same general trend appears-the U.S. return increases on day t, while the rest of the world (except China), increases. Chinese returns are negative on day t. Electronic copy available at: https://ssrn.com/abstract=3501160 Electronic copy available at: https://ssrn.com/abstract=3363006 A5.2. Visualizing Event Returns with a Cumulative Figure   Figure 2 in the main manuscript plots raw returns by event date. An alternative way to visualize event returns is by accumulating them in event time. Figure A6 is closely related to Figure   2, but makes two changes. First, raw returns are accumulated daily starting with day t-2 as a baseline (e.g., return shows at t-1 is equal to (1+Rett-2)*(1+Rett-1)-1, etc). Second, all daily returns are adjusted by subtracting the daily return from t-2. Figure A6. Cumulative Adjusted U.S. and Foreign Stock Returns around Event Dates Figure A6 plots value-weighted cumulative stock returns by event day (x-axis). Returns from Figure 2 accumulate by event day and are adjusted so all four groups' returns are equal to zero at the beginning of the five day accumulation window.

A5.3. Examining Stock Price Returns by Event Date
We also tabulate two-day stock returns [0, +1] by event to examine which events were most informative to investors. Each of the six main columns corresponds with one of the six event dates from 2017. The final column includes the sum of the six event date returns.   Table A5 reports value-weighted unadjusted returns for event-days [0, +1] by event. The first six columns report average returns for the six tax reform events listed in Table 1. We use event days [0, +1] as our event returns and present the total event return in the final column. Returns are first categorized as foreign or U.S. Returns by country are presented next. * indicates the total event return is statistically significance at the 5 percent level. Electronic copy available at: https://ssrn.com/abstract=3363006