The Bright Side of Corporate Diversification:

Transcription

1 The Bright Side of Corporate Diversification: Evidence from Policy Uncertainty Brian Clark Lally School of Management, Rensselaer Polytechnic Institute Troy, NY Bill B. Francis Lally School of Management, Rensselaer Polytechnic Institute Troy, NY Gilna Samuel* Lally School of Management, Rensselaer Polytechnic Institute Troy, NY *Contact author

2 The Bright Side of Corporate Diversification: Evidence from Policy Uncertainty Abstract This paper investigates the relative impact of economic policy uncertainty on real and financing decisions for single-segment and diversified firms. We show that policy uncertainty decreases the diversification discount and improves the efficiency of internal capital markets. Using segment-level data, we find that that policy uncertainty is more likely to reduce capital expenditures of single-segment firms relative to diversified firms. In particular, the impact of policy uncertainty is reduced by unrelated diversification, and more pronounced during periods of higher uncertainty and for firm financially constrained firms. Also, core segments and segments in politically sensitive industries are more affected by policy uncertainty. Using quarterly data, we find that diversified firms are more likely to increase intangible capital and cash holdings and reduce payouts and debt issues. Overall, this study shows that corporate diversification significantly influences the impact of economic policy uncertainty on corporate decisions. Importantly, accounting for diversification is necessary to get a more complete understanding of the impact of economic policy uncertainty at both the firm and aggregate levels. Key Words: diversification, economic policy uncertainty, corporate investments, internal capital markets

3 1 Introduction The impact of uncertainty about regulations, taxes, government spending or fiscal policies has drawn a lot of interest from policy makers, the media and academics. Policy uncertainty could increase financial constraints and lead to more costly external financing (Pastor and Veronesi, 2013; Gilchrist, Sim, and Zakrajsek, 2014; Brogaard and Detzel, 2015). This increase in external market frictions has been shown to significantly impact firm decision making. For example, around national elections firms decrease investments (Julio and Yook, 2014) and increase tax avoidance (Li, Maydew, Willis and Xu, 2016) and R&D (Atanassov, Julio and Leng, 2016). Also, around gubernatorial elections firms decrease investments and SEO offerings (Jens, 2017). The literature shows that economic policy uncertainty decreases investment (Gulen and Ion, 2016) and merger activity (Gulen, Ion and Bonaime, 2017; Nguyen and Phan, 2017). These studies although informative and important, do not consider the difference between diversified and single-segment firms. A key distinguishing factor of diversified firms is the presence of an internal capital market which can provide an alternative source of funding when external markets are constrained. Matvos, Seru and Silva (2018) find that firms become more diversified during periods of high external market frictions. Kuppuswamy and Villalonga (2015) show that the value of diversified firms increase during the 2008 financial crisis. Thus, the influence of corporate diversification should be accounted for when examining the effect of economic policy uncertainty on the firms financing and investment decisions. This paper fills this gap by investigating the impact of economic policy uncertainty on diversified firms relative to single-segment firms. The extant literature provides two alternative explanations for the impact of economic policy uncertainty on investments of diversified firms. On the one hand, diversified firms benefit from their internal capital market and have greater flexibility in addressing their

4 decision-making as external market frictions increase. For instance, a single-segment firm cannot undertake profitable investments if it lacks internal financing while diversified firms could decrease resources from less productive divisions to finance more profitable investments (Matvos and Seru, 2014). Also, as Stein (1997) points out, because lower information asymmetry and agency costs exist within diversified firms they are able to transfer resources more efficiently. As a result, corporate diversification should become more valuable when policy uncertainty increases and the negative impact of economic policy uncertainty on capital investments should be weaker for diversified firms relative to singlesegment firms. On the other hand, Jensen (1986) argues that agency conflicts could incentive managers to engage inefficient investments or waste available resources. Because internal capital markets provide more opportunities for managers to engage in inefficient allocation, managerial agency conflicts could lead to firms allocating resources from strong divisions to weaker divisions (Rajan, Servaes, and Zingales, 2000; Scharfstein and Stein, 2000). Notice that external market frictions could exacerbate this misallocation of resources within diversified firms. Based on these arguments, corporate diversification should be less valuable during times of higher uncertainty and the impact of economic policy uncertainty on investments should be stronger for diversified firms relative to single-segment firms. We empirically examine these competing hypotheses to determine the overall impact of policy uncertainty on diversified firms relative to single-segment firms using the Baker, Bloom and Davis (2016) Index (BBD subsequently) as our measure of economic policy uncertainty. This index is measured as a weighted average of three different components. The first component is a count of words related to policy uncertainty from newspaper articles. The second component contains information about future tax changes. The third component estimates uncertainty surrounding fiscal and monetary policy.

5 First, we examine whether the weaker association between policy uncertainty and investments of diversified firms impacts the value of the firm. We use the measure of excess value by Berger and Ofek (1995) to examine the impact of policy uncertainty on the diversification discount. Our results show that policy uncertainty has a positive and significant impact on the excess value of diversified firms, indicating that the diversification discount is reduced. We also examine the impact of policy uncertainty on capital allocation efficiency (Cho, 2015) and cross-divisional transfers (Rajan, Servaes and Zingales, 2000). We find that policy uncertainty increases capital allocation efficiency and transfers to high productivity divisions. Taken together, these results suggest that policy uncertainty decreases the diversification discount and improves internal capital market efficiency. Thus, corporate diversification becomes more valuable during times of higher policy uncertainty. To examine resource allocation within firms, we investigate the impact the overall BBD index on capital expenditures of diversified firms relative to single-segment firms using segment-level data over the period 1988 to The FASB No. 14 and SEC Regulation S-K require firms to report information on net sales, earnings before interest and taxes (EBIT), depreciation, capital expenditures, and assets for each segment. Given this limited number of variables, we focus on the decisions surrounding segment capital expenditures. First, we examine the impact on single-segment and diversified firms separately and find a strong negative and significant association between policy uncertainty and investments of singlesegment firms while this relationship is economically smaller for diversified firms. Then, we use the full sample and include the interaction of the diversification dummy and policy uncertainty variable. The coefficient of this interaction is positive and statistically significant, indicating that corporate diversification mitigates the negative impact of policy uncertainty on investments. We also run separate regressions using each component of the BBD index

6 and find similar results. In general, we show that the impact of policy uncertainty on investments is less pronounced for diversified firms compared to single-segment firms. Our findings are verified through a number of robustness tests. First, we add controls for future investment opportunities to address concerns that our results are driven by the countercyclical nature of policy uncertainty which could be capturing investment opportunities. We also control for several measures of overall economic uncertainty to verify that our results are driven by economic policy uncertainty and not general economic uncertainty. We add these investment opportunities and overall economic uncertainty controls to our model and we find consistent results. To address endogeneity concerns, we make use of the instrumental variable approach. Following Kaviani, Kryzanowski, Maleki and Savor (2017), we use the relative legislative power of the two major parties in the United States as an instrument for policy uncertainty and find consistent results. Finally, we address diversification self-selection issues by accounting for firms that choose to diversify when external market frictions increase. We exclude all firms that undertook a diversifying merger during each year and find similar results. Overall, these robustness and endogeneity test support our main findings. Next, we investigate the impact of unrelated and related diversification. Segment characteristics such as cash flows and investments opportunities differ more in unrelated diversified firms. Thus, unrelated diversifiers are regarded as more diversified relative to related diversifiers. We find that the relationship between policy uncertainty and investments is significantly reduced by unrelated diversification while related diversification has no significant effect of. Taken together, our results suggest that unrelated diversification mitigates the impact of policy uncertainty on capital expenditures. We also examine the impact during high and low periods of uncertainty. We find that the impact of policy uncertainty on capital expenditures is more pronounced during periods of higher uncertainty.

7 Third, we examine the impact of the 2008 financial crisis and find that our results are driven by economic policy uncertainty and not the crisis. Furthermore, we examine the impact of policy uncertainty on resource allocation within diversified firms. Agency conflicts within the firm could result in certain segments having stronger control rights. Duchin and Sosyura (2013) show that segment managers with social connections to the CEO receive more capital. Matvos and Seru (2014) also point out that diversified firms reallocate resources across divisions when external markets become more constrained. First, we examine the impact of policy uncertainty on investments of the core and non-core segments. The core segment usually has access to greater resources relative to non-core segments and these resources could be used to subsidize non-core segments during times of higher policy uncertainty. We find that there is a negative and significant relationship between policy uncertainty and investments for core segments while this relationship is insignificant for the non-core segments. Second, we investigate the impact on segments in politically sensitive industries and find that the negative effect of policy uncertainty on investments is more pronounced for business segments of diversified firms in politically sensitive industries. These results suggest that the investments of non-core and politically sensitive segments of diversified firms are more susceptible to policy uncertainty. In our analysis so far, we focus on the impact of policy uncertainty on capital expenditures due to the limitation of segment level data. However, other corporate decisions could be impacted by policy uncertainty. As such, we use quarterly data at the firm-level to examine the impact of policy uncertainty on additional real and financial decisions of diversified firms relative to single-segment firms. First, we examine the impact of policy uncertainty on intangible capital. We find that policy uncertainty increases intangible capital and this increase is greater for diversified firms. Next, we examine the impact of policy uncertainty on financing outcomes. We find that policy uncertainty is more likely to increase

8 cash holdings and decrease payouts and debt issues for diversified firms relative to singlesegment firms. We also verify these results by adding investment and overall economic uncertainty controls, performing instrumental variable analysis and excluding firms that undergo diversifying mergers. Additionally, we investigate alternative measures of diversification, the impact during the high and low uncertainty and the 2008 financial crisis period. Overall, the results of these tests are consistent with the main results. We contribute to the literature on the benefits corporate diversification and internal capital markets. Studies such as (Stein, 1997) argue that internal capital markets reduce information asymmetry which allows managers to more efficiently allocate resources. Also, diversified firms are better able to withstand constraints with external markets (Matvos and Seru, 2014) and the value of diversification increases during the financial crisis (Kuppuswamy and Villalonga, 2015). We add to this stream of literature by showing that economic policy uncertainty increases the excess value of firms and thus, reduces the diversification discount. This study also adds to the literature on the impact of policy uncertainty. The literature shows that policy uncertainty can impact corporate decisions such as tax avoidance (Li, Maydew, Willis and Xu, 2016), innovation (Atanassov, Julio and Leng, 2016), venture capital investment (Tian and Ye, 2017) and merger activity (Bonamie, Gulen and Ion, 2016). More specifically, studies such as Julio and Yook (2012), Gulen and Ion (2016) and Jens (2017) show that policy uncertainty decreases corporate investments. These studies examine pooled samples of single-segment and diversified firms; however, these two types of firms could behave differently during periods of uncertainty. Unlike single-segments firms, diversified firms have access to an internal capital market as an alternative source of financing. Our findings show that the impact of policy uncertainty on real and financing decisions is significantly different for single-segment firms compared to diversified firms.

9 Specifically, we find that diversified firms are less likely to decrease capital expenditures and more likely to increase intangible capital and cash holdings and more likely to reduce payouts and debt issues. 2 Literature Review 2.1 Corporate Diversification and Firm Value The seminal paper by Modigliani and Miller (1961) claims that in perfect markets diversification should be irrelevant to firm value. However, more than two-thirds of US corporations are diversified. The extant literature proposes several theories to explain why firms may diversify. For example, agency theories suggest that managers diversify to make it more difficult for them to be replaced. These managers overinvest and expand the firm beyond its optimal size to increase their power or compensation and to entrench themselves (Jensen, 1986; Jensen and Murphy, 1990; Stulz, 1990; Shleifer and Vishny, 1989). Other theories suggest managers diversify to exploit the benefits of an internal capital market. Lower information asymmetry exists within the firm, thus the CEO has insider information about the future investment opportunities of each segment and is better able to allocate resources (Stein, 1997) Also, diversification could provide debt co-insurance benefits. Lewellen (1971) posits that imperfect correlation between segment cash flows reduces overall firms risk. The extant literature also examines the impact of diversification on firm value. The seminal paper by Lang and Stulz (1994) first provides evidence of a diversification discount and shows that diversified firms have lower Tobin's Q relative to single-segment firms. Berger and Ofek (1997) also construct an excess value measure of diversified firms and find evidence supporting the existence of a diversification discount. Santalo and Becerra (2008) point out that the value of diversification varies across industries. Moreover, the literature argues that diversification can be more valuable during periods of uncertainty. Diversification

10 allows firm to obtain financing more easily when external financial constraints frictions exist. Billet and Mauer (2003) show that excess value of diversified firm increases when firms subsidies efficient but financial constrained segments. Also, Kuppuswamy and Villalonga (2015) find that the value of diversified firms increases during the financial crisis. 2.2 Policy Uncertainty and Corporate Decisions Policy uncertainty shocks are positively associated with the business cycles and can predict drops in GDP (Bloom, Floetotto, Jaimovich, Saporta-Eksten and Terry, 2014). The literature also demonstrates that policy uncertainty could impact corporate decisions. A number of variables have been used to capture uncertainty experienced by firms. For example, the literature uses variation in return assets (Leahy and Whited, 1996), volatility in stock returns and variation in analyst forecasts (Bond and Cummins, 2004), total factor productivity (Bloom, Jaimovich, Saporta-Eksten and Terry, 2012) and firm fundamentals (Stein and Stone, 2012) to measure uncertainty. However, measuring uncertainty due to political or policy changes is a challenge. One stream of literature looks at political uncertainty which is driven by political events such as presidential or gubernatorial elections. For example, Atanassov, Julio and Leng (2016) find that firms increase their R&D investments during U.S. gubernatorial election years. Li, Maydew, Willis and Xu (2016) use a sample of firms from thirty countries and show that tax avoidance increases around election years and decreases the year after the election. Jens (2017) find that investments decrease during U.S. gubernatorial elections. Using a cross country sample, Julio and Yook (2012) show that firms reduce capital expenditures during election years. Moreover, Baker, Bloom and Davis (2016) develop a measure of economic policy uncertainty. This measure is broader and includes uncertainty relating to regulations, taxes, government spending or fiscal policies, in addition to elections. Baker, Bloom and Davis

11 (2016) show that their index is associated with reductions in investments and employment in policy-sensitive industries. Gulen, Ion and Bonaime (2017) and Nguyen and Phan (2017) also use the BBD index to estimate economic policy uncertainty and show that the probability of a mergers is negatively associated with policy uncertainty. Also, Gulen and Ion (2016) use the BBD Index to show that firms decrease capital expenditures when policy uncertainty increases. Consequently, the impact of economic policy uncertainty on corporate decisions is a significant area of study. 2.3 Impact of Policy Uncertainty on Diversified Firms The extant literature shows that policy uncertainty could increase external frictions and reduce the financing available to firms (Pastor and Veronesi, 2013; Gilchrist, Sim, and Zakrajsek, 2014). On average, firms decrease corporate investments during times of high economic policy uncertainty (Julio and Yook, 2012; Gulen and Ion, 2016; Jens, 2017). However, diversified firms have access to internal capital markets which provides another source of financing. In this section, we discuss the potential impact of policy uncertainty on diversified firms. The bright side of diversification argues that diversified firms are able to use their internal capital market when external markets are constrained. The internal capital market provides an alternative source of financing to firms that would be financially constrained if functioning as single-segment firms (Fluck and Lynch, 1999). Moreover, fewer agency conflicts and lower information asymmetry exist, thus, managers are able to allocate resources between segments more efficiently (Stein 1997; Shin and Stulz, 1998). Therefore, policy uncertainty should increase the value of diversified firms and corporate diversification should reduce the negative impact of policy uncertainty on firms. The dark side of diversification posits that diversification leads to inefficient allocation of resources. Scharfstein (1998) find that diversified firms tend to subsidize low Q

12 segments while investing too little in higher Q segments. Rajan, Servaes, and Zingales (2000) find that greater diversity could lead to firms transferring resources to inefficient segments. Also, shocks to one segment can impact affect the investments of other segments. Lamont (1997) show that shocks to oil segments impact investments of non-oil segments. Hence, an increase in policy uncertainty could worsen inefficient resource allocation and reduce the value of diversified firms leading to a greater negative impact of policy uncertainty on these diversified firms. In summary, if internal capital markets are efficient then diversified firms are more equipped to withstand constraints due to policy uncertainty. Thus, the value of diversified firms increases and negative impacts of policy uncertainty will be weaken for diversified firms relative to single-segment firms. On the other hand, if corporate diversification leads to inefficient resource allocation then policy uncertainty could exacerbate this inefficiency. Hence, policy uncertainty should reduce the value of diversified firms and have a more pronounced negative impact on diversified firms relative to single-segment firms. 3 Sample and Variable Construction 3.1 Data Sources and Study Sample We use data from several sources. We obtain annual financial and accounting data at the firm-level from Compustat Annual database for the period 1988 to We include firms that report segment level data on the Compustat Industry Segment files. This database provides financial and accounting data for each segment of the firm. Following the literature we keep only business segments; we exclude segments with negative investments, assets or sales; exclude all firm years that have missing information for assets, investments or sales for all divisions; exclude firm years with missing SIC code for segments and exclude firms that have at least one segment in the financial (SIC 6) or utilities (SIC 9) industry. To ensure the integrity of segment data we require that sum of segment sales be within 1% of total firm-

13 level sales. Also, firms with assets less than 20 million are excluded to avoid small firms distorting our results. Lastly, the data on economic policy uncertainty data is obtained from Baker, Bloom, and Davis (2016). 3.2 Economic Policy Uncertainty To measure economic policy uncertainty we use the Baker, Bloom and Davis (2016), BBD Index, henceforth. This index is constructed as a weighted average of three components to measure policy-related economic uncertainty at the monthly level. The first component is obtained by searching the archives of ten large newspapers for policy related uncertainty news. The number of articles containing at least one of the terms uncertainty or uncertain, at least one of the terms economic or economy, and at least one of the terms congress, legislation, white house, regulation, federal reserve, or deficit. are counted and normalized by the total amount of articles in the newspaper during that month to control for changes in the volume of news over time. The time-series data from each of the ten newspapers are further normalized to unit standard deviation and summed within each month. Then, the index is scaled to have an average value of 100 for time period from 1985 to The second component of the BBD index captures policy uncertainty related to future changes in tax code. The discounted value of the revenue effects of tax provisions that will expire in the next ten years is estimated using data from the Congressional Budget Office. The third component measures forecaster disagreement about future monetary and fiscal policies using Consumer Price Index (CPI) forecasts from the Survey of Professional Forecasters provided by the Federal Reserve Board of Philadelphia and forecasts of purchases of goods and services by federal, state and local governments. The index is constructed using the average of the interquartile ranges of these two forecasts.

14 These three components are normalized and the overall BBD index is calculated using a weight of one-half for the news component, one-sixth for the tax component and one-third for the forecaster disagreement component. We define our policy uncertainty variable, PU, as the arithmetic average of the BBD index for each year (or quarter) during our sample period. Figure 1 shows that the BBD index does increase around events, such as elections, recessions or wars that are expected to increase policy-related uncertainty. 3.3 Measuring Diversification Following the extant literature, we measure diversification using a dummy variable which equals to one if the firm reported two or more business segments in different four-digit sic codes and zero otherwise. In additional tests, we measure unrelated and related diversification. Unrelated diversification is one if a firm reports two or more business segments in different two-digit SIC codes and zero otherwise. Related diversification is all other diversified firms and zero otherwise. 3.4 Diversification Discount and Internal Capital Market Measures The diversification discount is determined using a measure of excess value defined as the difference between actual value and the imputed value. If this value is negative (positive) then the diversification discount (premium) exist. We compute the excess value of diversified firms using single-segment firms to estimate the imputed value. First, we compute the market-to-sales for each single-segment firms by dividing its market value by total sales. Then the imputed value for each segment is found by multiplying the segment's sales by the median market-to-sales multiplier of single-segment firms in the same industry. The industry matching is based on the narrowest SIC grouping that includes at least five single-segment firms. The firm's imputed value is calculated as the sum of the segments imputed values. Lastly, the excess value is computed as the difference between the firm's market value and its imputed value.

15 To measure internal capital market efficiency we use the measure of capital allocation efficiency defined by Cho (2015) as the efficiency with which firms actively allocate resources from low to high opportunity segments. We follow the two step procedure by Cho (2015) to calculate this measure of capital allocation efficiency. First, the deviation of capital allocation from a hypothetical value of passive allocation based on segment sales is calculated as follows:, (3.1) where and are the capital expenditures and sales of segment j in firm i at time t and n is the number of segments. Second, we determine whether each segment has high (low) opportunities relative to other firm segments based on positive (negative) deviation from the passive capital allocation. The signed deviation for each segment is estimated as follows: (3.2) where is the q segment j in firm i for year t and of is the asset-weighted average q of the firms' remaining segments. Lastly the firm-level capital allocation efficiency is calculated as a weighted average of deviation across each segment of the firm: (3.3) where ' is the assets of segment j in firm i at time t and n is the number of segments. We also use the measure of cross-divisional transfers by Rajan, Servaes and Zingales, (2000). They develop a measure to capture fund transfers to/from a division using investment

16 ratio of single-segment in the same industry as a benchmark. Transfers to divisions are defined as: ), (3.4) where ss refers to the single-segment, j denotes segment, i denotes firms and t denotes t. is segment j total share of firm's assets. Then, we measure the efficiency of cross-divisional transfers by summing all transfers made to high-productivity and low-productivity divisions. 3.5 Control Variables and Fixed Effects In segment-level regressions we control for segment size and cash flow. We use also use data from Compustat Annual Database to control for firm-level leverage and Tobin's Q. In firm-level regressions, we control for firm size, cash flow, leverage and Tobin's Q. Also, we include firm or segment fixed effects to control for time-invariant differences across firms or segments. 4 Methodology and Descriptive Statistics 4.1 Model Our first model investigates the impact of policy uncertainty on the excess value of diversified firms relative to single-segment firms using the following model, (3.5) where i indexes firm, t indexes year and i is an indicator variable which accounts for firm fixed effects. Y is either the excess value of the diversified firms by Berger and Ofek (1997) or a measure of internal capital market efficiency (Cho, 2015; Rajan, Servaes and Zingales, 2000). X is a vector of control variables, namely, size, cash flow, leverage and Tobin's Q. We also cluster by both firm and year using two-way clustering. Two-way clustering allows us to account for potential correlation among error terms both within firm and across time (Cameron, Gelbach, and Miller, 2006).

17 Next, we examine resource allocation of diversified firms relative to single-segment firms using segment level. Specifically, we investigate whether the relationship between capital expenditures and economic policy uncertainty is weaker for diversified firms relative to single-segment firms using the following model: (3.6) where i indexes segment, t indexes year. i is an indicator variables that accounts for firm (or segment) fixed effects, respectively. measures the segment-level investment rate of firms. measures policy uncertainty and iv is a dummy which equals to one if the firms is diversified and zero otherwise. X is a vector of control variables, namely, segment size, segment cash flow, firm leverage and firm Tobin's Q. Finally, we cluster by both firm and year using two-way clustering. 4.2 Summary Statistics In Panel A of Table 1 we report summary statistics of the key variables of our model for the full sample. We find that the mean and standard deviation of the overall BBD index is 101 and 21, respectively. These values indicate that BBD varies significantly over the years. In particular, Figure 1 shows that policy uncertainty increases during events such as wars, financial crisis and elections. We also find that 39.3% of the firms in our sample are diversified. In Panel B examine the correlations between the components of the BBD Index. Although the overall component is highly correlated with the each component, in particular the news component, each component does contain some unique information. 4.3 Univariate Analysis

18 We compare single-segment and diversified firms at the segment-level. Panel C shows that on average, segments of diversified firms have larger sales and assets than singlesegment firms. Importantly, we find that segment capital expenditures scaled by segment sales, is greater for single-segment firms. This finding indicates that on average diversified firms invest less than single-segment firms. Also diversified firms have greater cash flow and sales growth than single-segment. Next, we examine firm-level variables using quarterly data. We find that diversified firms invest less in intangible capital, hold less cash and more likely to pay dividends, repurchase shares, issue debt and issue equity. 5 Empirical Results 5.1 Diversification Discount and Internal Capital Markets In this section, we examine the impact of economic policy uncertainty on the diversification discount and internal capital market. Lang and Stulz (1994) and Berger and Ofek (1997) compute the excess value of diversified firms relative to single-segment firms and document the presence of the diversification discount. However, Kuppuswamy and Villalonga (2015) find that the diversification discount was reduced during the financial crisis. Thus, if diversified firms are more beneficial when policy uncertainty is high then we would expect the value of diversification to be greater. We use the measure of excess value by Berger and Ofek (1997) and examine whether policy uncertainty decreases the diversification discount. These results are reported in Panel A of Table 2. In Column (1) we find that the coefficient of PU is positive and statistically significant, indicating that the value of diversification increases when policy uncertainty is high. For robustness, we examine the impact of each component of the policy uncertainty index and find consistent results. Overall, our results imply that policy uncertainty reduces the diversification discount.

19 Moreover, the literature contains mixed evidence on whether the diversification discount exist (Villalonga, 2004; Campa and Kedia, 2002). Firms operating at a discount have negative excess value indicating that these firms would be more efficient if operating as single-segment firms. On the other hand, firms operating at a premium have positive excess value implying that diversifying is more efficient for these firms. Thus, we expect that these more efficient diversified firms are better at withstanding policy uncertainty. We examine the impact of policy uncertainty on excess value for firms with a premium (Column 5) or discount (Column 6) separately. Our results show that the impact of policy uncertainty is more pronounced for firms with a diversification premium. These results indicate that diversified firms with more efficient internal capital markets are better able to withstand periods of higher policy uncertainty. To examine further examine the impact on internal capital markets we use the measure of capital allocation efficiency by Cho (2015) and cross-divisional transfers by Rajan, Servaes and Zingales (2000). These results are reported in Panel B of Table 2. Our findings show that policy uncertainty increases capital allocation efficiency of diversified firms during high periods of uncertainty. We also find that policy uncertainty increases the transfer resources to high productivity segments. These results suggest that diversified firms transfer resources among segments and use their internal capital market to finance investments when policy uncertainty reduces external financing. Overall, our results provide evidence that corporate diversification is more valuable when policy uncertainty increases and improves the efficiency of internal capital markets. 5.2 Impact on Segment-Level Investments First, we estimate the impact of policy uncertainty on investments of single-segment and diversified firms separately. Then, we use the full sample and estimate equation (1) using Ordinary Least Squares. These results are reported in Table 2. In Columns (1) and (2) the

20 variable of interest is. The coefficient of this variable represents the impact of policy uncertainty on capital expenditures. Column (1) shows that economic policy uncertainty significantly decreases investments of single-segment firms. We scale our variables by their standard deviations for ease of interpretation, thus our results indicate that capital expenditures decrease by 0.10 standard deviations the next quarter. However, the coefficient of is negative but economically smaller in Column (2) indicating that policy uncertainty impacts investments of diversified firms less. In Column (3), our variables of interest are and. The coefficient of is negative and statistically significant. Consistent with our summary statistics, this finding suggests that diversified firms invest less on average. The interaction variable is positive and statistically significant, indicating that diversification weakens the negative impact of policy uncertainty on investments. Taken together, our results show that capital expenditures of diversified firms are less sensitive to changes in economic policy uncertainty compared to single-segment firms. For robustness, in Columns 4, 5 and 6, we control for segment fixed effects instead of firm fixed effects and find consistent results. Also, in Panel B of Table 2, we examine the impact of each component of the BBD index. We find that similar results with most of the explanatory power being captured by the news component. For brevity, we use the overall BBD index in subsequent tests Controlling for Investment Opportunities and Economic Uncertainty In our baseline model, we do not include year fixed effects since this would mechanically absorb most of the explanatory power of the policy uncertainty variable. Bloom (2014) posits that economic policy uncertainty is countercyclical and could also proxy for availability of investment opportunities. Hence, following Gulen and Ion (2016) we include several variables to control for expectations about future investment opportunities. First, we

21 capture expected GDP growth using the one-year-ahead GDP forecasts from the Philadelphia Federal Reserve s biannual Livingstone survey. This variable is measured biannually as a percentage change between the mean GDP forecast and the current GDP level. In our analysis, we use the yearly average. Second, we use the Conference Board s monthly Leading Economic Index which consist of ten macroeconomic indicators that are associated with future GDP. We use the year-on-year log change of this index in our analysis. Third, we use Michigan Consumer Confidence Index from the University of Michigan to capture the future economic expectations of consumers. The BBD index could also be proxy for general economic uncertainty and not only policy-related uncertainty. Following Bloom (2009) we include macroeconomic measures of uncertainty. First, we use the coefficient of variation in GDP forecasts from the Livingstone survey of professional forecasters to calculate a proxy of uncertainty about future economic growth. Second, we use the within-quarter cross-sectional standard deviation of firm-level profit growth to proxy for uncertainty about future profitability. The monthly cross-sectional standard deviation of stock returns and the VXO (implied volatility) index from the Chicago Board Options Exchange are used to proxy for uncertainty within equity markets. We also use the index by Jurado, Ludvigson, and Ng (2015) which measures time-varying macroeconomic uncertainty. We include the proxies for both investment opportunities and general economic uncertainty as additional controls in our baseline model and report these results in Table 3, Panel A. Our results show that the impact of policy uncertainty on investments is greater for single-segment firms. Additionally, the interaction variable, remains positive indicating that corporate diversification reduces the impact of policy uncertainty. These results are consistent with our baseline results indicating that the BBD index captures

22 information not contained in proxies for future investment opportunities or overall economic uncertainty Instrumental Variable Analysis To account for endogeneity concerns we use an instrumental variable approach. We follow Kaviani, Kryzanowski, Maleki and Savor (2017) who use the relative legislative power of the Democratic and Republican parties each year. We use data from the Swank (2013) Comparative Political Parties Dataset to construct this variable. In the United States the Democratic and Republican parties are viewed as centre and right leaning. To estimate the relative legislative power of the majority and minority party, we calculate the difference in the percentage of legislative seats that are occupied by each party. This instrumental variable should be significantly related to policy uncertainty and only impact corporate investments through its relationship with economic policy uncertainty. We report the second stage results in Table 4, Panel B for single-segment and diversified separately. The F-statistics in our first stage are significant and our instrument is highly correlated with policy uncertainty indicating that our instrument is not weak. Our results show that the negative impact of policy uncertainty on investments remains greater for single-segment firms compared to diversified firms Exclusion of Firms Undertaking Diversifying Mergers We account for the fact that some firms will diversify when external market frictions increase. Matvos, Seru and Silva (2018) find that firms become more diversified during periods of higher uncertainty. Thus, we verify that our results are not driven by these firms choosing to diversify during periods of higher uncertainty by excluding all firms who undertake diversifying mergers each quarter and re-estimating our baseline model. These results are reported in Table 4, Panel C. We find that the results remain consistent, indicating

23 that our results are not driven by firms self-selecting to be diversified when uncertainty increases Unrelated and Related Diversification The relatedness of a firm's segments is a proxy for the extent of a firm's diversification strategy. This degree of relatedness could also affect the relationship between policy uncertainty and capital expenditures. We investigate the impact of related and unrelated diversification. Unrelated diversification indicates that business segments of the firm differ more in terms of segment characteristics such as cash flows, investments opportunities or industry. Unrelated business segments are less likely to have correlated cash flows or investment opportunities. Studies such as Duchin (2010) claim that correlation in investment opportunities and cash flows of different segments of the firm could impact the cost and benefits of internal capital markets. Berger and Ofek (1995) find that related diversification is more valuable. Khanna and Tice (2001) find that internal capital markets are more likely to be efficient for related diversifiers. Unrelated diversification is defined as firms that report two or more business segments at the two-digit SIC level. Related diversification is defined as all other diversified firms. First, we examine the impact of policy uncertainty on corporate investments for unrelated diversified firms and related diversified firms separately. These results are reported in Column (1) and Column (2) of Panel A, Table 5. The coefficient of the PU variable is statistically insignificant for both unrelated and related diversification indicating that policy uncertainty does not impact investments of these firms. In Column (3), we estimate our baseline model but include dummy variables to capture both unrelated and related diversification. We find that only the diversification dummy for unrelated diversification is statistically significant. We also find that the interaction of unrelated diversification and policy uncertainty is positive and statistically significant. In Columns (4), (5) and (6) we

24 obtain consistent results using segment fixed effects instead of firm fixed effects. In general our findings indicate that unrelated diversification helps mitigate the impact of policy uncertainty on investments High versus Low Uncertainty and the Financial Crisis In this section, we investigate whether our results are driven by high or low periods of uncertainty. We divide our sample into periods of high and low uncertainty based on the median and re-estimate our baseline regression. Panel B of Table 5 reports the impact of policy uncertainty on capital expenditures during high and low periods of uncertainty for diversified firms relative to single-segment firms. The interaction of the diversification dummy and policy uncertainty is positive and significant for high periods of uncertainty and insignificant for low periods of uncertainty. Taken together, these results indicate that corporate diversification is more likely to mitigate the impact of policy uncertainty on capital expenditures during higher periods of policy uncertainty. Also, we examine the impact of the financial crisis. Kuppuswamy and Villalonga (2015) find that the financial crisis significantly increase the excess value of firms. We define a crisis indicator as one if the calendar year falls between 2007 and We also interact this indicator variable with our policy uncertainty measure and re-estimate our model. These results are reported in Panel B of Table 5. We find that the indicator variables and the interactions with the policy uncertainty variables are not significant. These results indicate that our results are driven by periods of higher policy uncertainty but not the financial crisis Financial Constraints In this section, we examine the impact of policy uncertainty on financially constrained and non-financially constrained firms. Financial constraints may limit the amount of resources firms have to spend on investments. To measure financial constraints we use the text-based financial constraints measure by Maksimovic and Hoberg (2015) and the KZ index

25 by Kaplan and Zingales (1997). Higher indices indicate that a firm is more financially constraint. We report these results in Table 6. Our results show the impact of policy uncertainty on segment level investments is greater for financially constrained firms. Importantly, the interaction term of diversification dummy and policy uncertainty is positive and significant for financially constrained firms but insignificant for non-financially constrained firms. These results indicate that the impact of diversification on the relationship between investments and policy uncertainty is more pronounced for financially constrained firms Impact on Core and Politically Sensitive Segments Agency conflicts could impact the investment sensitivity to policy uncertainty of a segment. Duchin and Sosyura (2013) find that segment managers with more social connections with the CEO receive more capital. Rajan, Servaes and Zingales (2000) argue that power struggles between segment managers could result in resources being diverted to certain segments. Moreover, diversified firms reallocate resources across divisions when external markets become more constrained (Matvos and Seru, 2014). We compare the impact of policy uncertainty on the investments of the core segment and non-core segments. The core segment is usually the more influential segment and has greater access to resources. However, firm might use resources intended for the core segment to subsidize non-core segments during times of higher uncertainty. As a result, we expect the core segment to be more impacted by policy uncertainty compared to the other segments. We define the core segment as the segment with the largest sales and examine the impact of economic policy uncertainty on investments of core segments and non-core segments. These results are reported in Panel C, Table 6. In Column (1) and Column (2) we use only the core segment and find that the impact is negative and statistically significant. However, Columns (3) and (4) show that impact of economic policy uncertainty is negative and statistically insignificant

26 for non-core segments. These results indicate that firms are probably reluctant to cut investments in non-core segment and use cash flows from core segments to subsidize investments of the non-core segments. Therefore, the non-core segments are not significantly impacted by economic policy uncertainty while policy uncertainty significantly decreases investments in the core segments. Consistent with studies such as Matvos and Seru (2014), our results provide evidence that firms reallocate resources across divisions when external market frictions increase. Next, we examine the impact of policy uncertainty on capital investments for segments in politically sensitive industries relative to segments in non-politically sensitive industries. Julio and Yook (2010) find that investment cycles are more pronounced in politically sensitive industries. Following Herron, Lavin, Cram and Silver (1999) we classify segments in tobacco products, pharmaceuticals, health care services, defense, petroleum and natural gas, telecommunications and transportation industries as politically sensitive. Then investigate the impact of policy uncertainty on investments of segments in politically sensitive and non-politically sensitive industries. We report these results in Panel D of Table 6. Our findings show that that the impact policy uncertainty on investments is negative and statistically significant for politically sensitive segments while statistically insignificant for non-politically sensitive segments. 6 Impact on Additional Real and Financial Decisions Our findings so far suggest that diversified firms are better equipped to weather policy uncertainty. In particular, impact of policy uncertainty on investments is mitigated by related diversification. Also, investments of core segments and non-politically sensitive segments are less impact by policy uncertainty. In our previous tests, we focus on capital expenditures due to the limitation of segment-level data. In this section, we use quarterly data to examine the impact policy uncertainty on additional real and financial decisions. First, we examine the

27 impact on intangible capital. Next, we investigate the financial decisions of the firms. Specifically, we examine the impact on cash holdings, payouts, debt issues and equity issues. Similar to our previous models, we examine the impact on single-segment and diversified firms separately then use the full sample. Also, to account for the possible persistent impact of policy uncertainty, we examine the impact of policy uncertainty up to four quarters ahead. 6.1 Impact on Intangible Capital In this section, we examine the relative impact of policy uncertainty on intangible capital of diversified firms relative to single-segment. Intangible capital defined as the sum of the stock of organizational capital and R&D scaled by lagged assets. To measure the stock of organization capital we use the firm-specific measurement of organization capital using Selling, General and Administrative (SGA) expenses by Lev and Radhakrishnan (2005) and Eisfeldt and Papanikolaou (2013). The stock of organization capital is measured recursively using the following formula, (3.7) where is the depreciation rate of organization capital and is the consumer price index. The initial stock of organization capital is computed as follows, (3.8) where, g is the average real growth rate of firm-level SGA expenses. This initial value is computed for the first year in which SGA expense has a non-missing value in the Compustat database and the subsequent missing values become zero. We use an average real growth rate of firm-level SG&A expenses is 10%. Following Eisfeldt and Papanikolaou (2013) we use a depreciation rate of 15%. This depreciation rate is also used by the U.S. Bureau of

28 Economic Analysis (BEA) to estimate R&D capital in We follow a similar method to measure the stock of R&D. Panel A of Table 7 examines the impact of policy uncertainty on intangible capital. Column (1) and (2) indicate that policy uncertainty increases investment in intangible capital and this increase is greater for single-segment firms. Next, we examine the impact using the full sample and find that the increase in intangible capital is reduced for diversified firms up to four quarters ahead. Taken together, our results show that the impact of policy uncertainty on intangible capital is significantly less for diversified firms relative to single-segment firms Impact on Financing Decisions To capture financing outcomes we examine cash holdings, payouts, debt issues and equity issues. Cash holdings as ratio of cash and short term investments divided by lagged total assets. We define ), as one if the year-over-year change in dividends is negative and zero otherwise. We identify firms that repurchase shares if the amount spent on the purchase of common and preferred is greater than zero, during the fiscal quarter t and define the probability of repurchasing shares, ), is one if the firm repurchases shares and zero otherwise. We follow McKeon (2015) to measure both debt issues and equity issues. The probability of issuing debt, ), is one if the change in total debt is positive and zero otherwise. The probability of issuing equity, ), is one if net firm-initiated issues is strictly positive and zero otherwise. To obtain firm-initiated issues we compute the funds received from the sale of common shares as the difference between the sale of common and preferred shares and the sale of preferred shares if this difference is positive and zero otherwise. Then we define these issues as firm-initiated if the ratio of equity issues to the end-of-year market equity is greater than 3%. 12 Our results are robust to a depreciation rate between 10% and 50%.

29 We examine the impact on cash holdings, payouts, debt issues and equity issues in Table 7. Panel B reports the impact on cash holdings. Our results indicate that policy uncertainty increases cash holdings and this increase is greater for diversified firms. Panel C and Panel D examine the impact on payouts. We find policy uncertainty is associated with a decrease in dividends and share repurchases. This decrease is greater for diversified firms relative to single-segment firms. Next, we examine the impact on debt and equity issuance. These results are reported in Panel E and Panel F. We find that policy uncertainty decreases debt issuance for only diversified firms while the impact on equity issuance is not significant. Overall, the results indicate that the policy uncertainty impacts the financing decisions of diversified firms and single-segment firms differently. 6.3 Robustness Checks We include the proxies for both investment opportunities and general economic uncertainty as additional controls in our baseline model and report these results in Table 8. Panel A. For brevity we only report the impact one quarter ahead. Our results show that the impact of policy uncertainty on diversified and single-segment firms is consistent with our previous results indicating that policy uncertainty captures information not contained in proxies for future investment opportunities or overall economic uncertainty. In Panel B of Table 8 we use the instrumental variable approach using the relative legislative power as an instrument and report the second stage results. The F-statistics in our first stage are significant and our instrument is highly correlated with policy uncertainty indicating that our instrument is not weak. Our results show that the impact of policy uncertainty on real and financing outcomes remains consistent. We also examine the impact of policy uncertainty on real and financial decisions of diversified firms relative to single-segment firms after excluding firms that undertake a diversifying merger that quarter. We find that the results remain consistent, indicating that our results are not driven by firms that self-select to be

30 diversified when uncertainty increases. Overall, these tests indicate that our results are not driven general uncertainty, endogeneity or self-selection. Additionally, we examine the impact of related and unrelated diversification and report these results in Table 9. Our results indicate that the impact of policy uncertainty on intangible capital, cash holdings, payouts is driven by both related and unrelated diversification while debt issues is only impacted by unrelated diversification. Next, we examine the impact policy uncertainty during high and low periods of policy uncertainty. These results are reported in Table 10. The impact on payouts and debt issuance is more pronounced during high periods of policy uncertainty. We also examine the impact of the financial crisis in Panel C, Table 10. Our results indicate that the financial crisis had a significant impact cash holdings repurchases and debt issuance. 7 Conclusion In this study, we examine the impact of economic policy uncertainty on investments of diversified firms compared to single-segment firms. We use the policy uncertainty index developed by Baker, Bloom and Davis (2016). This index consist of three components which capture uncertainty in news, future tax changes and forecaster disagreement. We show that policy uncertainty increases the excess value of diversified firms. We use the excess value measure by Berger and Ofek (1995) to examine the impact of policy uncertainty on the diversification discount. Our results show that policy uncertainty increases the value of the firm. Also, we find that policy uncertainty is associated with an increase in capital allocation efficiency (Cho, 2015) and increase in transfers to high productivity segments (Rajan, Servaes and Zingales, 2000). Overall, these results show that policy uncertainty increases internal capital market efficiency and corporate diversification makes firms more valuable during periods of higher uncertainty.

31 Next, we examine the impact of policy uncertainty on corporate decisions. Using segment-level data, we find that the impact of policy uncertainty on investments is more pronounced for single-segment firms and corporate diversification mitigates the negative effect of policy uncertainty on investments. Also, we find consistent results using each component of the overall BBD index to measure policy uncertainty. Our results are robust to a number of endogeneity and robustness tests. First, we control for investment opportunities and general economic uncertainty. Second we follow an instrumental variable approach using relative legislative power as an instrument for policy uncertainty. Third, we exclude firms that undertook diversifying mergers to control for potential self-selection to diversify. Fourth, we show that the unrelated diversification mitigates the impact of policy uncertainty on investments. Also, the impact of diversification is more pronounced during periods of higher uncertainty and not driven by the 2008 financial crisis. Lastly, we find that the impact of policy uncertainty on investments is greater for financially constrained segments, core segments and segments in politically sensitive industries. We also examine the impact on additional firms decisions using quarterly data. We find that policy uncertainty increases intangible capital cash holdings and decrease payouts and debt issues for diversified firms relative to single-segment firms. We verify these results by including investment and overall economic uncertainty controls, conducting instrumental variable analysis and excluding firms that undergo diversifying mergers. Additionally, we investigate unrelated and related diversification and the impact during the high and low uncertainty and the 2008 financial crisis period. Overall the results of these endogeneity and robustness checks are consistent with our main results. This paper contributes to literature on the benefits of corporate diversification and shows that corporate diversification could mitigate negative impacts of policy uncertainty. We show that corporate diversification is particularly valuable when policy uncertainty

32 increases. Also, our findings improve the understanding of the effect of the economic policy uncertainty on corporate decisions. References Atanassov, Julian, Brandon Julio, and Tiecheng Leng, 2016, The bright side of political uncertainty: The case of R&D, Working paper. University of Nebraska, University of Oregon, and Sun Yat-sen University. Baker, Scott R., Nicholas Bloom, and Steven J. Davis, 2016, Measuring economic policy uncertainty, Quarterly Journal of Economics 131, Berger, Philip G., and Eli Ofek, 1995, Diversification's effect on firm value, Journal of Financial Economics 37, Billett, Matthew T., and David C. Mauer, 2003, Cross-subsidies, external financing constraints, and the contribution of the internal capital market to firm value, Review of Financial Studies 16, Bloom, Nicholas, 2009, The impact of uncertainty shocks, Econometrica 77, Bloom, Nicholas, Max Floetotto, Nir Jaimovich, I. Saporta Eksten, and Stephen Terry, 2014, Really uncertain business cycles, US Census Bureau Center for Economic Studies Paper No. CESWP, Bonaime, Alice A., Huseyin Gulen, and Mihai Ion, 2017, Does policy uncertainty affect mergers and acquisitions?, Journal of Financial Economics, forthcoming. Bond, Stephen R., and Jason G. Cummins, 2004, Uncertainty and investment: An empirical investigation using data on analysts' profits forecasts, Working paper, FEDS Brogaard, Jonathan, and Andrew Detzel, 2015, The asset-pricing implications of government economic policy uncertainty, Management Science 61, Campa, Jose Manuel, and Simi Kedia., 2002, Explaining the diversification discount, Journal of Finance 57, Cho, Young Jun, 2015, Segment disclosure transparency and internal capital market efficiency: Evidence from SFAS No. 131, Journal of Accounting Research 53, Duchin, Ran, and Denis Sosyura, 2013, Divisional managers and internal capital markets, Journal of Finance 68, Duchin, Ran, 2010, Cash holdings and corporate diversification, Journal of Finance 65, Eisfeldt, Andrea L., and Dimitris Papanikolaou, 2013, Organization capital and the cross section of expected returns, Journal of Finance 68,

33 Fluck, Zsuzsanna, and Anthony W. Lynch, 1999, Why do firms merge and then divest? A theory of financial synergy, Journal of Business 72, Gilchrist, Simon, Jae W. Sim, and Egon Zakrajšek, 2014, Uncertainty, financial frictions, and investment dynamics, National Bureau of Economic Research No. w Herron, Michael C., James Lavin, Donald Cram, and Jay Silver, 1999, Measurement of political effects in the United States economy: a study of the 1992 presidential election, Economics & Politics 11, Hoberg, Gerard, and Vojislav Maksimovic, 2014, Redefining financial constraints: A textbased analysis, Review of Financial Studies 28, Jens, Candace E., 2017, Political uncertainty and investment: Causal evidence from US gubernatorial elections, Journal of Financial Economics 124, Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and takeovers, American Economic Review 76, Jensen, Michael C., and Kevin J. Murphy, 1990, Performance pay and top-management incentives, Journal of Political Economy 98, Julio, Brandon, and Youngsuk Yook, 2012, Political uncertainty and corporate investment cycles, Journal of Finance 67, Jurado, Kyle, Sydney C. Ludvigson, and Serena Ng, 2015, Measuring uncertainty, American Economic Review 105, Kaplan, Steven N., and Luigi Zingales, 1997, Do investment-cash flow sensitivities provide useful measures of financing constraints?, Quarterly Journal of Economics 112, Kaviani, Somayeh Mahsa, Lawrence Kryzanowski, Hosein Maleki, and Pavel G. Savor, 2017, Policy uncertainty and corporate credit spreads, Working paper, Temple University Khanna, Naveen, and Sheri Tice, 2001, The bright side of internal capital markets, Journal of Finance 56, Kuppuswamy, Venkat, and Belén Villalonga, 2015, Does diversification create value in the presence of external financing constraints? Evidence from the financial crisis, Management Science 62, Lamont, Owen, 1997, Cash flow and investment: Evidence from internal capital markets, Journal of Finance 52, Lang, Larry HP, and Rene M. Stulz, 1994, Tobin's q, corporate diversification, and firm performance, Journal of Political Economy 102, Leahy, John V., and Toni M. Whited, 1995, The effect of uncertainty on investment: Some stylized facts, National Bureau of Economic Research. No. w4986. Lev, Baruch, and Suresh Radhakrishnan, 2005, The valuation of organization capital. In measuring capital in the new economy, University of Chicago Press

34 Lewellen, Wilbur G, 1971, A pure financial rationale for the conglomerate merger, Journal of Finance 26, Matvos, Gregor, Amit Seru, and Rui C. Silva, 2018, Financial market frictions and diversification, Journal of Financial Economics 127, Matvos, Gregor, and Amit Seru, 2014, Resource allocation within firms and financial market dislocation: Evidence from diversified conglomerates, Review of Financial Studies 27, McKeon, Stephen, 2015, Measuring equity issuance, Working paper, University of Oregon. Miller, Merton H., and Franco Modigliani, 1961, Dividend policy, growth, and the valuation of shares, Journal of Business 34, Nguyen, Nam H., and Hieu V. Phan, 2017, Policy uncertainty and mergers and acquisitions, Journal of Financial and Quantitative Analysis 52, Pástor, Ľuboš, and Pietro Veronesi, 2013, Political uncertainty and risk premia, Journal of Financial Economics 110, Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000,The cost of diversity: The diversification discount and inefficient investment, Journal of Finance 55, Santalo, Juan, and Manuel Becerra, 2008, Competition from specialized firms and the diversification performance linkage, Journal of Finance 63, Scharfstein, David S., 1998, The dark side of internal capital markets II: Evidence from diversified conglomerates, National Bureau of Economic Research No. w6352. Scharfstein, David S., and Jeremy C. Stein, 2000, The dark side of internal capital markets: Divisional rent seeking and inefficient investment, Journal of Finance 55, Shin, Hyun-Han, and René M. Stulz, 1998, Are internal capital markets efficient?, Quarterly Journal of Economics 113, Shleifer, Andrei, and Robert W. Vishny, 1989, Management entrenchment: The case of manager-specific investments, Journal of Financial Economics 25, Stein, Jeremy C., 1997, Internal capital markets and the competition for corporate resources, Journal of Finance 52, Stein, Luke, C.D. and Elizabeth Stone, 2012, The effect of uncertainty on investment. Hiring and R&D: Causal Evidence from Equity Options, Working paper, Arizona State University Stulz, René M., 1990, Managerial discretion and optimal financing policies, Journal of Financial Economics 26, Swank, Duane. (2013). Comparative political parties dataset: Electoral, legislative, and government strength of political parties by ideological group in 21 capitalist democracies,

35 Electronic Database. Milwaukee: Department of Political Science, Marquette University Villalonga, Belen, 2004, Does diversification cause the "diversification discount"?, Financial Management 5 27.

36 Table 1: Summary Statistics This table reports the summary statistics for key variables used in the analysis. Panel A: Summary Statistics for the full sample (1) (2) (3) (4) VARIABLES Mean SD Min Max Overall Index (PU) News component Tax component ,310 Gov component CPI component Diversified Panel B: Correlations News component Tax component Gov component CPI component Overall Index (PU) Overall Index (PU) 1 News component Tax component Gov component CPI component Panel C: Comparison of single-segment and diversified firms at segment-level (1) (2) (3) (4) (5) (6) Single-segment Diversified Difference VARIABLES N Mean SD N Mean SD Segment Variables Assets 110, ,170 66,483 1,008 2, *** Sales 110, ,698 66, , *** Capxs 110, , Capex/Sales 97, , *** Sales Growth 97, , *** Cash Flow 101, , Median Ind. Q 109, , *** Panel D: Comparison of single-segment and diversified firms at the firm-level (1) (2) (3) (4) (6) (6) Single-segment Diversified Difference VARIABLES N Mean SD N Mean SD Real and Financing Decisions Intangible Capital 131, , *** Cash Holdings 312, , *** P(Dividends) 313, , *** P(CutDividends) *** P(Repurchases) 313, , *** P(Equity Issues) 313, , *** P(Debt Issues) 313, , *** Control Variables Sales Growth 303, , *** Cash Flow 288, , *** Tobin Q 282, , ***

37 Table 2: Diversification Discount and Internal Capital Markets This table investigates the impact of policy uncertainty on the diversification discount and internal capital markets. Panel A examine the impact of policy uncertainty on the diversification discount. The dependent variable is the excess value of diversified firms. Panel B examines the impact on internal capital markets. The dependent variable in Columns (1) and (2) is the measure of overall efficiency by Cho (2015) and crossdivisional transfers by Rajan et. al (2000) in Columns (3) and (4). PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). The control variables are sales growth, cash flow, industry Q and election indicator. Variables are defined in Appendix C. Firm fixed effects are included but unreported. Robust standard errors are clustered by both firm and year and reported in parentheses. The time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Impact on Excess Value Dependent Variable ExcessValue (1) (2) (3) (4) (3) (4) Overall Forecaster Index News Tax-Code Disagreement Overall Index Premium Discount PU 0.159*** 0.125*** * 0.107** *** * (0.0387) (0.0370) ( ) (0.0436) (0.0302) (0.0292) Tobin Q 0.181*** 0.181*** 0.179*** 0.180*** *** 0.126*** (0.0259) (0.0258) (0.0248) (0.0255) ( ) (0.0125) CashFlow -6.6e-05*** -6.7e-05*** -6.9e-05*** -6.7e-05*** ** -2.31e-05 (1.56e-05) (1.61e-05) (1.57e-05) (1.48e-05) ( ) (3.74e-05) SalesGrowth * ( ) ( ) ( ) ( ) ( ) ( ) Election (0.0163) (0.0156) (0.0209) (0.0179) (0.0101) (0.0160) Observations 18,618 18,618 18,618 18,618 28,835 34,387 R-squared Controls YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel B: Impact on Internal Capital Market Dependent Variable Capital Allocation Efficiency Cho (2015) Cross Divisional Transfers Rajan et al. (2000) (2) (3) (4) (5) High Policy Uncertainty Low Policy Uncertainty High Productivity Low Productivity PU * * (0.0231) (0.0200) (4.452) (4.453) Tobin Q * ** ( ) ( ) (0.0302) (0.972) CashFlow 3.64e e-05** * (2.41e-05) (6.11e-06) ( ) ( ) SalesGrowth ** *** ( ) ( ) (0.0121) (0.0751) Election ( ) ( ) (1.909) (2.577) Observations 7,701 10,245 6,243 11,617 R-squared Controls YES YES YES YES Firm FE YES YES YES YES

38 Table 3: Impact of Policy Uncertainty on Segment-Level Capital Expenditures This table investigates the impact of policy uncertainty on capital investments. Columns (1) and (4) examine the impact on single-segment firms and Columns (2) and (5) examine the impact on diversified firms. Columns (3) and (6) use the full sample to estimate our baseline model in Equation (1). The dependent variable is the one year lead of segment investments scaled by segment sales. PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Panel A uses the overall baseline index and Panel B examines the individual components of the BBD Index. Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are segment sales growth, segment cash flow, industry Q, election indicator and GDP growth. Variables are defined in Appendix C. Firm fixed effects are included in Columns (1), (2) and (3) and segment fixed effects are included in Columns (4), (5) and (6) but unreported. Robust standard errors are clustered by both firm and year and reported in parentheses. The time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Singlesegment Diversified Full Sample PU ** ** *** ** *** *** (0.0379) (0.0178) (0.0408) (0.0369) (0.0175) (0.0392) Diverse ** ** (0.186) (0.171) Diverse*PU 0.101** ** (0.0406) (0.0369) Observations 54,030 30,388 84,839 53,801 29,237 83,560 R-squared Controls YES YES YES YES YES YES Firm FE YES YES YES NO NO NO Segment FE NO NO NO YES YES YES Panel B: Impact on Components of the overall policy index Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) News Tax-Code Forecaster Disagreement News Tax-Code Forecaster Disagreement PU *** *** *** *** *** 0.111*** (0.0336) ( ) (0.0273) (0.0329) ( ) (0.0256) Diverse ** *** (0.148) (0.0266) (0.180) (0.141) (0.0249) (0.178) Diverse*PU *** ** *** ** (0.0322) ( ) (0.0396) (0.0304) ( ) (0.0387) Observations 84,839 84,839 84,839 83,560 83,560 83,560 R-squared Controls YES YES YES YES YES YES Firm FE YES YES YES NO NO NO Segment FE NO NO NO YES YES YES

39 Table 4: Robustness Checks This table reports the results from our robustness checks. Panel A reports the estimation of the baseline model after including controls for investment opportunities and overall economic uncertainty. Panel B reports the second stage regression of the instrumental variable analysis using the relative legislative power of the Democratic and Republican parties each year as an instrument for policy uncertainty. Panel C reports the estimation after excluding firms that undertake diversifying mergers. Columns (1) and (4) examine the impact on single-segment firms and Columns (2) and (5) examine the impact on diversified firms. Columns (3) and (6) use the full sample to estimate our baseline model in Equation (1). The dependent variable is the one year lead of segment capital expenditures scaled by segment sales. PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are segment sales growth, segment cash flow, industry Q, election indicator. Variables are defined in Appendix C. Firm fixed effects are included in Columns (1), (2) and (3) and segment fixed effects are included in Columns (4), (5) and (6) but unreported. Robust standard errors are clustered by both firm and year and reported in parentheses. The time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Controlling for investment opportunities and overall economic uncertainty Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) Single-segment Diversified Full Sample Single-segment Diversified Full Sample PU ** *** ** ** (0.0648) (0.0223) (0.0516) (0.0657) (0.0281) (0.0551) Diverse * (0.258) (0.206) Diverse*PU * ** (0.0558) (0.0445) Consumer *** ** *** -8.42e *** Index ( ) ( ) ( ) ( ) ( ) ( ) Economic * Index (0.0611) (0.0230) (0.0421) (0.0640) (0.0257) (0.0507) Expected * * * ** *** ** GDPGrowth (0.0191) ( ) (0.0138) (0.0203) ( ) (0.0156) GDP * Dispersion ( ) ( ) ( ) ( ) ( ) ( ) JLN ** * ** ** (0.244) (0.0972) (0.177) (0.249) (0.0976) (0.200) VXO ( ) ( ) ( ) ( ) ( ) ( ) Returns SD 1.608*** 0.352** 1.177*** 1.656*** 0.502** 1.345*** (0.260) (0.157) (0.198) (0.279) (0.187) (0.227) Profit e GrowthSD ( ) (9.93e-05) ( ) ( ) ( ) ( ) Observations 60,847 25,370 86,694 60,461 24,213 85,274 R-squared Controls YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Segment FE NO NO NO YES YES YES

40 Panel B: Instrumental Variable Approach Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Singlesegment Diversified Full Sample PU ** ** ** ** ** (0.204) (0.116) (0.185) (0.218) (0.125) (0.207) Diverse * * (0.776) (0.586) Diverse*PU 0.354* 0.257* (0.173) (0.130) Observations 60,847 25,370 86,694 60,461 24,213 85,274 R-squared Controls YES YES YES YES YES YES Invest. Controls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Firm FE NO NO NO YES YES YES Segment FE NO NO NO YES YES YES Panel C: Exclusion of diversifying mergers Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Singlesegment Diversified Full Sample PU * ** * ** (0.0634) (0.0237) (0.0503) (0.0638) (0.0319) (0.0533) Diverse * (0.284) (0.227) Diverse*PU * (0.0614) (0.0490) Observations 52,257 21,110 73,846 51,861 19,850 72,346 R-squared Controls YES YES YES YES YES YES Invest. Controls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Firm FE NO NO NO YES YES YES Segment FE NO NO NO YES YES YES

41 Table 5: Unrelated and Related Diversification and High versus Low Periods of Uncertainty and Financial Crisis Panel A investigates the impact of related and unrelated diversification on the relationship between policy uncertainty and capital expenditures. Unrelated diversification is one if the firm has more than two segments at the two-digit SIC level and related diversification is all others. Panel B investigates the impact of periods of high policy uncertainty on capital expenditures. Panel C investigates the impact of periods of low policy uncertainty on capital expenditures. Panel D investigates the impact of the financial crisis. The dependent variable is the one year lead of segment capital expenditures scaled by segment sales. PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are segment sales growth, segment cash flow, industry Q, election indicator and GDP growth. Variables are defined in Appendix C. Firm fixed effects are included in Columns (1), (2) and (3) and segment fixed effects are included in Columns (4), (5) and (6) but unreported. Robust standard errors are clustered by both firm and year and reported in parentheses. The time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Unrelated and related diversification Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) Unrelated Related Full Sample Unrelated Related Full Sample PU ** ** ** (0.0324) (0.0573) (0.0557) (0.0289) (0.0589) (0.0588) Unrelated * ** (0.206) (0.186) Unrelated*PU 0.102** ** (0.0447) (0.0400) Related (0.248) (0.204) Related*PU (0.0538) (0.0444) Observations 24,547 5,780 84,839 23,505 5,546 83,560 R-squared Controls YES YES YES YES YES YES Invest. Controls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Firm FE YES YES YES NO NO NO Segment FE NO NO NO YES YES YES Panel B: Impact of high and low periods of uncertainty and financial crisis Dependent Variable Capex/Sales (1) (2) (3) (4) (5) (6) Low Policy High Policy Uncertainty Uncertainty Financial Crisis Log PU * * ** *** ** ** (0.0995) (0.0964) (0.0725) (0.0668) (0.0541) (0.0569) Diverse * * * ** (0.537) (0.467) (0.422) (0.406) (0.193) (0.172) Diverse*LogPU 0.231* 0.179* ** ** (0.113) (0.0995) (0.0955) (0.0920) (0.0419) (0.0371) Crisis (0.404) (0.416) Crisis* PU

42 (0.0911) (0.0941) Observations 44,115 42,435 62,152 60,189 84,839 83,560 R-squared Firm Controls YES YES YES YES YES YES Invest. Controls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Firm FE YES NO YES NO YES NO Segment FE NO YES NO YES NO YES

43 Table 6: Financial Constraints and Core or Politically Sensitive Segments Panel A examines the impact of financial constraints using Maksimovic and Hoberg, 2015 text-based measure and Panel B examines the impact of financial constraints using KZ (Kaplan and Zingales, 2000) measure. Panel C investigates the impact of policy uncertainty on corporate investments of the core and non-core segments. Panel D investigates the impact of policy uncertainty on corporate investments of the segments in politically sensitive industries and segments in other industries. The dependent variable is the lead of segment capital expenditures scaled by segment sales. PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are segment sales growth, segment cash flow, industry Q, election indicator and GDP growth. Variables are defined in Appendix C. Firm fixed effects are included in Columns (1) and (3) and segment level are included in Columns (2) and (4) but unreported. Robust standard errors are clustered by both firm and year and reported in parentheses. The time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Impact of Financial Constraints using Maksimovic and Hoberg, 2015 text-based measure Dependent Variable Capex/Sales (1) (2) (3) (4) Non-Financially Constrained Financially Constrained PU * * *** *** (0.0438) (0.0433) (0.0915) (0.0961) Diverse * ** ** (0.159) (0.162) (0.388) (0.393) Diverse*PU ** 0.205** (0.0374) (0.0377) (0.0839) (0.0845) Observations 16,774 16,202 15,553 14,930 R-squared Controls YES YES YES YES Invest. Controls YES YES YES YES Macro Controls YES YES YES YES Firm FE YES NO YES NO Segment FE NO YES NO YES Panel B: Impact of Financial Constraints using KZ (Kaplan and Zingales, 1997) Dependent Variable Capex/Sales (1) (2) (3) (4) Non-Financially Constrained Financially Constrained PU * * *** *** (0.0428) (0.0459) (0.0703) (0.0760) Diverse * ** (0.207) (0.191) (0.293) (0.275) Diverse*PU ** 0.160** (0.0451) (0.0419) (0.0644) (0.0604) Observations 32,395 31,573 25,938 25,061 R-squared Controls YES YES YES YES Invest. Controls YES YES YES YES Macro Controls YES YES YES YES Firm FE YES NO YES NO

44 Segment FE NO YES NO YES Panel C: Impact on Core and non-core segments Dependent Variable Capex/Sales Core Segment Non-Core Segments (1) (2) (3) (4) PU ** ** * (0.0274) (0.0264) (0.0450) (0.0416) Observations 14,163 13,721 18,933 18,013 R-squared Controls YES YES YES YES Invest. Controls YES YES YES YES Macro Controls YES NO YES NO Firm FE NO YES NO YES Segment FE NO YES NO YES Panel D: Impact on politically sensitive and non-politically sensitive Dependent Variable Capex/Sales Politically Sensitive Non-Politically Sensitive (1) (2) (3) (4) PU ** ** (0.130) (0.131) (0.0261) (0.0235) Observations 4,435 4,308 20,748 19,905 R-squared Controls YES YES YES YES Invest. Controls YES YES YES YES Macro Controls YES YES YES YES Firm FE YES NO YES NO Segment FE NO YES YES YES

45 Table 7: Impact on Additional Firms Decisions This table investigates the impact of policy uncertainty on real outcomes. Panel A examines the impact on intangible capital, Panel B examines cash holdings, Panel C examines the propensity to pay dividends, Panel D examines the propensity to repurchase shares, Panel E examines the propensity to issue debt and Panel F examines the propensity to issue equity. Columns (1) uses only single-segment firms and Columns (2) uses only diversified firms. Columns (1), (2) and (3) dependent variable has a lead of one period. In Columns (4), (5) and (6), the dependent variable has a lead of two, three and four respectively. PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are Tobin's Q, cash flow, sales growth, election dummy, and GDP growth. Variables are defined in Appendix C. Firm fixed effects and quarter dummies are included in all specifications but unreported. Robust standard errors are clustered by both firm and year and reported in parentheses. The data are quarterly and the time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Impact on Intangible Capital (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Dependent Variable PU 0.191*** ** 0.199*** 0.172*** 0.149*** 0.126** (0.0427) (0.0337) (0.0469) (0.0482) (0.0484) (0.0496) Diverse 0.460*** 0.393*** 0.341*** 0.309*** (0.103) (0.105) (0.100) (0.0998) Diverse*PU *** *** *** *** (0.0229) (0.0234) (0.0223) (0.0223) Observations 120,281 58, , , , ,680 R-squared Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel B: Impact on Cash Holdings (1) (2) (3) (4) (5) (6) Single-segment Diversified Full Sample Dependent Variable PU ** *** * ** ** *** (0.0213) (0.0155) (0.0202) (0.0226) (0.0248) (0.0242) Diverse *** ** (0.0624) (0.0694) (0.0744) (0.0731) Diverse*PU ** (0.0134) (0.0151) (0.0162) (0.0160) Observations 250, , , , , ,662 R-squared Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES

46 Panel C: Impact on the propensity to cut dividends (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Dependent Variable PU *** ** *** *** ( ) (0.0167) ( ) ( ) ( ) ( ) Diverse *** *** *** *** (0.0588) (0.0623) (0.0590) (0.0553) Diverse*PU *** *** *** *** (0.0127) (0.0135) (0.0128) (0.0120) Observations 251, , , , , ,749 R-squared Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel D: Impact on the propensity to repurchase shares (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Dependent Variable PU *** * * ** (0.0188) (0.0166) (0.0196) (0.0209) (0.0234) (0.0257) Diverse 0.298*** 0.297*** 0.299*** 0.230*** (0.0485) (0.0503) (0.0508) (0.0535) Diverse*PU *** *** *** *** (0.0106) (0.0110) (0.0111) (0.0117) Observations 251, , , , , ,749 R-squared Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel E: Impact on the propensity to issue debt (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Dependent Variable PU *** *** *** *** *** *** (0.0126) (0.0159) (0.0136) (0.0131) (0.0131) (0.0118) Diverse 0.149*** 0.171*** 0.160*** 0.143*** (0.0495) (0.0461) (0.0489) (0.0508) Diverse*PU *** *** *** ***

47 (0.0107) ( ) (0.0106) (0.0109) Observations 251, , , , , ,749 R-squared Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel F: Impact on the propensity to issue equity (1) (2) (3) (4) (5) (6) Singlesegment Diversified Full Sample Dependent Variable PU * * ( ) ( ) ( ) ( ) ( ) ( ) Diverse (0.0274) (0.0261) (0.0263) (0.0229) Diverse*PU e ( ) ( ) ( ) ( ) Observations 251, , , , , ,749 R-squared Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES

48 Table 8: Robustness Checks and Additional Firm Decisions This table reports the results of various robustness checks. Panel A reports the estimation of the previous models after including controls for several proxies for investment opportunities and overall economic uncertainty. In Panel B we implement an instrumental variable approach with the relative legislative power of the Democratic and Republican parties each year as an instrument for policy uncertainty. We report the second stage regression. Panel C we exclude all firms that undertake diversifying mergers. The dependent variable is the lead of intangible capital (Column 1), cash holdings (Column 2), propensity to pay dividends (Column 3) propensity to repurchase shares (Column 4), propensity to issue debt (Column 5) and propensity to issue equity (Column 6). PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are Tobin's Q, cash flow, sales growth, election dummy, and GDP growth. Variables are defined in Appendix C. Firm fixed effects and quarter dummies are included in all specifications but unreported. Robust standard errors are clustered by both firm and quarters and reported in parentheses. The data are quarterly and the time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Controlling for investment opportunities and overall economic uncertainty Dependent Variable (1) (2) (3) (4) (5) (6) PU *** * (0.0355) (0.0213) ( ) (0.0204) (0.0171) ( ) Diverse 0.400*** *** *** 0.356*** 0.185*** (0.0926) (0.0616) (0.0563) (0.0430) (0.0467) (0.0272) Diverse*PU *** *** *** *** *** (0.0204) (0.0134) (0.0122) ( ) (0.0101) ( ) Consumer *** *** ** *** * ** Index ( ) ( ) ( ) ( ) ( ) ( ) Economic *** *** ** Index (0.0319) (0.0145) ( ) (0.0146) (0.0120) ( ) Expected *** *** * *** GDP Growth (0.0118) ( ) ( ) ( ) ( ) ( ) GDP *** *** * *** Dispersion ( ) ( ) ( ) ( ) ( ) ( ) JLN *** * 0.552*** *** (0.161) (0.0972) ( ) (0.0765) (0.0744) (0.0352) VXO * * ( ) ( ) ( ) ( ) ( ) ( ) Returns SD 1.292*** * *** ** 0.153*** (0.245) (0.161) ( ) (0.105) (0.0910) (0.0532) Profit *** *** 4.54e e e-06 Growth SD ( ) ( ) ( ) (8.04e-05) (9.96e-05) (4.17e-05) *** Observations 163, ,288 ( ) 433, , ,439 R-squared *** Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES

49 Panel B: Instrumental Variable Analysis (1) (2) (3) (4) (5) (6) Dependent Variable PU (0.204) (0.132) (0.174) (0.188) (0.143) (0.0925) Diverse 0.680* ** *** 0.466*** * (0.350) (0.234) (0.161) (0.151) (0.150) (0.0880) Diverse*PU ** * *** *** * (0.0772) (0.0518) (0.0354) (0.0333) (0.0333) (0.0193) Observations 163, , , , , ,439 R-squared FirmControls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Investment Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel C: Exclusion of firms undergoing diversifying mergers (1) (2) (3) (4) (5) (6) Dependent Variable PU * (0.0355) (0.0212) (0.312) (0.0195) (0.0175) ( ) Diverse 0.409*** *** *** 0.186*** (0.0933) (0.0612) (0.160) (0.0425) (0.0473) (0.0271) Diverse*PU *** *** *** *** (0.0206) (0.0133) (0.0350) ( ) (0.0103) ( ) Observations 160, , , , , ,111 R-squared FirmControls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Investment Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES

50 Table 9: Unrelated and Related Diversification and Additional Firm Decisions This table investigates the impact of alternative measures of diversification and self-selection correction for the decision to diversify. Panel A examines the impact of related and unrelated diversification. Unrelated diversification is one if the firm has more than two segments at the two-digit SIC level and related diversification is all others. The dependent variable is the lead of intangible capital (Column 1), cash holdings (Column 2), propensity to pay dividends (Column 3) propensity to repurchase shares (Column 4), propensity to issue debt (Column 5) and propensity to issue equity (Column 6). PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are Tobin's Q, cash flow, sales growth, election dummy, and GDP growth. Variables are defined in Appendix C. Firm fixed effects and quarter dummies are included in all specifications but unreported. Robust standard errors are clustered by both firm and quarters and reported in parentheses. The data are quarterly and the time period used is 1988 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Dependent Variable (2) (3) (4) (5) (6) (7) PU *** * (0.0355) (0.0213) ( ) (0.0204) (0.0171) ( ) Unrelated 0.394*** *** *** 0.359*** 0.191*** (0.0942) (0.0612) (0.0569) (0.0439) (0.0477) (0.0276) Unrelated*PU *** *** *** *** *** (0.0207) (0.0133) (0.0124) ( ) (0.0103) ( ) Related 0.506** ** * 0.311** (0.202) (0.161) (0.0980) (0.120) (0.110) (0.0719) Related*PU *** ** ** (0.0438) (0.0348) (0.0213) (0.0265) (0.0238) (0.0159) Observations 163, , , , , ,439 R-squared FirmControls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Investment Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES

51 Table 10: High and Low Periods of Policy Uncertainty and Financial Crisis and Additional Firm Decisions This table investigates the impact of on real and financial decisions during high and low periods of uncertainty for diversified firms relative to single-segment. Panel A examines the impact of during high uncertainty. Panel B examines the impact of during low uncertainty. Panel C examines the impact of the crisis. The dependent variable is the lead of intangible capital (Column 2), cash holdings (Column 3), propensity to pay dividends (Column 4) propensity to repurchase shares (Column 5), propensity to issue debt (Column 6) and propensity to issue equity (Column 7). PU is the natural logarithm of the BBD Index by Baker, Bloom and Davis (2016). Diverse is one if a firm has more than two segments at the four-digit SIC level. The control variables are Tobin's Q, cash flow, sales growth, election dummy, and GDP growth. We also control for the proxies of investment opportunities and overall economic uncertainty. Variables are defined in Appendix C. Firm fixed effects and quarter dummies are included in all specifications but unreported. Robust standard errors are clustered by both firm and quarters and reported in parentheses. The data are quarterly and the time period used is 1987 to ***,**,and * denote the significance at the 1%, 5% and 10% levels respectively. Panel A: Impact during high uncertainty (2) (3) (4) (5) (6) (7) Dependent Variable PU *** (0.0929) (0.0311) (0.0222) (0.0349) (0.0391) (0.0101) Diverse *** 0.527*** 0.509*** (0.215) (0.135) (0.130) (0.119) (0.117) (0.0594) Diverse*PU *** *** *** (0.0446) (0.0282) (0.0272) (0.0250) (0.0245) (0.0122) Observations 63, , , , , ,046 R-squared FirmControls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Investment Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel B: Impact during low uncertainty (2) (3) (4) (5) (6) (7) Dependent Variable PU 0.153*** *** (0.0419) (0.0355) (0.0178) (0.0232) (0.0200) (0.0100) Diverse 0.363* *** (0.185) (0.177) (0.145) (0.0847) (0.0951) (0.0567) Diverse*PU ** *** (0.0421) (0.0397) (0.0331) (0.0191) (0.0214) (0.0129) Observations 99, , , , , ,976 R-squared

52 FirmControls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Investment Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES Panel C: Impact during 2008 financial crisis (2) (3) (4) (5) (6) (7) Dependent Variable PU *** * (0.0332) (0.0193) ( ) (0.0202) (0.0141) ( ) Diverse 0.409*** *** *** 0.350*** 0.183*** (0.0929) (0.0609) (0.0558) (0.0431) (0.0483) (0.0267) Diverse*PU *** *** *** *** *** (0.0205) (0.0133) (0.0121) ( ) (0.0105) ( ) Crisis*PU *** * ** *** (0.0745) (0.0341) (0.0204) (0.0346) (0.0298) (0.0125) Crisis *** * 0.344** *** (0.341) (0.163) (0.0931) (0.161) (0.146) (0.0579) Observations 163, , , , , ,439 R-squared FirmControls YES YES YES YES YES YES Macro Controls YES YES YES YES YES YES Investment Controls YES YES YES YES YES YES Quarter Dummies YES YES YES YES YES YES Firm FE YES YES YES YES YES YES

53 Figure 1: Policy Uncertainty Index Graph of the economic policy uncertainty index by Baker, Bloom and Davis (2016). Recessions periods are shaded. The time period used is 1985 to 2010.