Internationalization and zero leverage

Figures & data

Figure 1. The evolution of foreign sales and zero leverage. This figure presents the evolution of the median foreign sales ratio (left axis), the number of zero-leverage firm-year observations (right axis), and the number of MNC firm-year observations (right axis) for our sample of UK companies between 1990 and 2018. We define zero-leverage firm-year observations as those when market leverage equals zero. We define MNC firm-year observations as those when the foreign sales-to-total sales ratio exceeds 0.50. We obtain firm identification and accounting data from Bloomberg and foreign sales data from WorldScope.

Table 1. Number of firm-year observations for domestic and multinational firms by leverage status.

		Panel A: DOM					Panel B: MNC
ICB Code	Industry Name	All	Levered		Zero Leverage		All	Levered		Zero Leverage
		No.	No.	%	No.	%	No.	No.	%	No.	%
–	All	14,430	12,530	86.83	1,900	13.17	7,363	6,302	85.59	1,061	14.41
0001	Oil and Gas	452	340	75.22	112	24.78	653	515	78.87	138	21.13
1000	Basic Materials	503	428	85.09	75	14.91	927	816	88.03	111	11.97
2000	Industrials	4,954	4,528	91.40	426	8.60	2,348	2,189	93.23	159	6.77
3000	Consumer Goods	2,126	1,889	88.85	237	11.15	776	682	87.89	94	12.11
4000	Health Care	824	638	77.43	186	22.57	668	525	78.59	143	21.41
5000	Consumer Services	3,734	3,306	88.54	428	11.46	892	779	87.33	113	12.67
6000	Telecommunications	217	198	91.24	19	8.76	92	74	80.43	18	19.57
9000	Technology	1,620	1,203	74.26	417	25.74	1.007	722	71.70	285	28.30

Table 1 presents the distribution of the sample by industry, multinational status, and zero-leverage status. The sample is an unbalanced panel comprising 2,516 firms and 21,793 firm-year observations between 1990 and 2018. Bloomberg is the source of firm and accounting data. Column 1 presents Industry Classification Benchmark (ICB) codes and column 2 the industry name. Panel A shows domestic firms (DOM, foreign sales ratio < 0.50) and Panel B multinational firms (MNC, foreign sales ratio > 0.50).

Table 2. Characteristics of levered and zero-leverage firms by MNC status.

	Panel A: Levered			Panel B: Zero Leverage			Panel C: Δ Means (t-stats)
	1. All	2. LDOM	3. LMNC	4. All	5. ZDOM	6. ZMNC	7. (1–4)	8. (2–3)	9. (2–5)	10. (3–6)	11. (5–6)
LEV	0.229	0.236	0.220	–	–	–	–	3.910^a	–	–	–
FSALES	0.380	0.134	0.778	0.382	0.101	0.840	−0.201	−199.436^a	6.051^a	−8.646^a	−90.807^a
SIZE	5.184	4.812	5.789	3.521	3.397	3.722	28.946^a	−23.047^a	20.609^a	21.479^a	−3.962^a
CASH	0.101	0.095	0.111	0.287	0.286	0.289	−48.841^a	−6.216^a	−38.532^a	−30.150^a	−0.268
TANG	0.536	0.525	0.554	0.389	0.368	0.423	23.025^a	−6.192^a	18.513^a	13.799^a	−4.390^a
PROF	0.082	0.080	0.085	−0.004	0.005	−0.019	14.544^a	−1.206	10.115^a	10.629^a	1.334
NDTS	0.034	0.035	0.033	0.026	0.028	0.023	10.467^a	4.044^a	7.409^a	7.562^a	3.333^a
GROW	2.828	2.809	2.860	3.689	3.845	3.433	−5.583^a	−0.451	−5.027^a	−2.529^b	1.379
ACTAX	0.182	0.176	0.190	0.147	0.169	0.112	2.618^a	−1.387	0.467	3.591^a	2.698^a
Z	3.125	3.101	3.163	8.987	8.914	9.106	−30.586^a	−0.648	−23.890^a	−19.099^a	−0.240
HH	0.097	0.084	0.119	0.125	0.107	0.154	−8.248^a	−14.976^a	−6.043^a	−5.778^a	−6.806^a

Table 2 presents mean values of key variables for sample firms. We compute statistics using observations for which there are no missing values in any of the variables (case-wise deletion of observations). This produces an unbalanced panel comprising 1,580 firms and 11,412 firm-year observations between 1990 and 2018. Bloomberg is the source of firm and accounting data and WorldScope the source of foreign sales data. Panel A presents statistics for levered firms. Column 1 presents statistics for all levered firms while columns 2 and 3 present statistics for levered-domestic (LDOM) and levered-multinational (LMNC) firms, respectively. Panel B presents statistics for zero-leverage firms. Column 4 presents statistics for all zero-leverage firms while columns 5 and 6 present statistics for zero-leverage domestic (ZDOM) and zero-leverage multinationals (ZMNC), respectively. We classify a firm as MNC if its foreign sales ratio exceeds 0.50. Panel C presents t-statistics from tests of the hypothesis that the means of respective pairs of subsets are equal. LEV is leverage (ratio of short-term and long-term debt-to-short-term and long-term debt plus market capitalization). FSALES is foreign sales (ratio of foreign sales-to-total sales). SIZE is firm size (natural logarithm of total assets). CASH is the cash ratio (ratio of cash and cash equivalents-to-total assets). TANG is tangibility (ratio of fixed assets-to-total assets). PROF is profitability (ratio of operating income plus depreciation-to-total assets). NDTS is non-debt tax shields (ratio of depreciation-to-total assets). GROW is growth (ratio of market value-to-book value). ACTAX is actual tax (ratio of cash paid for taxes, net of any tax refunds-to-pre-tax income). Z is Z-score (following Altman’s specification). HH is a Herfindahl-Hirschman index capturing the level of industry concentration. a/b/c denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 3. Logistic regressions of firms’ zero-leverage status.

	Degree of Internationalization
	Model 1 (MNC 20)		Model 2 (MNC 50)		Model 3 (MNC 60)		Model 4 (MNC 80)
	A	B	A	B	A	B	A	B
Constant	−0.014	−0.174	−0.157	−0.500	−0.159	−0.521	−0.162	−0.327
	(−0.027)	(−0.321)	(−0.302)	(−0.946)	(−0.307)	(−0.993)	(−0.311)	(−0.625)
MNC	−0.047	0.013	0.228	0.475	0.249	0.460	0.369	0.525
	(−0.670)	(0.045)	(3.148)^a	(2.564)^b	(3.385)^a	(1.466)	(4.382)^a	(1.344)
SIZE	−0.335	−0.291	−0.348	−0.288	−0.348	−0.306	−0.346	−0.322
	(−18.198)^a	(−9.886)^a	(−19.081)^a	(−12.174)^a	(−19.138)^a	(−13.215)^a	(−19.249)^a	(−15.372)^a
CASH	4.230	3.822	4.177	4.227	4.173	4.389	4.138	4.297
	(16.686)^a	(10.818)^a	(16.563)^a	(14.043)^a	(16.538)^a	(14.213)^a	(16.375)^a	(15.038)^a
TANG	−0.043	−0.377	−0.074	−0.384	−0.077	−0.268	−0.113	−0.201
	(−0.247)	(−1.638)	(−0.422)	(−1.869)^c	(−0.442)	(−1.272)	(−0.648)	(−1.015)
PROF	0.245	0.555	0.266	0.586	0.264	0.571	0.273	0.543
	(1.257)	(1.901)^c	(1.384)	(2.286)^b	(1.374)	(2.359)^b	(1.422)	(2.388)^b
NDTS	−9.465	−6.717	−9.110	−7.817	−9.095	−8.325	−8.910	−8.783
	(−6.702)^a	(−3.793)^a	(−6.492)^a	(−4.860)^a	(−6.487)^a	(−5.149)^a	(−6.362)^a	(−5.765)^a
GROW	−0.026	−0.015	−0.026	−0.019	−0.026	−0.020	−0.025	−0.021
	(−3.543)^a	(−1.620)	(−3.569)^a	(−2.277)^b	(−3.548)^a	(−2.443)^b	(−3.517)^a	(−2.723)^a
ACTAX	0.145	0.213	0.144	0.249	0.146	0.218	0.145	0.212
	(2.493)^b	(2.453)^b	(2.475)^b	(3.347)^a	(2.516)^b	(3.126)^a	(2.499)^b	(3.289)^a
Z	0.061	0.054	0.061	0.053	0.061	0.049	0.061	0.052
	(6.747)^a	(4.249)^a	(6.828)^a	(5.087)^a	(6.825)^a	(5.222)^a	(6.836)^a	(5.726)^a
HH	−1.656	−1.032	−1.581	−0.940	−1.599	−1.243	−1.655	−1.457
	(−1.398)	(−0.835)	(−1.339)	(−0.772)	(−1.356)	(−1.039)	(−1.392)	(−1.212)
MNC × SIZE		−0.074		−0.133		−0.106		−0.081
		(−2.017)^b		(−3.936)^a		(−2.916)^a		(−2.034)^b
MNC × CASH		0.985		0.086		−0.494		−0.422
		(1.990)^b		(0.193)		(−0.930)		(−0.681)
MNC × TANG		0.694		0.885		0.579		0.382
		(2.006)^b		(2.832)^a		(1.507)		(0.849)
MNC × PROF		−0.618		−0.775		−0.911		−1.080
		(−1.580)		(−2.003)^b		(−2.269)^b		(−2.389)^b
MNC × NDTS		−5.061		−2.593		−1.787		0.311
		(−1.732)^c		(−0.886)		(−0.548)		(0.083)
MNC × GROW		−0.025		−0.024		−0.024		−0.020
		(−1.694)^c		(−1.446)		(−1.392)		(−1.034)
MNC × ACTAX		−0.117		−0.271		−0.229		−0.342
		(−0.997)		(−2.305)^b		(−1.845)^c		(−2.364)^b
MNC × Z		0.015		0.024		0.044		0.048
		(0.809)		(1.247)		(1.969)^b		(1.666)^c
MNC × HH		−0.767		−0.453		0.279		−0.156
		(−1.327)		(−0.794)		−(0.479)		(−0.252)
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
No. of obs.	11,412	11,412	11,412	11,412	11,412	11,412	11,412	11,412
Pseudo R^2	0.237	0.240	0.238	0.243	0.238	0.242	0.239	0.243
AUC	0.855	0.856	0.854	0.856	0.855	0.857	0.855	0.857

Table 3 presents estimated coefficients from logistic regressions of firms’ zero-leverage status. We define zero-leverage firms as those with firm-year observations when market leverage equals zero. Column 1 identifies independent variables. Each subsequent column presents results from models using different thresholds of internationalization. We estimate two model specifications: Model A is the baseline and Model B is augmented with interaction effects. MNC is a binary variable equal to unity when the ratio of foreign sales-to-total sales exceeds a predefined threshold, zero otherwise. The thresholds are 0.20 in Model 1, 0.50 in Model 2, 0.60 in Model 3, and 0.80 in Model 4. SIZE is firm size (natural logarithm of total assets). CASH is the cash ratio (ratio of cash and cash equivalents-to-total assets). TANG is tangibility (ratio of fixed assets-to-total assets). PROF is profitability (ratio of operating income plus depreciation-to-total assets). NDTS is non-debt tax shields (ratio of depreciation-to-total assets). GROW is growth (ratio of market-to-book value). ACTAX is actual tax (ratio of cash paid for taxes, net of any tax refunds-to-pre-tax income). Z is Z-score (following Altman’s specification). HH is a Herfindahl-Hirschman index capturing the level of industry concentration. Industry dummies are set to one for each industry, zero otherwise. Figures in parentheses show z-statistics (robust standard errors and covariances). AUC is the area under the ROC curve. a/b/c denote significance at the 0.01/0.05/0.10 levels, respectively.

Table 4. Logistic regressions of firms’ zero-leverage status: sample subsets.

	Sample Subsets
	Model 1 (Constraints)		Model 2 (Size)		Model 3 (Profitability)		Model 4 (Actual Tax)
	High	Low	Large	Small	High	Low	High	Low
Constant	0.742	−4.800	−4.727	0.281	−3.579	1.679	−2.366	0.677
	(1.239)	(−4.141)^a	(−4.228)^a	(0.452)	(−4.042)^a	(2.592)^a	(−2.547)^b	(1.031)
MNC	0.259	0.867	3.455	0.466	0.753	−0.535	0.348	−0.160
	(0.757)	(0.879)	(2.998)^a	(−1.265)	(1.345)	(−1.308)	(0.555)	(−0.430)
SIZE	−0.136	−0.417	−0.385	−0.032	−0.229	−0.351	−0.335	−0.279
	(−3.497)^a	(−4.745)^a	(−3.430)^a	(−0.679)	(−5.700)^a	(−9.944)^a	(−8.174)^a	(−8.246)^a
CASH	3.671	6.427	6.793	3.910	6.022	2.830	6.093	3.171
	(10.957)^a	(8.280)^a	(8.403)^a	(11.627)^a	(12.411)^a	(7.699)^a	(10.986)^a	(8.558)^a
TANG	−0.685	0.605	0.511	−0.572	0.233	−0.667	0.143	−0.630
	(−2.793)^a	(1.093)	(0.906)	(−2.359)^b	(0.651)	(−2.329)^b	-(0.420)	(−2.158)^b
PROF	0.140	2.355	8.403	−0.030	0.235	−0.238	−1.306	−0.039
	(0.578)	(1.667)^c	(6.107)^a	(−0.125)	(0.184)	(−0.891)	(−1.407)	(−0.159)
NDTS	−9.447	−6.799	−15.413	−9.400	−4.343	−17.737	−1.385	−15.284
	(−4.961)^a	(−1.583)	(−3.023)^a	(−5.235)^a	(−1.685)^c	(−5.715)^a	(−0.498)	(−5.711)^a
GROW	−0.021	−0.047	−0.070	−0.015	−0.066	−0.024	−0.079	−0.006
	(−1.980)^b	(−2.103)^b	(−4.985)^a	(−1.546)	(−4.850)^a	(−2.040)^b	(−4.282)^a	(−0.638)
ACTAX	0.254	0.166	0.158	0.212	0.159	0.228	−0.217	0.383
	(3.103)^a	(0.967)	(0.795)	(2.507)^b	(0.865)	(2.560)^b	(−1.235)	(2.020)^b
Z	0.040	0.148	0.125	0.042	0.213	0.029	0.237	0.031
	(4.302)^a	(3.214)^a	(2.742)^a	(4.289)^a	(4.768)^a	(3.738)^a	(5.337)^a	(4.005)^a
HH	−4.110	7.595	6.083	−3.692	2.098	−3.512	0.616	−1.932
	(−3.013)^a	(3.599)^a	(2.920)^a	(−2.652)^a	(1.055)	(−2.397)^b	(0.312)	(−1.290)
MNC × SIZE	−0.100	−0.443	−0.669	−0.115	−0.186	−0.086	−0.250	−0.044
	(−1.615)	(−3.514)^a	(−4.071)^a	(−1.462)	(−3.227)^a	(−1.552)	(−4.093)^a	(−0.855)
MNC × CASH	0.029	1.540	−0.212	−0.190	−0.576	1.426	−1.803	1.203
	(0.053)	(1.194)	(−0.155)	(−0.345)	(−0.686)	(2.225)^b	(−1.705)^c	(1.957)^c
MNC × TANG	0.799	2.507	2.455	0.494	1.217	1.422	1.847	0.847
	(1.879)^c	(2.918)^a	(2.817)^a	(1.162)	(1.988)^b	(2.806)^a	(2.847)^a	(1.745)^c
MNC × PROF	−0.858	2.399	−6.622	−0.661	0.858	−0.601	0.507	−0.549
	(−2.472)^b	(1.273)	(−3.527)^a	(−1.801)^c	(0.451)	(−1.495)	(0.314)	(−1.507)
MNC × NDTS	−3.722	−4.414	11.225	−3.370	−8.120	−0.628	−7.765	3.013
	(−0.956)	(−0.729)	(1.760)^c	(−0.869)	(−1.922)^c	(−0.109)	(−1.479)	(0.670)
MNC × GROW	−0.030	−0.033	0.025	−0.031	−0.002	−0.026	0.001	−0.042
	(−1.516)	(−0.977)	(0.958)	(−1.621)	(−0.084)	(−1.207)	(0.018)	(−2.270)^b
MNC × ACTAX	−0.302	−0.055	0.040	−0.275	−0.125	−0.314	−0.351	−0.131
	(−2.167)^b	(−0.203)	(0.166)	(−1.814)^c	(−0.490)	(−1.924)^c	(−1.003)	(−0.431)
MNC × Z	0.009	0.016	−0.011	0.017	−0.013	0.014	0.090	0.018
	(0.543)	(0.287)	(−0.200)	(0.959)	(−0.205)	(0.930)	(1.304)	(1.170)
MNC × HH	0.054	−1.905	−1.914	−0.240	−1.170	0.127	0.201	−0.870
	(0.085)	(−1.520)	(−1.723)^c	(−0.368)	(−1.111)	(0.179)	(0.183)	(−1.269)
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
No. of obs.	5,926	5,486	5,706	5,706	5,706	5,706	5,706	5,706
Pseudo R^2	0.182	0.343	0.295	0.179	0.282	0.27	0.293	0.257
AUC	0.805	0.916	0.904	0.803	0.888	0.860	0.890	0.856

Table 4 shows estimated coefficients from logistic regressions of the augmented Model 2 of firms’ zero-leverage status for cohorts of firms. Model 1 compares high versus low financial constraints firms. Model 2 compares large versus small firms. Model 3 compares high versus low profitability firms. Model 4 compares high versus low actual tax firms. The median value of the pertinent variable in each model is the cut-off point for each cohort. MNC is a binary variable equal to unity when the ratio of foreign sales-to-total sales exceeds 0.50. SIZE is firm size (natural logarithm of total assets). CASH is the cash ratio (ratio of cash and cash equivalents-to-total assets). TANG is tangibility (ratio of fixed assets-to-total assets). PROF is profitability (ratio of operating income plus depreciation-to-total assets). NDTS is non-debt tax shields (ratio of depreciation-to-total assets). GROW is growth (ratio of market-to-book value). ACTAX is actual tax (ratio of cash paid for taxes, net of any tax refunds-to-pre-tax income). Z is Z-score (following Altman’s specification). HH is a Herfindahl-Hirschman index capturing the level of industry concentration. Industry dummies are set to one for each industry, zero otherwise. Figures in parentheses show z-statistics (robust standard errors and covariances). AUC is the area under the ROC curve. a/b/c denote significance at the 0.01/0.05/0.10 levels, respectively.

Table 5. Firm investment and zero leverage.

	${\mathit{\text{Invest}}}_{i,t+1}=\alpha +{\beta }_1{D}_{\mathrm{ZL}}+{\beta }_{2}C{F}_{i,t}+{\beta }_{3}C{F}_{i,t}{D}_{\mathrm{ZL}}+{\beta }_4{Q}_{i,t}+{\beta }_5{Q}_{i,t}{D}_{\mathrm{ZL}}+{\epsilon }_{i,t}$
	(2) ALL	(3) MNC	(4) DOM
Constant	0.043	0.042	0.044
	(83.183)^a	(48.614)^a	(63.782)^a
$D_{\mathrm{ZL}}$	0.005	−0.002	0.009
	(3.057)^a	(−0.749)	(4.617)^a
$\mathrm{CF}$	0.013	0.017	0.011
	(7.906)^a	(6.143)^a	(4.968)^a
$\mathrm{CF}\times D_{\mathrm{ZL}}$	−0.004	−0.009	−0.001
	(−1.238)	(−1.944)^c	(−0.332)
$Q$	0.005	0.003	0.006
	(18.342)^a	(8.167)^a	(15.242)^a
$Q\times D_{\mathrm{ZL}}$	−0.003	−0.001	−0.004
	(−5.949)^a	(−1.535)	(−5.783)^a
No. of obs.	18,745	6,170	12,575
R^2	0.024	0.022	0.024

Table 5 presents results from panel regressions of the Tobin's Q model of investment for firms in our sample with non-missing foreign sales data. The dependent variable $(\mathrm{Inves}{t}_{i,t+1})$ is firm investment, defined as the ratio of capital expenditures-to-total assets. Column 1 identifies the independent variables, while subsequent columns present results for different sub-samples. We define zero-leverage firms as those with firm-year observations when market leverage equals zero and specify a $D_{\mathrm{ZL}}$ binary equal to 1, zero otherwise. $C{F}_{i,t}$ is cash flow$,$ calculated as net income plus depreciation-to-total assets. $Q_{i,t}$ is the Tobin’s Q ratio, defined as market value of equity plus book value of debt-to-total assets. Column 2 presents the results for all firms. Columns 3 and 4 present results for multinational (MNC) and domestic (DOM) firms, respectively; MNC (DOM) are defined as firm-year observations where the foreign sales-to-total sales ratio exceeds (is below) 0.50. Figures in parentheses show t-statistics (robust standard errors and covariances). a/b/c denote significance at the 0.01/0.05/0.10 levels, respectively.

Table 6. Robustness checks.

			Degree of Internationalization
			Model 1	Model 2	Model 3	Model 4
			MNC 20	MNC 50	MNC 60	MNC 80
Alternative Proxies		Obs.	Coef.	Coef.	Coef.	Coef.
(1)	ZL binary: Almost zero-leverage	11,412	−0.080	0.114^c	0.131^c	0.205^a
(2)	MNC binary: foreign assets ratio > threshold	9,367	−0.117	0.237^b	0.245^b	0.442^a
(3)	MNC binary: foreign assets ratio > 0 & foreign sales ratio > threshold	9,196	−0.009	0.196^b	0.242^a	0.391^a
(4)	MNC binary: both foreign assets ratio & foreign sales ratio > threshold	9,196	−0.130	0.362^a	0.350^a	0.490^a
Classification
(5)	Firth logit (penalized likelihood)	11,412	−0.047	0.228^a	0.249^a	0.369^a
Model Specification
(6)	Lagged independent variables ($X_{t-1}$)	10,198	−0.111	0.221^a	0.274^a	0.364^a
(7)	Standard errors clustered at firm-level	11,412	−0.047	0.228^c	0.249^c	0.369^b
(8)	Standard errors clustered at year-level	11,412	−0.047	0.228^a	0.249^a	0.369^a

Table 6 presents robustness tests to the baseline model (Model A) provided in Table 3. To conserve space, we suppress results for the intercept and control variables, although they are included in the estimations. Column 1 describes the robustness checks in consecutively numbered rows. Row 1 considers almost zero-leverage firms, defined as those with firm-year observations where $L_{i,t}<0.05$ and specifies a binary variable equal to unity to identify almost zero-leverage firms, zero otherwise. Rows 2–4 employ alternative proxies for internationalization, specifying a binary variable equal to unity to identify MNC when (i) the foreign assets-to-total assets ratio exceeds the predefined thresholds (row 2), (ii) the foreign assets-to-total assets ratio exceeds zero and the foreign sales-to-total sales ratio exceeds the predefined thresholds (row 3), and (iii) both the foreign assets-to-total assets ratio and the foreign sales-to-total sales ratio exceed the predefined thresholds (row 4). Rows 5–8 present alternative model specifications; row 5 considers the Firth model, row 6 introduces lagged independent variables, while rows 7 and 8 employ clustered standard errors at the firm- and year-level, respectively. Column 2 presents the number of firm-year observations, while each of the subsequent four columns presents coefficient estimates of the MNC binary for pertinent thresholds of internationalization. The thresholds are 0.20 in Model 1, 0.50 in Model 2, 0.60 in Model 3, and 0.80 in Model 4. a/b/c denote significance at the 0.01/0.05/0.10 levels (robust standard errors and covariances, unless otherwise stated), respectively.

Figure 2. Feature importance using random forest classifier. This figure presents results using a Random Forest classifier that evaluates the importance of baseline model variables (features) in discriminating between zero-leverage and levered firms. Feature importance scores are presented on the vertical axis. The features are presented on the horizontal axis and defined in Appendix Table A.1. and in Table 1. AUC is the area under the ROC curve.

Table A.1. Definitions of Variables and Key Output

Variable	Variable Full Name	Definition	Motivation/ CS Theories	Expected Sign	Observed Sign (ALL)	Observed Sign (DOM)	Observed Sign (MNC)	Sum Coeffs $\widehat{{\beta }_i}{X}_i+\widehat{{\beta }_j}{X}_i{D}_{\mathrm{MNC}}$
ZL	Zero leverage	Binary variable equals 1 if market leverage (short-term debt (BS047) and long-term debt (BS051)-to-short-term debt (BS047), long-term debt (BS051) and market capitalisation (RR902)) equals zero (0), 0 otherwise.
MNC	Internationalization	Binary variable equals 1 if the ratio of foreign sales-to-total sales (08731) exceeds a specified threshold, 0 otherwise. Source: WorldScope.	Trade-off Financial Constraints Financial Flexibility	– + +	+
SIZE	Size	Natural logarithm of total assets (BS035).	Pecking Order Trade-off Financial Constraints	+/– – –	–	–	–	–
CASH	Cash ratio	Ratio of cash and cash equivalents (BS010)-to-total assets (BS035).	Pecking Order Trade-off Financial Constraints^c Financial Flexibility	+ – + +	+	+	NS	+
TANG	Tangibility (asset structure)	Ratio of fixed tangible assets (RR879)-to-total assets (BS035).	Pecking Order Trade-off Financial Flexibility Financial Constraints	– – + –	NS	–	+	+
PROF	Profitability	Return on assets (operating income (IS033) plus depreciation (IS070)-to-total assets (BS035)).	Pecking Order Trade-off	+ –	NS	+	–	NS
NDTS	Non-debt tax shields	Ratio of depreciation (IS070)-to-total assets (BS035).	Trade-off	+	–	–	NS	–
GROW	Growth	Market-to-book value ratio (F0940) (market capitalization-to-total book value).	Pecking Order Trade-off Financial Flexibility	– + +	–	–	NS	–
ACTAX	Actual tax	Ratio of cash paid for taxes, net of any tax refunds (CF073)-to-pre-tax income (RR001).	Trade-off Financial Constraints	– +	+	+	–	NS
Z	Z-score	Altman’s (1968) Z-score:	Pecking Order Trade-off Financial Constraints	+ – –	+	+	+^†	+
		$Z=1.2X_1+1.4X_2+3.3X_3+0.6X_4+X_5$
		where X1 = ratio of working capital (RR150)-to-tangible assets (RR879); X2 = ratio of retained earnings (BS065)-to-tangible assets (RR879); X3 = earnings before interest and tax (RR803)-to-tangible assets (RR879); X4 = market value of equity (RR250)-to-total liabilities (RR005); X5 = sales (RR800)-to-tangible assets (RR879).
HH	Herfindahl-Hirschman index	HH measure of industry concentration as follows: $\mathrm{HH}=\sum _{i=1}^I{s}_{\mathrm{ijt}}^2$ where, $s_{i,t}$ is the market share of firm $i$ in industry $j$ and year $t$. We calculate HHI for each industry and year. Lower values imply many competing firms operate in an industry. Higher values infer that market share is concentrated in a few large firms.	Agency Theory Financial Constraints	– –	NS	NS	NS	NS

This table provides additional details on our variables obtained from Bloomberg and WorldScope (all variables sourced from Bloomberg except the foreign sales ratio; bracketed codes are respective Bloomberg or WorldScope variable codes). The Expected Sign column shows the expected signs on zero leverage. The columns on Observed Signs show signs observed in Table 3. ALL shows results from specification A; DOM and MNC present results from specification B from the standalone ($\widehat{{\beta }_i}{X}_i$) and interacted terms ($\widehat{{\beta }_j}{X}_i{D}_{\mathrm{MNC}}$), respectively. The final column shows signs of the summed coefficients ($\widehat{{\beta }_i}{X}_i+\widehat{{\beta }_j}{X}_i{D}_{\mathrm{MNC}}$) from specification B; significance is determined by tests of the null: $\widehat{{\beta }_i}{X}_i+\widehat{{\beta }_j}{X}_i{D}_{\mathrm{MNC}}=0$. NS indicates no significance.

^cCash can also proxy for future financial constraints.

^†Coefficient is statistically significant in models 3 and 4 (see Table 3).