The information content of corporate social responsibility disclosure in Europe: an institutional perspective

Figures & data

Table 1. Selection and distribution of our sample.

Panel A: Sample selection
	Total
Initial population of firm-year observations for 600 firms (2008-2016)	5400
Less: observations for which the annual report is not available	−377
= firm-year observations with available annual reports	5023
Less: observations that cannot be processed in textual analysis	−850
= firm-year observations with processable annual reports	4173
Less: observations with missing main variables	−212
Total sample	3961
Panel B: Sample distribution by industry group
	Frequency.
1 Consumer Non-durables	291
2 Consumer Durables	130
3 Manufacturing	515
4 Energy	92
5 Chemicals	209
6 Business equipment	260
7 Communication	227
8 Utilities	164
9 Wholesale	306
10 Healthcare	265
11 Finance	824
12 Other	678
Total sample	3961
Panel C: Sample distribution by countries
	Freq.
1 Austria	61
2 Belgium	102
3 Switzerland	351
4 Germany	267
5 Denmark	136
6 Spain	204
7 Finland	121
8 France	598
9 United Kingdom	1226
10 Ireland	71
11 Italy	148
12 Luxembourg	35
13 Netherlands	191
14 Norway	76
15 Portugal	25
16 Sweden	349
Total sample	3961

Figure 1. Topic-specific CSR disclosure by year.

Table 2. Descriptive Statistics for all variables.

	Mean	Std. Dev	Minimum	P25	Median	P75	Maximum	Observations
P	41.9968	99.7624	0.7497	6.5767	16.0843	37.2521	797.7345	3961
BVE	22.1115	44.6695	−0.0841	3.3875	8.2736	20.3687	311.8097	3961
EARN	2.4112	5.5593	0.0000	0.3300	0.9200	2.1800	43.4600	3961
POST	0.3575	0.4793	0.0000	0.0000	0.0000	1.0000	1.0000	3961
TENV	0.0940	0.0393	0.0005	0.0688	0.0940	0.1228	0.2509	3961
TSOC	0.0794	0.0339	0.0004	0.0559	0.0753	0.0990	0.2875	3961
TEMPL	0.1015	0.0432	0.0002	0.0683	0.0970	0.1330	0.2485	3961
THR	0.0323	0.0191	0.0003	0.0205	0.0294	0.0398	0.2504	3961
TCORR	0.0166	0.0103	0.0001	0.0098	0.0145	0.0209	0.1084	3961
TBRIB	0.0117	0.0070	0.0000	0.0066	0.0104	0.0150	0.0614	3961

This table presents descriptive statistics for the variables used in equations (1) and (2).

Table 3. The results of equations (1) and (2): nested price panel regressions Price is regressed on book value of equity and earnings, both per share and on the respective CSR disclosure measures T_i.

Equation	(1)	(2)	(2)	(2)	(2)	(2)	(2)
BVE	0.9662***	0.9559***	0.9602***	0.9563***	0.9638	0.9576***	0.9483***
	(3.5071)	(3.5084)	(3.5078)	(3.5074)	(3.5114)	(3.5049)	(3.4985)
EARN	2.7013***	2.6779***	2.6887***	2.6838***	2.6895	2.6863***	2.6706***
	(3.0271)	(3.0487)	(3.0505)	(3.0535)	(3.0381)	(3.0490)	(3.0684)
TENV		17.0286
		(0.7302)
TENV *POST		−119.7881***
		(−3.0810)
TSOC			47.3939*
			(1.8149)
TSOC*POST			−104.5475***
			(−2.7880)
TEMPL				27.1312
				(1.3825)
TEMPL*POST				−105.1878****
				(−3.4179)
THR					129.5665***
					(2.1932)
THR*POST					−169.4093***
					(−2.6723)
TCORR						187.2301*
						(1.9189)
TCORR*POST						−367.6679***
						(−3.1384)
TBRIB							78.3899
							(0.7799)
TBRIB*POST							−733.3985***
							(−4.4339)
Constant	1.0608	−0.2413	−2.1731	−1.3256	−2.3523	−1.1750	0.6983
	(0.2154)	(−0.0433)	(−0.3920)	(−0.2443)	(−0.4299)	(−0.2260)	(0.1407)
Cluster	FIRM	FIRM	FIRM	FIRM	FIRM	FIRM	FIRM
Firm fixed effects	YES	YES	YES	YES	YES	YES	YES
Year fixed effects	YES	YES	YES	YES	YES	YES	YES
Adj. R-squared	0.9543	0.9548	0.9546	0.9547	0.9545	0.9546	0.9551
F	18.46	16.97	17.11	16.70	17.62	16.67	17.41
Observations	3961	3961	3961	3961	3961	3961	3961

In this table, we used Ordinary Least Squares coefficient estimates and t-statistics (in parentheses) based on the standard variance estimator. ***, **, and * indicate statistical significance at the 1%, 5% and 10% levels (two-tailed) respectively. This table reports the results of the effect of a topic-specific CSR disclosure on share prices analyzed via panel with fixed effects for years and cross-section. A description of the variables can be found in Appendix I.

Table 4. Descriptive statistics for the R² of pooled equations (1) and (2) on the firm level.

	$R_{}^2\left(B\right)$	$R_{}^2\left(T_{ENV}\right)$	$R_{}^2\left(T_{SOC}\right)$	$R_{}^2\left(T_{EMPL}\right)$	$R_{}^2\left(T_{HR}\right)$	$R_{}^2\left(T_{CORR}\right)$	$R_{}^2\left(T_{BRIB}\right)$
Mean	0.4934	0.6147	0.6218	0.6182	0.6259	0.6251	0.6264
Std. Dev.	0.3773	0.3616	0.35929	0.3524	0.3502	0.3556	0.3504
Median	0.5820	0.7403	0.74398	0.7343	0.7411	0.7377	0.7456
Observations	3303	3303	3303	3303	3303	3303	3303

This table presents the results of incremental changes. The first column displays the adjusted R² for the pooled estimation of equation (1), whereas the remaining columns show the adjusted R² of how the respective textual disclosure affects prices. For this analysis, we used pooled regressions of equation (2).

Table 5. The results of equation (3): the incremental value-relevance of the institutional environment In columns (I) we regress R² _inc(T_i) on the institutional factors, controlling for industry and country-fixed effects; in columns (II) we also add financial characteristics including size, leverage and profitability.

	$R_{incr}^2\left(T_{ENV}\right)$		$R_{incr}^2\left(T_{SOC}\right)$		$R_{incr}^2\left(T_{EMPL}\right)$		$R_{incr}^2\left(T_{HR}\right)$		$R_{incr}^2\left(T_{CORR}\right)$		$R_{incr}^2\left(T_{BRIB}\right)$
	(I)	(II)	(I)	(II)	(I)	(II)	(I)	(II)	(I)	(II)	(I)	(II)
factor_csr	−0.1608***	−0.1668***	−0.2406***	−0.2467***	−0.2266***	−0.2343***	−0.2164***	−0.2229***	−0.1693***	−0.1759***	−0.2322***	−0.2388***
	(−4.1494)	(−4.3071)	(−6.2234)	(−6.3923)	(−6.1269)	(−6.3624)	(−5.4616)	(−5.6343)	(−4.2202)	(−4.3926)	(−5.8580)	(−6.0501)
factor_empl	−0.2443***	−0.2584***	−0.3158***	−0.3357***	−0.3255**	−0.3454***	−0.1649**	−0.1804***	−0.1322*	−0.1488**	−0.1580**	−0.1804***
	(−3.7194)	(−3.9309)	(−4.8196)	(−5.1264)	(−5.1931)	(−5.5268)	(−2.4557)	(−2.6858)	(−1.9439)	(−2.1903)	(−2.3516)	(−2.6928)
factor_enf	1.1803***	1.2473***	1.4657***	1.5624***	1.6101***	1.7045***	0.6395**	0.7115**	0.3067	0.3840	0.4533	0.5574*
	(3.7776)	(3.9865)	(4.7022)	(5.0122)	(5.4000)	(5.7300)	(2.0022)	(2.2262)	(0.9483)	(1.1871)	(1.4182)	(1.7481)
factor_legal	0.1095	0.0924	0.2329	0.2031	0.0427	0.0186	0.5100***	0.4928***	0.7075***	0.6885***	0.6979***	0.6699***
	(0.6423)	(0.5412)	(1.3688)	(1.1941)	(0.2626)	(0.1145)	(2.9257)	(2.8254)	(4.0089)	(3.9006)	(4.0009)	(3.8500)
size		−0.0145***		−0.0151***		−0.0173***		−0.0148***		−0.0147***		−0.0149***
		(−4.2123)		(−4.4269)		(−5.3182)		(−4.2300)		(−4.1438)		(−4.2711)
leverage		0.0267		0.0640***		0.0520**		0.0336		0.0440*		0.0891***
		(1.0901)		(2.6222)		(2.2336)		(1.3439)		(1.7382)		(3.5694)
profitability		−0.1061***		−0.1008***		−0.1183***		−0.1177***		−0.1249***		−0.1254***
		(−3.1691)		(−3.0213)		(−3.7148)		(−3.4402)		(−3.6085)		(−3.6738)
Constant	−0.1443	0.0850	−0.2672**	−0.0329	−0.1554	0.1141	−0.3403***	−0.1066	−0.5110***	−0.2806**	−0.5206***	−0.2961***
	(−1.1682)	(0.6266)	(−2.1677)	(−0.2435)	(−1.3181)	(0.8848)	(−2.6947)	(−0.7696)	(−3.9961)	(−2.0012)	(−4.1193)	(−2.1417)
Industry fixed effects	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Country fixed effects	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Observations	3303	3285	3303	3285	3303	3285	3303	3285	3303	3285	3303	3285
R-squared	0.05753	0.0828	0.0924	0.1021	0.0979	0.1107	0.0712	0.0794	0.0640	0.0729	0.0746	0.0876
F	10.68	10.49	13.34	13.22	14.22	14.47	10.05	10.04	8.964	9.142	10.57	11.16

This table reports Ordinary Least Squares coefficient estimates and t-statistics (in parentheses) based on the standard variance estimator. ***, **, and * indicate statistical significance at the 1%, 5% and 10% levels (two-tailed) respectively. The dependent variable is R²_incr(T_i), i.e. the incremental value-relevance that is attributed to the inclusion of a topic-specific CSR disclosure. factor_csr, factor_empl, factor_enf and factor_legal are four factors measuring the strength of the institutional environment with respect to CSR, employee, enforcement and legal issues. To generate these factors, we used a PCA with Oblimin rotations. The institutional variables are described in Appendix IV.

Table 6. Results for equation (3): controlling for textual characteristics We regress $R_{incr}^2\left(T_i\right)$ on the Institutional Environment and Textual Characteristics of the Annual Report, namely readability and tone, controlling for industry- and country-fixed effects.

	$R_{incr}^2\left(T_{ENV}\right)$	$R_{incr}^2\left(T_{SOC}\right)$	$R_{incr}^2\left(T_{EMPL}\right)$	$R_{incr}^2\left(T_{HR}\right)$	$R_{incr}^2\left(T_{CORR}\right)$	$R_{incr}^2\left(T_{BRIB}\right)$
factor_csr	−0.1615***	−0.2413***	−0.2271***	−0.2171***	−0.1700***	−0.2329***
	(−4.1775)	(−6.2634)	(−6.1548)	(−5.4956)	(−4.2525)	(−5.8916)
factor_empl	−0.2442***	−0.3148***	−0.3335***	−0.1671**	−0.1310*	−0.1566**
	(−3.6985)	(−4.7817)	(−5.2887)	(−2.4766)	(−1.9174)	(−2.3189)
factor_enf	1.1474***	1.4247***	1.6200***	0.6152*	0.2643	0.4129
	(3.6526)	(4.5495)	(5.4012)	(1.9162)	(0.8135)	(1.2852)
factor_legal	0.1375	0.2656	0.0525	0.5360***	0.7411***	0.7293***
	(0.8053)	(1.5600)	(0.3220)	(3.0712)	(4.1952)	(4.1748)
discl_tone	−2.0290***	−2.2841***	−1.4547***	−2.1027***	−2.3322***	−2.1535***
	(−3.8329)	(−4.3281)	(−2.8780)	(−3.8867)	(−4.2590)	(−3.9773)
discl_readability	0.0297	0.0309	0.0426**	0.0370*	0.0312	0.0281
	(1.5244)	(1.5921)	(2.2897)	(1.8569)	(1.5502)	(1.4087)
Constant	−0.1404	−0.2640**	−0.1418	−0.3331***	−0.5079***	−0.5182***
	(−1.1351)	(−2.1413)	(−1.2010)	(−2.6355)	(−3.9702)	(−4.0960)
Industry fixed effects	YES	YES	YES	YES	YES	YES
Country fixed effects	YES	YES	YES	YES	YES	YES
Observations	3303	3303	3303	3303	3303	3303
R-squared	0.0808	0.0990	0.1023	0.0773	0.0706	0.0803
F	10.66	13.33	13.82	10.16	9.218	10.59

This table reports Ordinary Least Squares coefficient estimates and t-statistics (in parentheses) based on standard variance estimator. ***, **, and * indicate statistical significance at the 1%, 5% and 10% levels (two-tailed) respectively. The dependent variable is R²_incr(T_i), i.e. the incremental value relevance that is attributed to the inclusion of the topic-specific CSR disclosure. factor_csr, factor_empl, factor_enf and factor_legal are four factors measuring the strength of the institutional environment with respect to CSR, employee, enforcement and legal issues We generated these factors through a PCA with Oblimin rotations. See Appendix IV for a description of the institutional variables. All other variables are shown in Appendix I and winsorized at the top and bottom 1%.

Table 7. Results of equation (3): the incremental value-relevance of the institutional environment among good and bad CSR performers.

	$R_{incr}^2\left(T_{ENV}\right)$		$R_{incr}^2\left(T_{SOC}\right)$		$R_{incr}^2\left(T_{EMPL}\right)$		$R_{incr}^2\left(T_{HR}\right)$		$R_{incr}^2\left(T_{CORR}\right)$		$R_{incr}^2\left(T_{BRIB}\right)$
factor_inst	−0.0368	−0.0542	−0.0837**	−0.1065***	−0.0723**	−0.0861**	−0.0744*	−0.0855**	−0.0895**	−0.1038***	−0.1235***	−0.1395***
	(−0.9763)	(−1.4152)	(−2.2305)	(−2.7868)	(−2.0088)	(−2.3550)	(−1.9331)	(−2.1844)	(−2.2941)	(−2.6161)	(−3.2067)	(−3.5595)
djsi = 1	0.0549***		0.0634***		0.0496***		0.0533***		0.0490***		0.0540***
	(4.5292)		(5.2425)		(4.2808)		(4.2966)		(3.8956)		(4.3477)
djsi*factor_inst	−0.0611***		−0.0863***		−0.0703***		−0.0749***		−0.0498**		−0.0493**
	(−3.0805)		(−4.3715)		(−3.7108)		(−3.6954)		(−2.4223)		(−2.4293)
a4ir_good		−0.0060		−0.0109		−0.0089		−0.0004		−0.0058		−0.0069
		(−0.5741)		(−1.0399)		(−0.8897)		(−0.0369)		(−0.5352)		(−0.6410)
a4ir_good*factor_inst		−0.0067		−0.0043		−0.0145		−0.0226		−0.0092		−0.0083
		(−0.4020)		(−0.2580)		(−0.9114)		(−1.3251)		(−0.5309)		(−0.4887)
Constant	0.0431*	0.0617	0.0314	0.0566**	0.0191	0.0385*	0.0574**	0.0733***	0.0342	0.0502**	0.0110	0.0283
	(1.8069)	(2.5368)	(1.3218)	(2.3337)	(0.8390)	(1.6570)	(2.3545)	(2.9477)	(1.3850)	(1.9929)	(0.4492)	(1.1385)
Industry fixed effects	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Country fixed effects	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Observations	3297	3303	3297	3303	3297	3303	3297	3303	3297	3303	3297	3303
R-squared	0.0823	0.0755	0.1022	0.0929	0.1044	0.0987	0.0780	0.0719	0.0687	0.0643	0.0804	0.0749
F	10.85	9.912	13.78	12.42	14.11	13.28	10.24	9.391	8.935	8.334	10.59	9.822

We regress $R_{incr}^2\left(T_i\right)$ on the factor_inst and an interaction between factor_inst and good CSR performers measured either by an inclusion in the DJSI or an above median a4ir performance, controlling for industry and country-fixed effects.

This table reports Ordinary Least Squares coefficient estimates and t-statistics (in parentheses) based on the standard variance estimator. ***, **, and * indicate statistical significance at the 1%, 5% and 10% levels (two-tailed) respectively. The dependent variable is R²_incr(T_i), i.e. the incremental value relevance that is attributed to the inclusion of the topic-specific CSR disclosure. The dependent variable is R²_incr(T_i), i.e. the incremental value relevance that is attributed to the inclusion of a topic-specific CSR disclosure. factor_inst is our composite measure of the strength of the institutional environment that is equal to the mean of the four factors generated through a PCA. A4IR_good takes the value 1 if the company’s score is above the median and 0 otherwise. See Appendix IV for a description of the institutional variables.

Table

Variable	Description	Source of data
Main Variables
P	Share Price at fiscal year end	Bloomberg
BVE	Book value of equity is common equity at fiscal year end, divided by the number of outstanding shares	Datastream
EARN	Earnings per share at fiscal year end	Datastream
TENV	The tf-idf similarity between the respective disclosure in a firm’s annual report and the compiled vocabulary that reflects disclosure on environmental matters; the search term is ‘ecology.’	textual analysis
TSOC	The tf-idf similarity between the respective disclosure in a firm’s annual report and the compiled vocabulary that reflects disclosure on social matters; the search term is ‘social.’	textual analysis
TEMPL	The tf-idf similarity between the respective disclosure in a firm’s annual report and the compiled vocabulary that reflects disclosure on employee matters; the search term is ‘employee.’	textual analysis
THR	The tf-idf similarity between the respective disclosure in a firm’s annual report and the compiled vocabulary that reflects disclosure on human rights; the search term is the bigram ‘human right.’	textual analysis
TCORR	The tf-idf similarity between the respective disclosure in a firm’s annual report and the compiled vocabulary that reflects disclosure on corruption; the search term is ‘corruption.’	textual analysis
TBRIB	The tf-idf similarity between the respective disclosure in a firm’s annual report and the compiled vocabulary that reflects disclosure on bribery; the search term is ‘bribery.’	textual analysis
Institutional Variables
factor_csr	Institutional factor derived from a PCA on the institutional variables (Appendix III, Panel A). Panel B of Appendix III provides an overview of which variable feeds into which factor.	Appendix III, Panel A
factor_empl	Institutional factor derived from a PCA on the institutional variables (Appendix III, Panel A). Panel B of Appendix III provides an overview of which variable feeds into which factor.	Appendix III, Panel A
factor_enf	Institutional factor derived from a PCA on the institutional variables (Appendix III, Panel A). Panel B of Appendix III provides an overview of which variable feeds into which factor.	Appendix III, Panel A
factor_legal	Institutional factor derived from a PCA on the institutional variables (Appendix III, Panel A). Panel B of Appendix III provides an overview of which variable feeds into which factor.	Appendix III, Panel A
Control Variables
discl_tone	The tone of the annual report, calculated as [(number of positive words – number of negative words)/number of total words] of the annual report.The word lists are based on those of Loughran and McDonald (2011); higher values reflect a more positive tone.	textual analysis
discl_readability	The readability of the annual report, measured as the average mean of the percentile ranks for the Fog Index, and the Flesch-Kincaid and the (101 – percentile ranks) for the Flesch-Reading-Ease, divided by 100, multiplied with (−1); the Fog Index, the Flesch-Kincaid and the Flesch-Reading-Ease are calculated for the annual reports; higher values reflect higher readability.	textual analysis
leverage	A firm’s financial leverage, measured as total debt divided by total assets at fiscal year end.	datastream
size	The firm size, measured as the logarithm of the revenues at fiscal year end.	datastream
profitability	Profitability is measured as EBIT divided by total assets at fiscal year end.	datastream
DJSI	Dummy variable that equals 1 if the company was listed in the Dow Jones Sustainability Index between 2010 and 2016.	DJSI
a4ir	ASSET4 score; details provided by Thomson Reuters (2015).	datastream

Table

Panel A: Top-20 words
Ti	Top-20 words	Total number of words	Number of unique words
TENV	‘ecology’ (989), ‘development’ (418), ‘sustainable’ (406), ‘energy’ (363), ‘ministry’ (299), ‘management’ (176), ‘environmental’ (164), ‘director’ (123), ‘sustainability’ (122), ‘biodiversity’ (120), ‘project’ (113), ‘company’ (105), ‘report’ (101), ‘use’ (99), ‘economy’ (98), ‘environment’ (91), ‘climate’ (86), ‘material’ (80), ‘transport’ (76), ‘building’ (75)	18,280	3343
TSOC	‘social’ (162,132), ‘responsibility’ (38,693), ‘environmental’ (38,634), ‘corporate’ (31,345), ‘report’ (24,775), ‘security’ (24,360), ‘employee’ (23,410), ‘company’ (20,735), ‘information’ (20,324), ‘cost’ (16,074), ‘economic’ (15,513), ‘business’ (15,256), ‘financial’ (14,435), ‘management’ (14,410), ‘development’ (13,309), ‘contribution’ (12,035), ‘benefit’ (11,749), ‘policy’ (11,741), ‘risk’ (11,651), ‘tax’ (10,969)	2,565,511	42,115
TEMP	‘employee’ (865,083), ‘share’ (224,953), ‘benefit’ (136,486), ‘plan’ (133,910), ‘company’ (128,974), ‘number’ (88,554), ‘director’ (59,063), ‘management’ (57,249), ‘option’ (57,222), ‘board’ (52,515), ‘financial’ (52,502), ‘business’ (52,492), ‘pension’ (51,067), ‘service’ (48,306), ‘cost’ (47,934), ‘report’ (47,747), ‘scheme’ (46,254), ‘executive’ (45,483), ‘performance’ (45,046), ‘based’ (43,771)	11,125,751	76,627
THR	‘right’ (25,784), ‘human’ (24,758), ‘principle’ (4891), ‘business’ (4817), ‘policy’ (3430), ‘respect’ (3302), ‘supplier’ (3064), ‘employee’ (2843), ‘risk’ (2439), ‘global’ (2420), ‘labor’ (2398), ‘company’ (2296), ‘environment’ (2072), ‘standard’ (2025), ‘corruption’ (2016), ‘compact’ (1964), ‘declaration’ (1880), ‘code’ (1877), ‘social’ (1875), ‘conduct’ (1778)	338,393	8621
TCORR	‘corruption’ (18,136), ‘anti’ (8467), ‘risk’ (4339), ‘bribery’ (4275), ‘policy’ (3818), ‘business’ (3768), ‘compliance’ (3466), ‘employee’ (2708), ‘right’ (2481), ‘code’ (2252), ‘human’ (2209), ‘conduct’ (2069), ‘training’ (2039), ‘company’ (2024), ‘law’ (1937), ‘ethic’ (1794), ‘principle’ (1726), ‘practice’ (1506), ‘procedure’ (1490), ‘management’ (1430)	278,992	8134
TBRIB	‘bribery’ (7334), ‘corruption’ (4296), ‘anti’ (3868), ‘policy’ (2259), ‘act’ (1736), ‘business’ (1735), ‘risk’ (1,695), ‘compliance’ (1549), ‘employee’ (1317), ‘code’ (1053), ‘conduct’ (996), ‘law’ (995), ‘training’ (986), ‘including’ (949), ‘company’ (807), ‘procedure’ (788), ‘report’ (689), ‘fraud’ (672), ‘control’ (643), ‘management’ (632)	118,180	4996

Table

Panel B: Word window examples
	20-word window	Corresponding CSR disclosure
TENV	TENV: ‘system,’ ‘energy,’ ‘optimization,’ ‘thermal,’ ‘insulation,’ ‘infrastructure,’ ‘relevant,’ ‘value,’ ‘relation,’ ‘operational,’ ‘ecology,’ ‘collected,’ ‘consolidated,’ ‘consolidated,’ ‘value,’ ‘assessed,’ ‘used,’ ‘benchmark,’ ‘verify,’ ‘realized‘	Roche, Annual Report 2012, p. 112:‘[…] measures to reduce energy consumption are implemented, whether through production installations equipped with intelligent energy management systems or through the energy optimization and thermal insulation of infrastructures. All the relevant values in relation to operational ecology are collected and consolidated annually. These consolidated values are assessed and used as a benchmark to verify realized objectives.’
TSOC	TSOC: ‘corporate,’ ‘social,’ ‘responsibility,’ ‘includes,’ ‘support,’ ‘social,’ ‘institution,’ ‘local,’ ‘basis,’ ‘demonstrate,’ ‘social,’ ‘responsibility,’ ‘making,’ ‘donation,’ ‘various,’ ‘organization,’ ‘provide,’ ‘direct,’ ‘efficient,’ ‘unbureaucratic‘	Wienerberger, Annual Report 2011, p. 51: ‘This central principle of active corporate social responsibility includes support for social institutions on a local basis. We do not demonstrate our social responsibility by making donations to various international organizations, but provide direct, efficient and unbureaucratic help.’
TEMP	TEMPL: ‘called,’ ‘coaching,’ ‘performance,’ ‘initiated,’ ‘manager,’ ‘goal,’ ‘strengthening,’ ‘feedback,’ ‘culture,’ ‘improving,’ ‘employee,’ ‘performance,’ ‘crucial,’ ‘make,’ ‘high,’ ‘performing,’ ‘learning,’ ‘organization,’ ‘recognized,’ ‘best‘	Electrolux, Annual Report 2013, p. 45:‘[…] a global training program called (Coaching for Performance) was initiated in 2013 for all managers with the goal of strengthening the feedback culture and improving employee performance. This is crucial to make Electrolux a high-performing learning organization. To be recognized as the best appliance company by our employees is an important vision for Electrolux.’
THR	THR: ‘business,’ ‘human,’ ‘right,’ ‘ungp,’ ‘base,’ ‘human,’ ‘right,’ ‘commitment,’ ‘policy,’ ‘bill,’ ‘human,’ ‘right,’ ‘universal,’ ‘declaration,’ ‘human,’ ‘right,’ ‘covenant,’ ‘civil,’ ‘political,’ ‘right‘	Unilever, Annual Report 2013, p. 17:‘In line with the UN Guiding Principles on Business and Human Rights (UNGP), we base our human rights commitment and policy on the International Bill of Human Rights (the Universal Declaration of Human Rights, the International Covenant on Civil and Political Rights and the International Covenant on Economic, Social and Cultural Rights).’
TCORR	TCORR: ‘act,’ ‘lawful,’ ‘manner,’ ‘particular,’ ‘emphasis,’ ‘placed,’ ‘compliance,’ ‘antitrust,’ ‘legislation,’ ‘avoidance,’ ‘corruption,’ ‘risk,’ ‘compliance,’ ‘measure,’ ‘supplemented,’ ‘range,’ ‘internal,’ ‘policy,’ ‘guideline,’ ‘instruction‘	BMW Group, Annual Report 2014, p. 185:A coordinated set of instruments and measures is employed to ensure that the BMW Group, its representative bodies, its managers and staff act in a lawful manner. Particular emphasis is placed on compliance with antitrust legislation and the avoidance of corruption risks. Compliance measures are supplemented by a whole range of internal policies, guidelines and instructions.’
TBRIB	TBRIB: ‘regulation, ‘ ‘relating, ‘ ‘import, ‘ ‘export, ‘ ‘control, ‘ ‘money, ‘ ‘laundering, ‘ ‘false, ‘ ‘accounting, ‘ ‘anti, ‘ ‘bribery, ‘ ‘anti, ‘ ‘boycott, ‘ ‘provision, ‘ ‘non, ‘ ‘compliance, ‘ ‘expose, ‘ ‘fine, ‘ ‘penalty,’ ‘suspension‘	BAE Systems, Annual Report 2013, p. 108:These include, without limitation, regulations relating to import export controls, money laundering, false accounting, anti-bribery and anti-boycott provisions. Non-compliance could expose the Group to fines, penalties, suspension or debarment, which could have a material adverse effect on the Group.

Panel A of the table presents the top-20 word windows for each topic-specific vocabulary. The frequency of the respective word appears in the vocabulary and is indicated in parentheses. Panel B shows examples retrieved from our sample annual reports for every T.

Table

Panel A: Description of institutional variables
Variable	Variable measurement	Source
Empl_Law	Employment laws based on alternative employment contracts, cost of increasing hours worked, cost of firing workers and cost of dismissal procedures.	Botero et al. (2004)
SoSec_Law	Social security laws index based on benefits for old age, disability, death, sickness, health and unemployment.	Botero et al. (2004)
Coll_Law	Collective relations laws based on labor-union power and collective disputes.	Botero et al. (2004)
HRI	Human Rights Index reflecting human-rights protection. Higher scores indicate better protection.
Rule_Law	Extent to which agents have confidence in, and abide by, the rules of society.	World Bank
Voice_Acct	Extent to which a country’s citizens are able to select their government and voice other concerns.	World Bank
Govt_Eff	Perceptions of the quality of public services and policies, and the government to promote such policies.	World Bank
Reg_Qual	Perceptions of the government’s ability to formulate and implement sound policies.	World Bank
Corr_Contr	Perceptions of the extent to which public power is exercised for private gain.	World Bank
Sust_Dev	Sustainable development priority: ‘Sustainable development is a priority in countries.’	IMD World Competitiveness Database (WCD)
Eth_Pract	Ethical practice implementation: ‘Ethical practices are implemented in companies.’	IMD WCD
Soc_Resp	Social responsibility: ‘The social responsibility of business leaders is high.’	IMD WCD
Brib_Cor	Bribery and corruption: ‘Bribery and corruption do not exist.’	IMD WCD
Env_Laws	Environmental laws: ‘Environmental laws and compliance do not hinder the competitiveness of businesses.’	IMD WCD
EPI	Environmental Performance Index.^30	Yale Law School
Jour_Free	Degree of journalistic freedom.	Reporters without borders
Law_Order	Strength and impartiality of the legal system; strength of popular observance of the law.	ICRG
Gov_Stab	Measure of the government’s ability to carry out its declared programs and to stay in office.	ICRG
ASDI	Anti Self-Dealing Index, measured as ex-ante and ex-post private control of self-dealing.	Djankov et al. (2008)
PEI	Public Enforcement Index, measured as the strength of public enforcement of self-dealing.	Djankov et al. (2008)

Table

Panel B: Description of the composition of the institutional factors
CSR Awareness (factor_csr)	Employee Protection (factor_empl)	Enforcement (factor_enf)	Legal Environment (factor_legal)
Sust_Dev	Empl_Law	Gov_Stab	SoSec_Law
Eth_Pract	Coll_Law	PEI	HRI
Soc_Resp	ASDI		Rule_Law
Env_Laws			Voice_Acct
			Gov_Eff
			Reg_Qual
			Brib_Cor
			EPI
			Jour_Free
			Corr_Contr
			Law & Order

Loughran, T., & McDonald, B. (2011). When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. Journal of Finance, 66(1), 35–65. https://doi.org/10.1111/j.1540-6261.2010.01625.x

 Web of Science ®Google Scholar

Botero, J., Djankov, S., La Porta, R., Lopez-De-Silanes, F., & Shleifer, A. (2004). The regulation of labor. Quarterly Journal of Economics, 119(4), 1339–1382. https://doi.org/10.1162/0033553042476215

 Web of Science ®Google Scholar

Djankov, S., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2008). The law and economics of self-dealing. Journal of Financial Economics, 88(3), 430–465. https://doi.org/10.1016/j.jfineco.2007.02.007

 Web of Science ®Google Scholar