Annual Report Narratives and the Cost of Equity Capital: U.K. Evidence of a U-shaped Relation

Figures & data

Table 1. Sample selection and distribution of observations across years

	Observations Lost	Observations Remaining
Panel A. Sample Selection
Disclosure scores from full CFIE dataset		10,443
./. Missing Datastream company codes	−9
./. Multiple annual report documents per firm-year	−43	10,391
./. Missing cost of equity capital estimates	−4405	5986
./. Missing control variables	−834	5152
Panel B. Distribution of Observations Across Years	Observations	in %
2003	322	6.3
2004	388	7.5
2005	460	8.9
2006	399	7.7
2007	452	8.8
2008	441	8.6
2009	455	8.8
2010	462	9.0
2011	540	10.5
2012	455	8.8
2013	414	8.0
2014	364	7.1
Total	5152	100

Notes: Panel A reports our sample selection criteria and the number of observations lost. Panel B shows the distribution of remaining observations across the sample period years.

Table 2. Descriptive statistics

Variable	OBS	Mean	STD	Minimum	First Quartile	Median	Third Quartile	Maximum
Panel A. Ranked Disclosure Scores
Discindex	5152	0.59	0.28	0.01	0.37	0.63	0.84	1.00
Causal	5152	0.58	0.28	0.01	0.35	0.61	0.82	1.00
Fog	5152	0.53	0.28	0.00	0.28	0.54	0.77	0.99
Forward	5152	0.59	0.28	0.00	0.36	0.63	0.83	1.00
Gov	5152	0.60	0.28	0.09	0.39	0.65	0.84	1.00
Perf	5152	0.57	0.29	0.01	0.33	0.60	0.82	1.00
Rearend	5152	0.58	0.28	0.00	0.34	0.61	0.82	1.00
Restfront	5152	0.56	0.29	0.00	0.32	0.59	0.82	1.00
Strategy	5152	0.57	0.29	0.00	0.34	0.61	0.83	1.00
Discindex_squared	5152	0.43	0.31	0.00	0.14	0.40	0.71	1.00
Causal_squared	5152	0.41	0.31	0.00	0.12	0.38	0.68	1.00
Fog_squared	5152	0.36	0.30	0.00	0.08	0.29	0.60	0.99
Forward_squared	5152	0.42	0.31	0.00	0.13	0.39	0.69	1.00
Gov_squared	5152	0.44	0.31	0.01	0.15	0.42	0.70	1.00
Perf_squared	5152	0.40	0.31	0.00	0.11	0.36	0.67	1.00
Rearend_squared	5152	0.41	0.31	0.00	0.12	0.37	0.67	1.00
Restfront_squared	5152	0.40	0.31	0.00	0.10	0.35	0.68	1.00
Strategy_squared	5152	0.41	0.31	0.00	0.11	0.37	0.69	1.00
Panel B. Cost of Equity Capital Estimates
Coec	5152	0.17	0.14	0.02	0.08	0.12	0.21	0.69
Coec_gls	5152	0.13	0.10	0.01	0.07	0.11	0.16	0.69
Coec_ct	5145	0.16	0.15	0.01	0.06	0.10	0.19	0.76
Coec_oj	3310	0.20	0.15	0.03	0.10	0.16	0.26	0.78
Coec_peg	3435	0.23	0.18	0.02	0.10	0.17	0.30	0.86
Panel C. Control Variables
Beta	5152	0.61	0.49	−0.37	0.25	0.54	0.89	2.11
Leverage	5152	0.25	0.23	0.00	0.02	0.22	0.40	0.88
MTBV	5152	2.83	3.60	0.18	1.09	1.79	3.12	26.2
LogTA	5152	11.8	2.20	6.93	10.3	11.7	13.3	17.1
AbsAwca	5152	0.17	0.19	0.00	0.05	0.12	0.22	1.15
LogAge	5152	3.14	0.93	1.39	2.40	3.04	3.93	4.81
Panel D. Annual Median Scores for Discindex and its Components
	Discindex	Causal	Fog	Forward	Gov	Perf	Rearend	Restfront	Strategy
2003	0.45	0.57	0.66	0.51	0.59	0.51	0.23	0.43	0.47
2004	0.45	0.58	0.66	0.52	0.59	0.52	0.25	0.42	0.47
2005	0.53	0.59	0.62	0.57	0.62	0.54	0.30	0.43	0.50
2006	0.64	0.65	0.65	0.61	0.67	0.61	0.58	0.50	0.64
2007	0.66	0.64	0.56	0.62	0.68	0.60	0.61	0.59	0.64
2008	0.68	0.65	0.53	0.63	0.65	0.61	0.68	0.62	0.63
2009	0.68	0.64	0.50	0.65	0.64	0.62	0.71	0.62	0.59
2010	0.69	0.64	0.49	0.68	0.67	0.64	0.72	0.65	0.63
2011	0.68	0.59	0.46	0.64	0.61	0.63	0.69	0.64	0.60
2012	0.74	0.64	0.46	0.72	0.70	0.65	0.71	0.71	0.68
2013	0.78	0.64	0.42	0.78	0.74	0.67	0.74	0.76	0.75
2014	0.74	0.53	0.43	0.76	0.83	0.61	0.75	0.74	0.78

Notes: This table reports descriptive statistics for the estimation sample. Descriptive statistics for ranked disclosure scores, implied cost of equity capital estimates, and control variables, are reported, respectively, in Panel A, B and C. We report descriptive statistics for ranked disclosure scores by year in Panel D. All variables are as defined in Appendix 2 and Equation (A1) in Appendix 1. All variables other than ranked disclosure scores have been winsorized at the 1st and 99th percentile. OBS = Observations. STD = Standard Deviation.

Table 3. Panel regressions for cost of equity capital on overall disclosure

Dependent Variable: Cost of Equity Capital (Coec)	(1)	(2)	(3)	(4)
Discindex	−0.085***	−0.415***	−0.387***	−0.169***
	(−6.80)	(−10.68)	(−9.73)	(−4.39)
Discindex_squared		0.323***	0.311***	0.154***
		(10.40)	(9.57)	(5.14)
Beta	−0.008*	−0.013***	−0.004	0.002
	(−1.78)	(−3.19)	(−1.00)	(0.37)
Leverage	0.101***	0.099***	0.091***	0.075***
	(8.78)	(8.88)	(7.53)	(5.06)
MTBV	−0.007***	−0.007***	−0.008***	−0.005***
	(−11.11)	(−11.46)	(−11.47)	(−7.35)
LogTA	−0.033***	−0.036***	−0.040***	−0.049***
	(−16.99)	(−18.68)	(−18.95)	(−9.86)
AbsAwca	0.012	0.007	0.008	0.005
	(0.94)	(0.56)	(0.66)	(0.54)
LogAge	0.002	0.002	0.001	0.013
	(0.85)	(0.89)	(0.42)	(1.28)
Constant	0.599***	0.694***	0.706***	0.733***
	(31.38)	(33.06)	(24.48)	(13.53)
Observations	5152	5152	5152	5152
Adj R-squared	0.43	0.46	0.50	0.43
Fixed Effect	No	No	Industry	Firm
Optimum = –β1 / 2β2		0.64	0.62	0.55

Notes: The table shows the relation between a firm’s (ranked) overall disclosure index and the cost of equity capital, with a linear term (only) for disclosure in column (1) and an additional quadratic term for disclosure in columns (2) to (4). Next to our standard control variables (see Appendix 2 for definitions) we include no fixed effects (columns (1) and (2)), industry fixed effects (column 3), and firm fixed effects (column 4). The t-statistics are adjusted for firm error clustering; *, **, and *** denote significance at 10%, 5%, and 1%. We calculate the optimal Discindex as –β₁ / 2β₂ where β₁ is the coefficient on Discindex and β₂ is the coefficient on Discindex_squared (as in Equation (1)).

Figure 1. U-shaped relation between cost of equity capital and volume of annual report narratives. This figure illustrates the reduction in the cost of equity capital as a function of the (ranked) volume of annual report narratives, as implied by the regression results in Table 3, column (4) (that is after allowing for firm fixed effects)

Table 4. Three-year rolling window regressions showing variation over time in optimum levels of disclosure

Dependent Variable: Cost of Equity Capital (Coec)	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Beginning Year	2003	2004	2005	2006	2007	2008	2009	2010	2011	2012
Ending Year	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014
Discindex	−0.435***	−0.388***	−0.438***	−0.417***	−0.419***	−0.415***	−0.408***	−0.383***	−0.313***	−0.345***
	(−5.87)	(−5.92)	(−7.66)	(−6.44)	(−6.45)	(−6.26)	(−6.56)	(−5.71)	(−4.93)	(−5.68)
Discindex_squared	0.434***	0.374***	0.384***	0.361***	0.347***	0.310***	0.281***	0.258***	0.203***	0.243***
	(6.03)	(6.11)	(7.27)	(6.12)	(6.10)	(5.50)	(5.33)	(4.54)	(3.77)	(4.84)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1170	1247	1311	1292	1348	1358	1457	1457	1409	1233
Adj R-squared	0.53	0.52	0.52	0.48	0.55	0.56	0.60	0.56	0.53	0.52
Fixed Effect	Industry	Industry	Industry	Industry	Industry	Industry	Industry	Industry	Industry	Industry
Optimum	0.50	0.52	0.57	0.58	0.60	0.67	0.73	0.74	0.77	0.71
% of Firms Below Optimum	52%	47%	45%	41%	43%	48%	56%	55%	56%	45%

Notes: The table shows the relation between a firm’s overall disclosure index and the cost of equity capital with a linear term and a quadratic term for disclosure. We calculate the optima based on the coefficients on Discindex and Discindex_squared as described in the text and in Table 3. The regressions are based on a three-year rolling window and include industry fixed effects. We include our standard control variables (untabulated) (see Appendix 2 for definitions). The t-statistics are adjusted for firm error clustering; *, **, and *** denote significance at 10%, 5%, and 1%.

Table 5. Firm learning effects and disclosure adjustments

Dependent Variable: ΔDiscindex	(1)	(2)	(3)
Lag_distance_overdisclosure	−0.383***	−0.386***	−0.323***
	(−14.2)	(−11.9)	(−11.3)
Lag_distance_underdisclosure	0.101***	0.112***	0.066***
	(5.81)	(5.29)	(3.53)
Timetrend	−0.006***	−0.058***	−0.055***
	(−6.07)	(−5.31)	(−5.65)
LogTA	0.021***	0.023***	0.017***
	(13.43)	(11.4)	(9.67)
MTBV	−0.001	−0.001	−0.001*
	(−0.78)	(−0.84)	(−1.75)
AbsAwca	0.030**	0.039***	0.036**
	(2.52)	(2.69)	(2.56)
ROA	−0.012	0.018	0.024
	(−0.29)	(0.32)	(0.44)
Exantefinance	0.020	0.035	0.032
	(1.29)	(1.55)	(1.43)
SNetincome	0.015	0.000	0.002
	(0.39)	(0.01)	(0.03)
Constant	−0.202***	0.340***	0.377***
	(−9.95)	(2.96)	(3.70)
Lag_ΔDiscindex			−0.255***
			(−11.6)
Observations	3176	2095	2095
Adj R-squared	0.15	0.17	0.22
Year Fixed Effect	Yes	Yes	Yes

Notes: The table reports the year-on-year change in the disclosure index, ΔDiscindex, depending on whether the firm, in the prior year, was over- or under-disclosing, relative to the sample-wide year-specific optimum level of disclosure in that year. Lag_distance_overdisclosure is set equal to the prior year actual disclosure level minus prior year estimated optimal disclosure level if this value is positive, and zero otherwise. Lag_distance_underdisclosure is set equal to the prior year estimated optimal disclosure level minus the prior year actual disclosure level if this value is positive, and zero otherwise. The t-statistics are adjusted for firm error clustering; *, **, and *** denote significance at 10%, 5%, and 1%.

Table 6. Regulatory change and clutter

	(1)	(2)
	Distance_Overdisclosure	Increase_Discindex
Post_BR	0.038***	0.018***
	(8.19)	(2.60)
Post_CGREV	−0.035***	−0.001
	(−8.97)	(−0.17)
Timetrend	−0.002**	−0.006***
	(−2.47)	(−4.34)
LogTA	0.026***	0.009
	(6.42)	(1.51)
MTBV	−0.000	−0.001
	(−0.10)	(−1.03)
AbsAwca	0.011	0.058***
	(1.61)	(4.03)
ROA	−0.016	−0.015
	(−0.74)	(−0.30)
Exantefinance	−0.018***	−0.004
	(−7.24)	(−0.77)
SNetincome	−0.043**	0.074
	(−2.49)	(1.56)
Observations	5065	3176
Adj R-squared	0.11	0.04
Fixed Effects	Firm	Firm

Notes: The table reports the effects of the introduction of the Business Review in 2006 (Post_BR) and the revision of the Corporate Governance Code in 2010 (Post_CGREV) on the amount of firm-year-specific over-disclosure (column (1)) and the amount of year-on-year increases in Discindex (column (2)). In periods of year-on-year decreases in Discindex, Increase_Discindex is set equal to 0. Post_BR is equal to 1 for all firms-years with a financial year-end equal to 31st of March 2006 or later. Post_CGREV is equal to 1 for all firm-years with a financial year-end equal to 30th of June 2011 or later. The t-statistics in columns (1) and (2) are adjusted for firm error clustering; *, **, and *** denote significance at 10%, 5%, and 1%.

Table 7. Individual component analysis

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Scaled_Perf	−0.060***								−0.081***
	(−4.59)								(−3.58)
Scaled_Gov		−0.045***							−0.038**
		(−3.57)							(−2.43)
Scaled_Strategy			−0.543***						−0.299
			(−3.47)						(−1.54)
Scaled_Causal				−1.638***					1.602
				(−2.68)					(1.52)
Scaled_Forward					−0.521*				0.461
					(−1.85)				(1.06)
Fog						0.003			0.000
						(0.34)			(0.05)
Log_Rearend							0.000		−0.428***
							(0.03)		(−4.92)
Log_Rearend_squared									0.022***
									(4.86)
Log_Restfront								−0.001	−0.004
								(−0.67)	(−0.76)
Log_Restfront_squared									0.000
									(0.75)
Constant	0.671***	0.645***	0.658***	0.666***	0.661***	0.661***	0.662***	0.667***	2.760***
	(23.73)	(22.07)	(22.86)	(23.39)	(23.03)	(23.11)	(14.86)	(23.44)	(6.47)
Observations	5152	5152	5152	5152	5152	5152	5152	5152	5152
Adj R-squared	0.48	0.47	0.47	0.47	0.47	0.47	0.47	0.47	0.49
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Fixed Effects	Industry	Industry	Industry	Industry	Industry	Industry	Industry	Industry	Industry

Notes: The table reports the effect of individual components of Discindex on the cost of equity capital after scaling these individual word count measures by Rearend. We include Beta, Leverage, MTBV, LogTA, AbsAwca, LogAge as control variables and control for industry-fixed effects. t-statistics are adjusted for firm error clustering; *, **, and *** denote significance at 10%, 5%, and 1%.

Table

Variable	Definition
Panel A. Disclosure Variables (before Ranking)
Discindex	Overall disclosure index
Perf	Word count in management performance commentary sections
Gov	Word count in governance & remuneration sections
Restfront	Word count in rest of front-end sections (i.e., excluding performance commentary and governance & remuneration sections)
Rearend	Word count in rear-end sections (including financial statements and notes)
Strategy	Strategic keyword count across all sections of the annual report (with each keyword being weighted by the probability that the keyword features in strategic commentary sections)
Causal	Number of causal keywords in the management performance commentary sections
Forward	Number of forward-looking keywords in all annual report sections
Fog	Fog readability index
Panel B. Cost of Equity Capital Variables
Coec	Average cost of equity capital (across four valuation models)
Coec_gls	Cost of equity capital implied by the Gebhardt et al. (2001) valuation model
Coec_ct	Cost of equity capital implied by the Claus and Thomas (2001) valuation model
Coec_oj	Cost of equity capital implied by the Ohlson and Juettner-Nauroth (2005) valuation model
Coec_peg	Cost of equity capital implied by the Easton (2004) valuation model
Panel C. Control Variables
Beta	Estimated via the market model and from weekly returns over the prior two years
Leverage	Debt-to-equity ratio defined as total debt (Datastream item WC03255) over total debt plus total equity (Datastream item WC03995)
MTBV	Market-to-book value (Datastream item MTBV)
LogTA	Natural logarithm of total assets (Datastream item WC02999)
AbsAwca	Absolute abnormal working capital accruals (calculated via the modified Jones model)
LogAge	Natural logarithm of number of years since company incorporation (with incorporation date given by Datastream item WC18273)
ROA	Return on Assets (net income (Datastream item WC0175) divided by total assets (Datastream item WC02999))
Exantefinance	Indicator variable equal to 1 if either operating cash flows minus average capital expenditure from years t−3 through to year t−1 (scaled by current assets in t−1) is less than 0.5 (Dechow, Sloan, & Sweeney, 1996)
SNetincome	Standard deviation of net income over year t, t−1 and t−2
Panel D. Further Variables
Lag_distance_overdisclosure	Difference between the one-year lagged actual firm-year-specific disclosure score and the one-year lagged year-specific optimum disclosure score if positive, and 0 otherwise
Lag_distance_underdisclosure	Difference between the one-year lagged year-specific optimum disclosure score and the one-year lagged actual firm-year-specific disclosure score if positive, and zero otherwise.
Post_BR	Indicator variable equal to 1 for all financial year-ends after 31/03/2006
Post_CGREV	Indicator variable equal to 1 for all financial year-ends after 30/06/2011
Timetrend	Year minus 2002

Gebhardt, W. R., Lee, C. M. C., & Swaminathan, B. (2001). Toward an implied cost of capital. Journal of Accounting Research, 39(1), 135–176.

 Web of Science ®Google Scholar

Claus, J., & Thomas, J. (2001). Equity premia as low as three percent? Evidence from analysts’ earnings forecasts for domestic and international stock markets. The Journal of Finance, 56(5), 1629–1666.

 Web of Science ®Google Scholar

Ohlson, J. A., & Juettner-Nauroth, B. E. (2005). Expected EPS and EPS growth as determinants of value. Review of Accounting Studies, 10(2–3), 349–365.

 Web of Science ®Google Scholar

Easton, P. D. (2004). PE ratios, PEG ratios, and estimating the implied expected rate of return on equity capital. The Accounting Review, 79(1), 73–95.

 Web of Science ®Google Scholar

Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1996). Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research, 13(1), 1–36.

 Google Scholar