Skip to main content
SpringerLink
Log in

A new measure of firm-group accounting closeness

  • Original Research
  • Published:
Review of Quantitative Finance and Accounting Aims and scope Submit manuscript

Abstract

In this study we develop a new measure of firm-group accounting similarity which captures co-movement in accounting fundamentals over time. We calculate the mean of (1) the average pair-wise correlation between earnings and cash flows for a group of firms and (2) the average pair-wise \(R^2\) from regressing firm i’s earnings (cash flows) on firm j’s earnings (cash flows) within-group. We analyze how the four most widely used industrial classification schemes perform in relation to our measure. We document that the within-industry information transfer contagion effect is increasing in our measure. We confirm that within-industry similarity for the average industry varies widely and is the reason why announcement contagion effects are not observed in certain industries. Lastly, we show that the number of analysts following an industry increases in our measure of industry closeness. From an investor’s standpoint, the new measure enables them to better discern whether the performance of a group of firms is related over time. From a researcher’s standpoint, the new measure enables them to more accurately control for industry effects by capturing the fact that within-industry similarity varies across industries.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3

Similar content being viewed by others

Notes

  1. e.g. Standard Industrial Classification (SIC), Global Industrial Classification (GIC), North American Industrial Classification (NAICS) and Fama-French (FF).

  2. The correlation matrix determinant captures co-movement in a different way than individual off-diagonal correlations.

  3. See Table 2.

  4. See Figs. 1 and 2.

  5. The highest decile of closeness.

  6. De Franco et al. (2011) provide evidence that their measure is positively related to analyst forecast accuracy and negatively related to analyst forecast dispersion.

  7. Our results are robust to using only earnings or cash flows in the calculation of \(C_1\).

  8. Based on the properties of the Pearson Correlation and determinants of two-by-two correlation matrices.

  9. Both types of co-movement indicate close groups since either type aids in predicting the fundamentals of the whole group by observing the fundamentals of one or more firms in that group.

  10. For example, certain accounting regulations may affect certain firms differently in how they account for specific economic transactions. This difference in accounting treatment would effect the earnings portion of our ex post closeness measure (not the cash flow portion). Our measure cannot delineate between the effect on ex post co-movement of the difference in accounting for given economic events and the ex ante exposure to different economic events.

  11. See, Docking et al. (1997), Aharony and Swary (1983), Bowen et al. (1983), Olsen and Dietrich (1985), Lang and Stulz (1992), Firth (1996), Ferris et al. (1997), Ettredge and Richardson (2003), Akhigbe et al. (2015) and Chang et al. (2005).

  12. GIC 6-digit industries are at the same level of granularity as SIC 2-digit industries.

  13. See Sect. 5 for the criteria we used in determining the “leader” of a given industry in a given year and a more detailed discussion on the trading strategy results.

  14. Results are robust to using 8 or 16 previous quarterly numbers as well as using 5 previous annual numbers. Also, although the GIC, NAICS and FF schemes came along before the start of our sample time period, COMPUSTAT back fills the firm industry codes for these schemes.

  15. Co-movement in earnings and cash flows needs to be measured over the same calendar time period. Otherwise there will be some macro-level economic events which some firms experience that other firms don’t experience. This will add noise to the closeness measures. 12/31 fiscal year-end firms comprise 67% of the firm-years in COMPUSTAT.

  16. We leave out the tables for each scheme displaying industry-year closeness for our entire sample time period due to space constraints. These tables are available upon request.

  17. Mean of the 25 \(\mathcal {C}\) values over 1990–2014 for each industry.

  18. e.g. See Pyo and Lustgarten (1990), Freeman and Tse (1992) and Gleason et al. (2008)

  19. Technically our sample period is from 1988–2014 but we only look at contagion effects resulting from earnings announcements between 1990–2014. We need the extra years because the closeness is calculated using earnings and cash flow data over the previous 12 quarters for each year between 1990–2014. Also, quarterly cash flow data (OANCFY) is not reported in the merged database until 1987.

  20. We use earnings before extraordinary items (IBQ) and net cash flows from operating activities (OANCFY) to measure earnings and cash flows respectively. To calculate operating cash flows in a quarter we take the quarterly change in OANCFY.

  21. Co-movement in earnings and cash flows needs to be measured over the same calendar time period. Otherwise there will be some macro-level economic events which some firms experience that other firms don’t experience. This will add noise to the closeness measures.

  22. Our results are robust although not as significant when we use “largest assets” to delineate the leader.

  23. There are 67, GIC 6-digit industries. See https://www.msci.com/gics.

  24. Using the Fama and French (1993) 3-factor model produces similar results.

  25. http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/

  26. We did not use analysts forecasts to evaluate the “news” due to limited data availability over our sample period. I/B/E/S has much more limited data coverage than CRSP (especially over the early years in our sample period) and we wanted to maintain a representative sample.

  27. A simple regression of the peer cumulative abnormal returns on the news of the leaders pooled over all three windows produces a coefficient of 0.0873 (p value = 0.0000 and \(R^2=0.28\%.\))

  28. The results are similar if comparing the contagion effects between the 1st and 5th quintiles of industry closeness.

  29. Petersen (2009)

  30. The Table 6 result is robust to the following specifications: (1) Using quarterly data over the previous 16 months (instead of 12 months) to measure \(\mathcal {C}\) for each industry-year; (2) Using annual accounting data over the previous 5 years to measure \(\mathcal {C}\) instead of quarterly data to measure \(\mathcal {C}\) for each industry-year; (3) Using net income and total cash flows instead of earnings before extraordinary items and operating cash flows as inputs in calculating \(\mathcal {C}\); (4) Measuring industry closeness using one of the other industrial classification schemes such as SIC2, NAICS3 or FF.

  31. Recall Fig. 3.

  32. We determine this in unreported simulation analysis.

  33. As an alternative to the Eq. (3) specification, we replaced cumulative abnormal returns with raw returns and included the cumulative market return as a control variable. The results were very similar to our main specification results from Table 6.

  34. See footnote 10 in GL (2008). They add an additional 185 restatement events identified through their extensive review of approximately 2200 pertinent WSJ articles. These 185 events make up less than 20% of their initial sample (before applying their sample criteria cuts) and there is no reason to believe these events are qualitatively different enough from the ones contained in the GAO database to justify their inclusion to maintain a “representative” sample. Furthermore, this process requires considerable amount of time and effort since 2200 WSJ articles must be read carefully. Consequently we do not augment our sample following GL (2008).

  35. Nearly every eight-digit GICS code was represented.

  36. We follow GL (2008) in this respect. Our results are entirely robust to using six-digit GIC codes or two-digit SIC codes.

  37. Again, we care about the closeness of the industry, not just the pair-wise closeness of firm i with each of its peers for the same reason as discussed earlier after H1 and H2.

  38. This is the same regression as regression equation (10) in De Franco et al. (2011) but we use our measure of closeness instead of their measure of comparability. The reader should compare our Table 9 analyst coverage results to their Table 5 results.

  39. These control variables are defined in Table 10.

  40. We also estimate Eq. (3) where \(\mathcal {C}\) is the average of all pair-wise closeness values between peer i and each of its peers instead of the usual specification which considers all possible pairs of firms in firm i’s industry. The coefficient on \(\mathcal {C}\) in this specification was also significantly greater than zero but not as significant (\(\beta _2 = 0.332\), t-stat = 4.73).

References

  • Aharony J, Swary I (1983) Contagion effects of bank failures: evidence from capital markets. J Bus 56:305–322

    Article  Google Scholar 

  • Akhigbe A, Madura J, Martin AD (2015) Intra-industry effects of negative stock price surprises. Rev Quant Financ Acc 45(3):541–559

    Article  Google Scholar 

  • Baginski SP (1987) Intraindustry information transfers associated with management forecasts of earnings. J Account Res 25:196–216

    Article  Google Scholar 

  • Barth ME, Kasznik R, McNichols MF (2001) Analyst coverage and intangible assets. J Account Res 39:1–34

    Article  Google Scholar 

  • Bhojraj S, Lee C (2002) Who is my peer? a valuation-based approach to the selection of comparable firms. J Account Res 40(2):407–439

    Article  Google Scholar 

  • Bhojraj S, Lee C, Oler DK (2003) What’s my line? a comparison of industry classification schemes for capital market research. J Account Res 41(5):745–774

    Article  Google Scholar 

  • Bhushan R (1989) Firm characteristics and analyst following. J Account Econ 11(2):255–274

    Article  Google Scholar 

  • Bowen RM, Castanias RP, Daley LA (1983) Intra-industry effects of the accident at three mile island. J Financ Quant Anal 18(01):87–111

    Article  Google Scholar 

  • Brennan MJ, Hughes PJ (1991) Stock prices and the supply of information. J Financ 46(5):1665–1691

    Article  Google Scholar 

  • Chan LK, Lakonishok J, Swaminathan B (2007) Industry classifications and return comovement. Financ Anal J 63(6):56–70

    Article  Google Scholar 

  • Chang S-C, Lai J-H, Yu C-H (2005) The intra-industry effect of share repurchase deregulation: evidence from taiwan. Rev Pac Basin Financ Mark Polic 8(02):251–277

    Article  Google Scholar 

  • Clinch GJ, Sinclair NA (1987) Intra-industry information releases: a recursive systems approach. J Account Econ 9(1):89–106

    Article  Google Scholar 

  • De Franco G, Kothari S, Verdi RS (2011) The benefits of financial statement comparability. J Account Res 49(4):895–931

    Article  Google Scholar 

  • Docking DS, Hirschey M, Jones E (1997) Information and contagion effects of bank loan-loss reserve announcements. J Financ Econ 43(2):219–239

    Article  Google Scholar 

  • Ettredge ML, Richardson VJ (2003) Information transfer among internet firms: the case of hacker attacks. J Inf Syst 17(2):71–82

    Google Scholar 

  • Fama EF, French KR (1993) Common risk factors in the returns on stocks and bonds. J Financ Econ 33(1):3–56

    Article  Google Scholar 

  • Ferris SP, Jayaraman N, Makhija AK (1997) The response of competitors to announcements of bankruptcy: an empirical examination of contagion and competitive effects. J Corp Financ 3(4):367–395

    Article  Google Scholar 

  • Firth M (1996) Dividend changes, abnormal returns, and intra-lndustry firm valuations. J Financ Quant Anal 31(02):189–211

    Article  Google Scholar 

  • Foster G (1981) Intra-industry information transfers associated with earnings releases. J Account Econ 3(3):201–232

    Article  Google Scholar 

  • Freeman R, Tse S (1992) An earnings prediction approach to examining intercompany information transfers. J Account Econ 15(4):509–523

    Article  Google Scholar 

  • Gleason CA, Jenkins NT, Johnson WB (2008) The contagion effects of accounting restatements. Account Rev 83(1):83–110

    Article  Google Scholar 

  • Han JC, Wild JJ (1990) Unexpected earnings and intraindustry information transfers: further evidence. J Account Res 28:211–219

    Article  Google Scholar 

  • Hoberg G, Phillips G (2016) Text-based network industries and endogenous product differentiation. J Polit Econ 124(5):1423–1465

    Article  Google Scholar 

  • Hrazdil K, Scott T (2013) The role of industry classification in estimating discretionary accruals. Rev Quant Financ Acc 40(1):15–39

    Article  Google Scholar 

  • Lang MH, Lundholm RJ (1996) Corporate disclosure policy and analyst behavior. Account Rev 467–492

  • Lang LH, Stulz R (1992) Contagion and competitive intra-industry effects of bankruptcy announcements: an empirical analysis. J Financ Econ 32(1):45–60

    Article  Google Scholar 

  • Lee CM, Ma P, Wang CC (2015) Search-based peer firms: aggregating investor perceptions through internet co-searches. J Financ Econ 116(2):410–431

    Article  Google Scholar 

  • O’Brien PC, Bhushan R (1990) Analyst following and institutional ownership. J Account Res 55–76

  • Olsen C, Dietrich JR (1985) Vertical information transfers: the association between retailers’ sales announcements and suppliers’ security returns. J Account Res 23:144–166

    Article  Google Scholar 

  • Petersen MA (2009) Estimating standard errors in finance panel data sets: comparing approaches. Rev Financ Stud 22(1):435–480

    Article  Google Scholar 

  • Pownall G, Waymire G (1989) Voluntary disclosure choice and earnings information transfer. J Account Res 27:85–105

    Article  Google Scholar 

  • Pyo Y, Lustgarten S (1990) Differential intra-industry information transfer associated with management earnings forecasts. J Account Econ 13(4):365–379

    Article  Google Scholar 

  • Ramnath S (2002) Investor and analyst reactions to earnings announcements of related firms: an empirical analysis. J Account Res 40(5):1351–1376

    Article  Google Scholar 

  • Veldkamp LL (2006) Information markets and the comovement of asset prices. Rev Econ Stud 73(3):823–845

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. School of Management, Binghamton University, Binghamton, USA

    Jonathan Ross & Anthony Meder

  2. Von Allmen School of Accounting, University of Kentucky, Lexington, USA

    David Ziebart

Authors
  1. Jonathan Ross
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. David Ziebart
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Anthony Meder
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Jonathan Ross.

Additional information

We thank Charles Lee for his helpful comments and suggestions which have improved the flow and contribution of our paper. We also gratefully acknowledge the helpful comments and suggestions of John Fellingham, Hong Xie and workshop participants at the University of Kentucky, Rutgers University and Binghamton University.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ross, J., Ziebart, D. & Meder, A. A new measure of firm-group accounting closeness. Rev Quant Finan Acc 52, 1137–1161 (2019). https://doi.org/10.1007/s11156-018-0739-0

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11156-018-0739-0

Keywords

JEL Classification

Search

Navigation