We examine whether the contribution of firm-level accounting earnings to the informativeness of the aggregate is tilted towards earnings with specific financial reporting characteristics. Specifically, we investigate whether considering the smoothness of firm-level earnings increases the informativeness of aggregate earnings for future real GDP, and if so, whether macroeconomic forecasters use this information efficiently. Using recently-developed mixed data sampling methods, we find that the aggregate is tilted towards firms with smoother earnings and that this composition of aggregate earnings outperforms traditional weighting schemes. Further, this tilted aggregate has a stronger positive association with forecast revisions; in fact, analysts who utilize earnings the most in their forecasts appear to fully impound the informativeness of earnings smoothness. Our results synthesize and span parallel yet distinct streams of research on the role of accounting earnings in firm-level and macroeconomic outcomes and suggest an important role for financial reporting characteristics in the aggregate.
This is a preview of subscription content, access via your institution.
Buy single article
Instant access to the full article PDF.
Price excludes VAT (USA)
Tax calculation will be finalised during checkout.
Smoothness is often considered a measure of a firm’s earnings quality in other contexts. In this study, we do not take a position on whether smoother earnings reflects higher or lower earnings quality. Instead, our interest lies in simply understanding whether and to what extent smoothness influences the differential informativeness of firm-level earnings for aggregate outcomes.
Although we focus on smoothness, the goal of this study is to measure, test, and report whether financial reporting characteristics can survive aggregation to enhance the aggregate informativeness of earnings, rather than attempting to identify the best financial reporting characteristic from a menu of options. We leave examinations of other firm-level reporting attributes to future research for which our empirical framework provides guidance.
Mixed data sampling models, developed by Ghysels et al. (2006), have been applied almost exclusively to temporal differences in the frequency of observation among variables, such as quarterly versus daily (e.g., Ball and Gallo 2018; Ball and Ghysels 2018). In our setting, mixed data sampling refers to cross-sectional differences in the frequency of observations because thousands of firm-level quarterly earnings are observed in the cross section for every one quarterly observation of real GDP. Along with Anderson et al. (2009) and Ball et al. (2018), our study is among the first to apply mixed data sampling methods in the cross-section.
We incorporate both equal and value weighting into our aggregation benchmark, as these are the two commonly used weighting schemes in the aggregate earnings literature.
For example, an equal-weighted aggregate treats each dollar of (scaled) earnings the same (i.e., applies the same coefficient). In this study, we allow each firm’s contribution to the aggregate to tilt away from the benchmark aggregate earnings as a function of firm-specific characteristics, which allows the coefficient to vary across firms in accordance with each firm’s earnings smoothness parameter.
Finding a significant tilt towards firms with smoother earnings in the association between aggregate earnings and future GDP does suggest that smoothness, on average, reflects either an attribute of the fundamental earnings process or management’s communication of relevant private information, as opposed to nefarious earnings manipulation (which would not be expected to have a stronger relationship with real economic outcomes).
We do not assert whether forecasters explicitly use earnings smoothness in constructing their forecasts. Smoothness may be associated with some other characteristic that makes these firms more relevant for their forecasts. Nevertheless, our findings do suggest that discriminating by earnings smoothing is a potentially fruitful approach.
Specifically, corporate profits are measured primarily using tax data, which are collected with a significant lag and then extrapolated to estimate corporate profits. In addition to tax returns, the BEA uses surveys as well as some financial statement information to measure corporate profits. This number is then revised over time.
See Dechow et al. (2010) for an overview of the earnings quality literature.
We choose market value of assets as our scaling variable in an effort to abstract away from the firm’s financing choice. Scaling by sales or market value of equity—two common approaches in the aggregate earnings literature—yields qualitatively similar results.
For example, firms with a fiscal quarter ending in January, February, or March of 2014 would all be included in our aggregate of 2014 first quarter earnings, so long as the earnings are announced before the end of April 2014.
Konchitchki and Patatoukas (2014a, pg. 80) state that “according to Thomas Stark, Assistant Director and Manager of the Real-Time Data Research Center, ‘almost no one uses the advance estimate’ when measuring GDP growth forecast errors.”
The Philadelphia Fed sends the questionnaire to its panel of forecasters at the end of the first month of the calendar quarter, after the advance release of GDP. The deadline for responses is set for the second or third week of the middle month of the calendar quarter. The results are always released before the second report is released by the BEA at the end of the middle month of the quarter. For example, in the first quarter, the questionnaire is sent to panelists at the end of January and is due back in the middle of February, with results released in mid- to late-February. In our sample, the average number of forecasters is 36.
In our setting, we observe firm-level quarterly earnings growth for hundreds of firms every calendar quarter for every one quarterly observation of real GDP growth.
Using a predetermined representative agent utility function, Brandt et al. (2009) formulate approximated portfolio allocations, using a similar weighting scheme. Their analysis does not involve the estimation of regression models with the purpose of aggregating cross-sectional data, as we do in our analysis.
For example, the benchmark component is equivalent to applying an equal weight to all observations if the estimated value of 𝜃 = 0. Conversely, the benchmark component is equivalent to applying a value weight to all observations if the estimated value of 𝜃 = 1. We allow 𝜃 to take on a range of values, which increases the flexibility of the benchmark component.
Technically speaking, the estimated smoothness-weighted aggregate earnings presumably reduces the measurement error that results from imposing an equal weight when constructing aggregate earnings, which ultimately biases the coefficient βEARN.
Anderson, E., Ghysels, E., Juergens, J. (2009). The impact of risk and uncertainty on expected returns. Journal of Financial Economics, 94(2), 233–263.
Ball, R., & Sadka, G. (2015). Aggregate earnings and why they matter. Journal of Accounting Literature, 34, 39–57.
Ball, R.T., Bonham, J., Hemmer, T. (2018). Does it pay to “Be Like Mike”? Aspirational peer firms and relative performance evaluation. Working paper, University of Michigan, University of Chicago, and Rice University.
Ball, R.T., & Gallo, L. (2018). A mixed data sampling approach to accounting research, Working paper, University of Michigan.
Ball, R.T., & Ghysels, E. (2018). Automated earnings forecasts: beat analysts or combine and conquer? Management Science, 64(10), 4936–4952.
Bonsall, S., Bozanic, Z., Fischer, P. (2013). What do management earnings forecasts convey about the macroeconomy? Journal of Accounting Research, 51(2), 225–266.
Brandt, M., Santa-Clara, P., Valkanov, R. (2009). Parametric portfolio policies: exploiting characteristics in the cross-section of equity returns. Review of Financial Studies, 22(9), 3411–3447.
Crawley, M. (2014). Macroeconomic consequences of accounting: the effect of accounting conservatism on macroeconomic indicators and the money supply. The Accounting Review, 90(3), 987–1011.
Dechow, P. (1994). Accounting earnings and cash flows as measure of firm performance: the role of accounting accruals. Journal of Accounting and Economics, 18(1), 3–42.
Dechow, P., Ge, W., Schrand, C. (2010). Understanding earnings quality: a review of the proxies, their determinants and their consequences. Journal of Accounting and Economics, 50(2–3), 344–401.
DeFond, M., & Park, C. (1997). Smoothing income in anticipation of future earnings. Journal of Accounting and Economics, 23(2), 115–139.
Demerjian, P., Lewis-Western, M., McVay, S. (2017). How does intentional earnings smoothing vary with managerial ability? Journal of Accounting, Auditing and Finance. https://doi.org/10.1177/0148558X17748405.
Dichev, I. (2018). Comparing GAAP to NIPA earnings. Working paper, Emory University.
Gallo, L., Hann, R., Li, C. (2016). Aggregate earnings surprises, monetary policy, and stock returns. Journal of Accounting and Economics, 62(1), 103–120.
Ghysels, E., Santa-Clara, P., Valkanov, R. (2006). Predicting volatility: getting the most out of return data sampled at different frequencies. Journal of Econometrics, 131(1–2), 59–95.
Konchitchki, Y., & Patatoukas, P. (2014a). Accounting earnings and gross domestic product. Journal of Accounting and Economics, 57(1), 76–88.
Konchitchki, Y., & Patatoukas, P. (2014b). Taking the pulse of the real economy using financial statement analysis: implications for macro forecasting and stock valuation. The Accounting Review, 89(2), 669–694.
Kothari, S.P. (2001). Capital markets research in accounting. Journal of Accounting and Economics, 31(1–3), 105–231.
Kothari, S.P., Lewellen, J., Warner, J. (2006). Stock returns, aggregate earnings surprises, and behavioral finance. Journal of Financial Economics, 79(3), 537–568.
Kothari, S.P., Lewellen, J., Warner, J. (2014). The behavior of aggregate corporate investment. Working paper, MIT, Dartmouth College, and University of Rochester.
Laurion, H., & Patatoukas, P. (2016). From micro to macro: does conditional conservatism aggregate up in the national income and product accounts? Journal of Financial Reporting, 1(2), 21–45.
Leuz, C., Nanda, D., Wysocki, P. (2003). Earnings management and investor protection: an international comparison. Journal of Financial Economics, 69 (3), 505–527.
Nallareddy, S., & Ogneva, M. (2016). Predicting restatements in macroeconomic indicators using accounting information. The Accounting Review, 92(2), 151–182.
Newey, W., & West, K. (1987). A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica, 55(3), 703–708.
Shivakumar, L., & Urcan, O. (2017). Why does aggregate earnings growth reflect information about future inflation? The Accounting Review, 92(6), 247–276.
Subramanyam, K. (1996). The pricing of discretionary accruals. Journal of Accounting and Economics, 22(1–3), 249–281.
Tucker, J., & Zarowin, P. (2006). Does income smoothing improve earnings informativeness? The Accounting Review, 81(1), 251–270.
We appreciate helpful comments from Robert Bushman, Peter Easton (editor), Dennis Fixler, Rebecca Hann, Greg Miller, Zack Miller, Scott Richardson, Gil Sadka, Lakshmanan Shivakumar, an anonymous reviewer, and participants at the 2018 CARE conference on Firm-Level Information and the Macroeconomy and the Southern Economics Association’s Annual Meeting as well as seminars at the University of Iowa, the University of North Carolina at Chapel Hill, the University of Washington, and the University of Michigan’s Paton Center brown bag series.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: timeline and variable descriptions
Appendix: timeline and variable descriptions
1.1 A.1 Timeline
1.2 A.2 Macro-level variables
GDPq+ 1 Third (or final) estimate of real GDP growth for calendar quarter q+ 1 (released approximately three months after the end of quarter q+ 1) obtained from the Real-Time Data Set for Macroeconomists provided by the Philadelphia Fed.
REVq+ 1 SPF forecast revision of real GDP growth for calendar quarter q+ 1 defined as the difference between the mean of individual SPF forecasts issued during q+ 1 (Eq+ 1 [GDPq+ 1]) and the mean of individual SPF forecasts issued during q (Eq [GDPq+ 1]).
FEq+ 1 Real GDP growth forecast error defined as the difference between the third estimate of real GDP growth in calendar quarter q+ 1 (GDPq+ 1) and the mean of the most recent SPF individual forecasts of real GDP growth in q+ 1 (Eq+ 1 [GDPq+ 1]) issued approximately 45 days prior to the end of q + 1.
GDPq The first estimate of real GDP growth for quarter q (released approximately one month after the end of calendar quarter q) obtained from the Real-Time Data Set for Macroeconomists provided by the Philadelphia Fed.
GDPq−k Third (or final) estimate of real GDP growth for calendar quarter q −k (released approximately three months after the end of quarter q −k) obtained from the Real-Time Data Set for Macroeconomists provided by the Philadelphia Fed.
YLDq Yield on the one-year Treasury bill with constant maturity measured one month after the end of calendar quarter q retrieved from the Federal Reserve Bank of St. Louis (FRED).
SPRDq Difference between the yields, measured one month after the end of calendar quarter q, on the 10-year Treasury bond with constant maturity and the one-year Treasury bill with constant maturity, which are both retrieved from FRED.
RETq Buy-and-hold market return, including distributions, on the value-weighted market portfolio over a three-month window that ends one month after the end of calendar quarter q.
Nq Number of firms with a fiscal quarter ending in calendar quarter q that report earnings no later than one month after the end of calendar quarter q and have non-missing data for all required variables.
1.3 A.3 Firm-level variables
earnn, q Quarterly earnings growth of firm n for the fiscal quarter ending in calendar quarter q defined as the difference between n’s quarterly net income before extraordinary items in q and q − 1 scaled by the market value of assets (i.e., market value of equity plus book value of liabilities) at the beginning of the fiscal quarter.
cfon, q Quarterly growth of cash from operations of firm n for the fiscal quarter ending in calendar quarter q, which is equal to the difference between n’s quarterly cash from operations in q and q − 1 scaled by the market value of assets (i.e., market value of equity plus book value of liabilities) at the beginning of the fiscal quarter. Cash from operations is net income before extraordinary items minus the change in working capital accruals, which is defined as the change in the book value of current assets net of changes in cash minus the change in book value current liabilities net of changes in the current portion of long-term debt.
smthn, q Firm n’s earnings smoothness measured as the standard deviation of quarterly earnings growth, σ(earnn, q), divided by the standard deviation of quarterly growth in cash from operations, σ(cfon, q), where standard deviations are estimated over a 20-quarter rolling window ending with calendar quarter q − 1.
About this article
Cite this article
Ball, R.T., Gallo, L. & Ghysels, E. Tilting the evidence: the role of firm-level earnings attributes in the relation between aggregated earnings and gross domestic product. Rev Account Stud 24, 570–592 (2019). https://doi.org/10.1007/s11142-019-09493-0