HomeFundamental ValueThe predictability of market-wide earnings revisions

The predictability of market-wide earnings revisions

-

Forward earnings yields are a key metric for the valuation of an equity market. Helpfully, I/B/E/S and DataStream publish forward earnings forecasts of analysts on a market-wide index basis. Unfortunately, updates of these data are delayed by multiple lags. This can make them inaccurate and misleading in times of rapidly changing macroeconomic conditions. Indeed, there is strong empirical evidence that equity index price changes predict future forward earnings revisions significantly and for all of the world’s 25 most liquid local equity markets. This predictability can be used to enhance the precision of real-time earnings yield data and avoid misleading trading signals.

The post ties in with SRSV’s summary lecture on Macro Information Efficiency/ Fundamental Value.

The post is based on proprietary technical notes of Macrosynergy Partners and SRSV Ltd.

Market-wide equity earnings estimates and their publication lags

I/B/E/S global aggregates provide bottom-up company earnings forecasts and related data for country and major international equity indices. The data are based on predictions of equity analysts of various institutions.  Company level data presented on a 12-month-forward basis are calculated prior to aggregation. The 12 months are counted from the current date. They are effectively a weighted average of current fiscal year earnings and next fiscal year earnings, with the weights reflecting the number of months in the 12-month period falling into current and next fiscal year respectively. Data is also presented on an 18-month-forward basis, comprising the 12 month period starting in 6 months’ time.

Until 2006 I/B/E/S global aggregates had been updated only once per month, typically on the Tuesday following the 3rd Friday of the month based on snapshot taken on the Thursday before the 3rd Friday of every month at close of business in New York. In March, 2006 weekly-updated aggregates were introduced.

Weekly aggregate data are calculated every Thursday. The earnings data used in this calculation is as of the previous Thursday, however. Moreover, after calculation data are only being made available on Datastream by the following Wednesday (by close of business). This procedure implies that the minimum delay between updated earnings forecasts and their publication is 13 calendar days or 9 working days. The actual delay between analyst earnings revisions and publication on an index basis should be longer for two reasons. First, data are not updated for 4 days subsequent to the release weekly. Second, analysts themselves cannot update forecasts continuously and the publication of their numbers naturally comes with a delay. All considered, the effective average updating delay of new information is probably between 3 and 6 weeks, depending on circumstances. And a lot can happen in equity markets over 3-6 weeks.

The predictability of market-wide earnings forecast revisions

Given the explicit and implicit delays explained above, it is implausible that any event over the 2-3 weeks prior to release of index-level forward earnings estimates could have had any impact on the released data. By contrast, market prices over this period do include the information of relevant changes over the lag period. Thus, it stands to reason that equity price changes between the lag period and a previous reference period contain information on forthcoming revisions to earnings estimates.

For the empirical analysis we investigate the predictive power of past equity index price changes with respect to earnings expectations revisions over the subsequent month. We conduct our analysis for a panel of 25 local-currency country equity indices, and a history from 2006 (beginning of weekly index earnings estimates) to 2018. For details of indices and the meaning of currency symbols see the annex below.

The equity index price increase is calculated as the percent change of the average equity index price of the last 3 weeks relative to the preceding three months, at a monthly rate. The price change horizon assumes that earnings predictions at the beginning of a 3-week lag period have been stale on average 1 1/2 months, which we consider a realistic assumption. We do, however, cross-check our result using an equity price increase of the last 2 weeks versus the previous 2 months, which would posit considerably better earnings update efficiency.

The empirical analysis confirms the predictive power of market changes on earnings revisions across the world. Since 2006 the relation between the past index price changes (% 3-week average over previous 3 months, monthly rate) and subsequent 1-month changes in forward earnings has been positive in all 25 countries of our panel.

An estimate of elasticity of future earnings-per-share revisions to index price changes can be derived from a global panel regression based on a simple pooling model. The estimate of the global average elasticity of forward EPS changes over the next month to past index price changes has been roughly 0.65, with very high significance. This means that for each 10% (monthly rate) decline in the index price, the forward EPS forecasts has on averaged been revised down by about 6.5%. Accounting for this expected change hence significantly mitigates short-term distortions of earnings yield estimates that arise from delayed updating of earnings estimates.

There is a possibility that, even under consideration of lags, the impact of new information on earnings revisions does not occur over one month but is spread, or distributed, over several time periods. Put simply, some institutions may be even more inert in updating earnings estimates than we suggested. Econometrically we can test for such inertia through a “distributed lag structure” by including 1-month lagged earnings revisions as explanatory variable. Indeed, the coefficient of a lagged dependent variable has been highly significant, albeit small at just above 0.06. That means a 10% revision of forward EPS in one month points to a 0.6% revision over the subsequent month. This suggests that while the chosen lag and lookback periods do not capture the full extent of the inertia of the revision process, the neglected part of the inertia is small.

Using a shorter lookback period for equity returns (percent change of 2-week average over previous 2 months, monthly rate) reduces the elasticity of forward EPS revisions to past index price changes 0.46, but increases its elasticity to lagged revisions to 0.11. This is plausible, because the shorter the considered lookback, the more uncaptured revision inertia there should be. The main take-away is that shorter lookback horizons leave a more meaningful quantity of unexplained inertia and are thus not advisable for the purpose of using market changes in order to predict earnings forecast revisions.

The “market adjustment” of forward earnings ratios

Plausibility and empirical findings suggest that past equity index price changes can be used to predict forthcoming EPS revisions. The predictions can then be employed to adjust earnings yields. This should increase the precision of real-time available market-wide forward earnings yields. This seems to be particularly appropriate in times of large recent information shocks that have altered the market-wide outlook for corporate profits. For example, in case of a large negative shock the resulting drawdown in prices alongside stale unrevised forward earnings would bias forward earnings yields to the high side, turning them into a misleading trading signal.

This “market-adjustment”” does, by construction, not alter general medium-term pattern of forward earnings yields across times. It only corrects some short-term distortions. The normal range of correction has been modest, with a panel standard deviation of 1.2% of the yield. For an average forward earnings yield of 8%, the resulting standard correction of the earnings yield in %-points based on previous price changes would have been 0.1%-point, giving a range of 7.9-8.1%, a small quantity given the 2.75%-points overall panel standard deviation of earnings yields. In case of large market shocks the adjustment of forward earnings yields would have been more meaningful. Indeed, adjustments of earnings yields by 5% or more have not been uncommon in turbulent times. A 5% forward EPS adjustment would change an 8% forward earnings yield to 7.6%/8.4% on the low and high side respectively.

Annex: The 25 markets of the empirical analysis

Futures contracts have been chosen for the following local indices (alphabetically by currency symbol):

AUD: Australian Stock Exchange (ASX) 200 or ASX 200 (200 constituents as of August 2018).
BRL: Brazil Bovespa (67 constituents).
CAD: Toronto Stock Exchange 60 Index (60 constituents).
CHF: Swiss Market or SMI (20 constituents).
CNH: Hang Seng China Enterprises (50 constituents, “H-shares”, actually quoted in HKD).
CNY: Shanghai Shenzhen CSI 300 (300 constituents, “A-shares”).
DEM: Germany DAX 30 Performance/ Xetra (30 constituents, actually quoted in EUR).
ESP: Spain IBEX 35 (35 constituents, actually quoted in EUR).
FRF: France CAC 40 (40 constituents, actually quoted in EUR).
GBP: UK FTSE 100 (101 constituents).
INR: India CNX Nifty (50 constituents).
ITL: Italy FTSE MIB Index (40 constituents).
JPY: Nikkei 225 Stock Average (225 constituents).
KRW: Korea Stock Exchange KOSPI 200 (201 constituents).
MXN: Mexico IPC (35 constituents).
MYR: FTSE Bursa Malaysia KLCI (30 constituents).
NLG: Netherlands AEX Index (25 constituents, actually quoted in EUR).
PLN: Warsaw General Index 20 (20 constituents).
SEK: OMX Stockholm 30 (30 constituents).
SGD: MSCI Singapore Free (25 constituents).
THB: Bangkok SET 50 (50 constituents).
TRY: Turkey Bist National 30 (30 constituents).
TWD: MSCI Taiwan (89 constituents).
USD: Standard and Poor’s 500 Composite (500 constituents).
ZAR: South africa FTSE / JSE Top 40 (42 constituents).

Editor
Editorhttps://research.macrosynergy.com
Ralph Sueppel is managing director for research and trading strategies at Macrosynergy. He has worked in economics and finance since the early 1990s for investment banks, the European Central Bank, and leading hedge funds.