Discussion Paper: Are Analysts Providing Measurable ‘Added-Value’?

02:34 0 Comments

Once companies have provided financial guidance: Are analysts’ forecasts able to add any extra value? If so, how often and by how much?
A study by AKRO investiční společnost, comparing the accuracy of analyst forecasts and management forecasts, shows that so called ‘post event’ analyst forecasts, i.e. those made post recent results/management guidance, are in general more accurate than management forecasts. Both the frequency and magnitude of the greater accuracy prove significant, a somewhat reassuring conclusion for research analysts.
If analysts are able to provide insights with regard to tangible measures of value, it seems logical to assume analysts are also able to provide insights with regard to less tangible measures of value, e.g. management quality, industry outlook. At a time when the ‘active’ asset management industry is getting a bad press, and many research departments are being downsized[i], the results should give pause for thought.


Analysts’ Forecasts ‘Closer to the Mark’ More Often…
The study compared the accuracy of annual forecasts for constituents of the Japanese Nikkei 225 index, over an eleven year period, between the years 2006-2016. The study took advantage of an interesting feature of the Japanese equity market, whereby listed companies provide updated annual forecasts whenever they release their quarterly results.

Source: Bloomberg Earnings/ AKRO investiční společnost, a.s.
Nb. Over the 11 year period there are potentially 2475 (225x11) comparatives for each measure. Comparatives were made only when there was both a management forecast and a ‘post event’ analyst consensus prior to the final results being released.
Two measures of forecast accuracy were examined. The first measure simply compared the frequency with which ‘post event’ analyst forecasts proved more accurate than management guidance. Four key forecast variables were analysed: Sales, Operating Income, Net Income and EPS. A summary of the results are presented in Table: 1. Regardless of which item on the P&L account was being forecast, ‘post event’ analyst forecasts turn out to be closer to the final result significantly more often than management forecasts. Overall, analyst forecasts were closer on 57% of occasions, management on 39% of occasions while comparable accuracy was recorded on 4% of occasions. With the occasional exception, this bias in favour of analysts’ forecasts persisted over the entire period of the study.

…And By a Significant Margin
The second measure of forecast accuracy looked at the percentage the reported result deviated from forecast: Sometimes referred to as the degree of ‘surprise’. The lower the percentage deviation, the more accurate was the forecast.

Source: Bloomberg Earnings/ AKRO investiční společnost, a.s.
Forecasts which are more accurate, i.e. lower median percentage deviation, than management guidance, are highlighted in green.
The median percentage deviation of forecasts from the actual result, the level of surprise, is shown in table 2. For example, in terms of sales forecasts, the median percentage surprise was 0.87% for management forecasts, 0.86% for the ‘standard’ consensus forecast, and 0.73% for the ‘post event’ consensus. What is most striking from the table is that, whatever the P&L item, the ‘post event’ consensus is the most accurate predictor (lowest percentage deviation) of the final result, more accurate than either the ‘standard’ consensus or management forecasts. It is also clear that as one moves down the income statement, while the overall forecasting accuracy diminishes, the degree of ‘value-add’ analysts provide increases. For example, the median level of surprise for management forecasts of earnings per share (EPS) is 7.43% compared with 6.18% (a difference of 1¼%) for analysts’ ‘post-event’ forecasts.
This difference in forecast accuracy is really quite large: Remember, we are comparing the accuracy of forecasts prior to the announcement of the final results. With results for the first 9 months already announced, the 1 ¼ % difference in forecast accuracy for the 12 month results must come about in the final 3 months of the year. To illustrate this point, assuming there is no seasonality in the quarterly results, for the analysts’ 12 month eps forecasts to be 1 ¼ % more accurate, their forecasts for the final quarter must be approximately 5% (1¼% x 4)more accurate than those implied by the management forecast. A really impressive result!

Timeliness Matters
The practical implications are clear:
1.     In general, investors should utilise forecasts made ‘post event’, i.e. post the most recent results and management guidance.
2.     Where no ‘post event’ forecasts are available, investors should focus on company guidance.
3.     Least accurate are forecasts made prior to recent results/guidance.  Where possible these ‘out-of-date’ forecasts should be completely excluded as they materially detract from forecast accuracy[ii].
These common sense findings call into question the widespread use of ‘standard’ consensus forecasts. In table 2, the ‘standard’ consensus (which also includes older forecasts) is shown as significantly less accurate than the ‘post event’ consensus. In other words, the timeliness of forecasts is more important than the inclusion of a large number of forecasts. Part of the problem may be brokers’ willingness to publish forecasts even when they are blatantly out of date. In this context, ‘Post event’ consensus forecasts, compiled by information providers, by excluding ‘out of date’ estimates[iii], in fact provide considerable value-added. The findings also suggest, not unrealistically, that research has most potential for value-added when directed to ‘neglected’ stocks with no recent broker estimates.

Research Matters
At a time when ‘passive’ investing is in vogue, the study provides tangible evidence of the ‘value-added’ research analysts provide. Whilst forecasting results is but one tiny aspect of investment research, it is both tangible and measurable. That analysts can add-value with regard to financial forecasts is reassuring and probably means that analysts are also able to provide insights with regard to less tangible areas of valuation, e.g. competitive position, environmental responsibility, management quality, riskiness. In other words, scarce capital can be more efficiently allocated due to the efforts of research analysts.
Despite the added value which research clearly provides, the ‘revelation’ that the average investor cannot outperform the average investor has led many investors to abandon research-based investment in favour of low cost passive index based investment. As investment groups struggle to justify paying for research, the recent ‘unbundling ‘of research from commissions has, the author suspects, hastened this trend towards ‘dumbing down’ investment decisions. We have populated the investment universe with Lemmings, self-assured that they are no more a Lemming than other Lemmings and implicitly relying on the efforts of fewer and fewer genuinely active investors. Can we sleep soundly at night based on that premise?

Is There Anyone Left to Warn? Look Out!!!! There’s a Cliff Ahead!!!!








The Cost of Dumb Investment
There is, however, a hidden cost to the move away from fundamentally driven investment and it is a cost ultimately borne by society as a whole. Capital allocated on the basis of little or no fundamental research is more likely than not to be capital misallocated and we will all pay the price for that.
Poor capital allocation ultimately leads to lower growth in economic activity and more extreme economic cycles. It is difficult to see how an increasingly ‘research lite’ investment environment possibly helps. Chart 1 shows the trend in US nominal GDP since the end of WW2. Since the start of the 2008/9 financial crisis, nominal GDP has been markedly below trend. Indeed, if you believe that the stock market should reflect trends in underlying economy, the current valuation of the US S&P500 index needs to drop almost -44% to bring it back in line with GDP. In 2000 and 2007, the figures were -54% and –36% respectively. If investors had allocated capital more wisely, current levels of GDP would be much closer to trend (dashed line) and the stock market’s current valuation would look much less stretched. 

Chart 1: ‘’Mind the Gap’’ Nominal GDP, Trend GDP, and the US S&P500 Index Since 1947







Source: AKRO investiční společnost, a.s./Bloomberg
Nor is the deviation of GDP growth from its long-term trend just because of low inflation. Today’s somewhat deflationary environment is, of course, itself a consequence of poor investment decisions made prior to the 2000 dot-com bust and the 2007/8 crisis, e.g. reckless investing/ bank lending. However, even adjusting for inflation, there has been a step change down in the rate of GDP growth Chart 2). 

Chart 2: Growth in US Real GDP Has Moved Down a Gear 






Source: AKRO investiční společnost, a.s./Bloomberg
Would more research based investment have prevented the financial crisis? Quite possibly. Investing on the basis of index membership or partial 3rd party analysis (rating agencies, issuing banks) can never be a substitute for investment based on critical independent research.
One example, of such research, would be Terry Smith’s ‘’Accounting for Growth’’, 1992. A book, at the time both widely praised and condemned, which highlighted the aggressive accounting practices of many leading British companies. Should similarly well-argued research have appeared before the 2000 bust or the 2008/9 crisis, the economic consequences would have been profoundly positive. Instead, twice within the space of a decade, we witnessed capital destruction on a massive scale. Even relatively small insights, at the industry or corporate level, should therefore be encouraged.
With hindsight, investment blunders are generally avoidable and foreseeable. Analysts can add valuable insights but we need to promote a climate of intelligent investment; not today’s philosophy of scattergun investment. The investment industry should not aim for mediocrity because, if it leads to poor capital allocation, ultimately nobody benefits.

Diversification – No Free Lunch
If analysts, as our study suggests, are able to provide valuable insights, why doesn’t this translate into better performance? Some research suggests it does. Small boutique investment firms, with high conviction portfolios have, for example, been shown to be able to generate above average returns[iv].  However, for research based investment to flourish more generally, three issues, which have undermined the active management industry, need to be addressed: over-diversification, short-term investment horizons and low risk tolerances.
Until these issues are tackled, the ‘active’ management industry will remain under pressure.
Jeremy Monk MBA, ASIP, BSc (Hons), DIC
Investment Director, AKRO investiční společnost, a.s.
Prague.
18th October, 2016

For this article in Czech, please view this link: https://drive.google.com/open?id=0BySxYN0eDgfkRGtWbTF3d2tiUFU
The author would like to acknowledge the help of Sayo Ando, Takayuki Kutsuma and Beata Wijeratne, Equities – Global Data, Bloomberg London/Tokyo Markets in compiling the underlying data used in the study.
This article does not constitute investment advice or a recommendation to buy or sell any security. The opinions expressed in this article are the author’s and do not necessarily represent the views of AKRO investiční společnost, a.s.





[i] UK Investment Staff Increases by a Fifth. Rising Popularity of Cheap Passive Funds Could Cause Job Cuts, FTfm, October 2, 2016
[ii] Ota, Hiroshi. 2005. Yoso rieki no seido to kachi kanrensei: I/B/E/S, shikiho, keieisha yoso no hikaku (Accuracy of earnings forecasts and the correlation to valuation: A comparison of forecasts from I/B/E/S, Shikiho, and management guidance). Gendai Finance 28: 141-159. Ota’s research highlighted the poor accuracy of consensus estimates compared with management guidance.  AKRO investiční společnost’s research focused on comparing management forecasts with ‘post event’ forecasts. Excluded from the scope of the study were situations where the consensus consisted of only ‘out of date’, i.e. pre-event, forecasts. Visually inspecting the data shows that these were almost always less accurate than a subsequent management forecast. Inclusion of these ‘consensus’ forecasts, which don’t include a single up-to-date forecast, would have undoubtedly further reduced the overall accuracy the ‘standard’ consensus.
[iii] Bloomberg, for example, manually excludes those broker analysts that provide ‘pre-guidance/old estimates’ even when the company revised their estimates.
[iv] The Boutique Premium: Do Boutique Investment Managers Create Value? Affiliated Managers Group, Inc. (“AMG”), June 15, 2015

0 komentářů: