536 The Journal of Finance Section V we estimate the transactions costs of following the strategies of buying the most highly rated stocks and selling short those that are least favorably rated and discuss the profitability of these strategies.We partition our sample by firm size and reexamine the returns to our strategies in Sec- tion VI.A summary and conclusions section ends the paper. I.The Data,Sample Selection Criteria, and Descriptive Statistics The analyst recommendations used in this study were provided by Zacks Investment Research,which obtains its data from the written and electronic reports of brokerage houses.The recommendations encompass the period from 1985(the year that Zacks began collecting this data)through 1996. Each database record includes,among other items,the recommendation date, identifiers for the brokerage house issuing the recommendation and the an- alyst writing the report (if the analyst's identity is known),and a rating between 1 and 5.A rating of 1 reflects a strong buy recommendation,2 a buy,3 a hold,4 a sell,and 5 a strong sell.This five-point scale is commonly used by analysts.If an analyst uses a different scale,Zacks converts the analyst's rating to its five-point scale.Ratings of 6 also appear in the Zacks database and signify termination of coverage. Another characteristic of the database,one that has not been explicitly acknowledged in any prior study as far as we are aware,is that the data made available to academics does not constitute Zacks'complete set of rec- ommendations.According to an official at Zacks,some individual brokerage houses have entered into agreements that preclude their recommendations from being distributed by Zacks to anyone other than the brokerage houses' clients.Consequently,although the recommendations of most large and well- known brokers are included,the recommendations of several large broker- age houses are not part of this academic database (although they are represented in Zacks'consensus statistics).4,5 The Zacks database contains 378,326 observations for the years 1985 through 1996.Dropping those for the 1,286 firms not appearing on the CRSP file leaves a final sample of 361,620 recommendations.Table I provides descrip- tive statistics for these recommendations.As shown in column 3,the number of firms covered by Zacks has increased steadily over the years.For the year 4 For the first year in which we compute recommendation returns,1986,the Zacks database includes the recommendations of 12 of the 20 largest brokerage houses,in terms of capital employed.(Capital levels are taken from the Securities Industry Yearbook (1987,1997).)The capital of these 12 brokerage houses comprises 54 percent of the total capital of these largest houses.For the last year of recommendation returns,1996,the Zacks database includes the recommendations of 12 of the 19 largest brokerage houses(the 20th does not prepare analyst recommendations),whose capital comprises 49 percent of the total capital of these largest houses. 5 Supplementary tests performed using the First Call database(which includes these large brokerage house recommendations)suggest that these omissions do not have a significant effect on our results.See footnote 20
Section V we estimate the transactions costs of following the strategies of buying the most highly rated stocks and selling short those that are least favorably rated and discuss the profitability of these strategies. We partition our sample by firm size and reexamine the returns to our strategies in Section VI. A summary and conclusions section ends the paper. I. The Data, Sample Selection Criteria, and Descriptive Statistics The analyst recommendations used in this study were provided by Zacks Investment Research, which obtains its data from the written and electronic reports of brokerage houses. The recommendations encompass the period from 1985 ~the year that Zacks began collecting this data! through 1996. Each database record includes, among other items, the recommendation date, identifiers for the brokerage house issuing the recommendation and the analyst writing the report ~if the analyst’s identity is known!, and a rating between 1 and 5. A rating of 1 reflects a strong buy recommendation, 2 a buy, 3 a hold, 4 a sell, and 5 a strong sell. This five-point scale is commonly used by analysts. If an analyst uses a different scale, Zacks converts the analyst’s rating to its five-point scale. Ratings of 6 also appear in the Zacks database and signify termination of coverage. Another characteristic of the database, one that has not been explicitly acknowledged in any prior study as far as we are aware, is that the data made available to academics does not constitute Zacks’ complete set of recommendations. According to an official at Zacks, some individual brokerage houses have entered into agreements that preclude their recommendations from being distributed by Zacks to anyone other than the brokerage houses’ clients. Consequently, although the recommendations of most large and wellknown brokers are included, the recommendations of several large brokerage houses are not part of this academic database ~although they are represented in Zacks’ consensus statistics!. 4,5 The Zacks database contains 378,326 observations for the years 1985 through 1996. Dropping those for the 1,286 firms not appearing on the CRSP file leaves a final sample of 361,620 recommendations. Table I provides descriptive statistics for these recommendations. As shown in column 3, the number of firms covered by Zacks has increased steadily over the years. For the year 4 For the first year in which we compute recommendation returns, 1986, the Zacks database includes the recommendations of 12 of the 20 largest brokerage houses, in terms of capital employed. ~Capital levels are taken from the Securities Industry Yearbook ~1987, 1997!.! The capital of these 12 brokerage houses comprises 54 percent of the total capital of these largest houses. For the last year of recommendation returns, 1996, the Zacks database includes the recommendations of 12 of the 19 largest brokerage houses ~the 20th does not prepare analyst recommendations!, whose capital comprises 49 percent of the total capital of these largest houses. 5 Supplementary tests performed using the First Call database ~which includes these large brokerage house recommendations! suggest that these omissions do not have a significant effect on our results. See footnote 20. 536 The Journal of Finance
Security Analyst Recommendations and Stock Returns 537 22.5 的 号的 9 8 ici oici 点 222 SISATBuV 举 到 200.1 600.5 2691 sioxoig (OD) 6 9 品 三 互 超 强 三 S SWIH POJ0A0O g 9 g 2 日9 日日 日日 2 isAleuy iod 9 见 2 吧 Y 9 Y 2 9 9 9 I SIqeL swLnJ polaoo jo oqunu auI,eok Aq 'al umi01 yos bupseN/XWV/HSAN dSHO ay uo pos!I sunJ IIe sopnpou!swnJ poas!l jo oqunu oyL A[rep dsyo ou u posn somtmoos le jo on[ea axew ay pue swryj poioAoo Ile jo suorqez
538 The Journal of Finance 1996,59.8 percent of all firms on the NYSE,AMEX,or Nasdag have at least one recommendation in the database(column 4).The market capitalization of these firms constitute 95.6 percent of the capitalization of all firms in the market (column 5).This is consistent with the conventional wisdom that analysts tend to cover larger firms,because they offer more liquidity and allow the analysts'clients to more easily take large positions in the firms' shares (which,in turn,generates larger commissions revenues for the bro- kerage houses). From 1986 onward,the mean number of analysts per covered firm has generally been increasing (column 6),whereas the median number has re- mained constant (column 7).The mean and median number of covered firms per analyst has also been stable(columns 8 and 9).Additionally,the number of brokerage houses contributing recommendations to Zacks and the number of analysts providing forecasts has steadily increased over time(columns 10 and 11).The last column of the table reports the average of all of the analyst ratings,by year.It shows a rather steady decrease over time,indicating that analysts'recommendations have become more favorable.6 A 6 x 6 transition matrix of the analysts'recommendations appears in Table II.Each cell fi,j}of the matrix contains two numbers.The top one is the number of observations in the database in which an analyst moved from a recommendation of i to one of j;the bottom number is the median number of calendar days between the announcement of a recommendation of i and a revised recommendation ofj.The diagonal elements of the matrix reflect reiterations of analyst recommendations.Most of the entries in this matrix are concentrated in the upper 3 x 3 cells.This is to be expected,given the conventional wisdom that analysts are reluctant to issue sell recommenda- tions.Within this region,the bulk of the observations represent reiterations. The mean time between a recommendation and its reiteration is a little less than 300 days.This is much longer than the mean time between a recom- mendation and a revision by the analyst to a new rating,which is generally in the low 100-day range.To the extent that the Zacks database does not record all reiterations,such a difference is not surprising. The line entitled "First Zacks Recommendation"records the first recom- mendation in the database for a given analyst-company pair.Consistent with McNichols and O'Brien (1998),the first recommendation is usually a buy (1 or 2),less often a hold,and rarely a sell(4 or 5).This again reflects the reluctance of analysts to issue sell recommendations.This observation is also consistent with the numbers in the last two lines of the table.Of all the recommendations in the database,47.1 percent are buys whereas only 5.7 per- cent are sells.Excluding observations with a rating of 6,buys constitute 54.1 percent of the total,whereas sells make up only 6.5 percent. 6 The year 1985 has,by far,the smallest number of covered firms,brokerage houses,and analysts,likely because it is the first year that Zacks began tracking recommendations.Because the 1985 data is so sparse,we do not include the investment returns from that year in our analysis
1996, 59.8 percent of all firms on the NYSE, AMEX, or Nasdaq have at least one recommendation in the database ~column 4!. The market capitalization of these firms constitute 95.6 percent of the capitalization of all firms in the market ~column 5!. This is consistent with the conventional wisdom that analysts tend to cover larger firms, because they offer more liquidity and allow the analysts’ clients to more easily take large positions in the firms’ shares ~which, in turn, generates larger commissions revenues for the brokerage houses!. From 1986 onward, the mean number of analysts per covered firm has generally been increasing ~column 6!, whereas the median number has remained constant ~column 7!. The mean and median number of covered firms per analyst has also been stable ~columns 8 and 9!. Additionally, the number of brokerage houses contributing recommendations to Zacks and the number of analysts providing forecasts has steadily increased over time ~columns 10 and 11!. The last column of the table reports the average of all of the analyst ratings, by year. It shows a rather steady decrease over time, indicating that analysts’ recommendations have become more favorable.6 A 6 3 6 transition matrix of the analysts’ recommendations appears in Table II. Each cell $i, j% of the matrix contains two numbers. The top one is the number of observations in the database in which an analyst moved from a recommendation of i to one of j; the bottom number is the median number of calendar days between the announcement of a recommendation of i and a revised recommendation of j. The diagonal elements of the matrix reflect reiterations of analyst recommendations. Most of the entries in this matrix are concentrated in the upper 3 3 3 cells. This is to be expected, given the conventional wisdom that analysts are reluctant to issue sell recommendations. Within this region, the bulk of the observations represent reiterations. The mean time between a recommendation and its reiteration is a little less than 300 days. This is much longer than the mean time between a recommendation and a revision by the analyst to a new rating, which is generally in the low 100-day range. To the extent that the Zacks database does not record all reiterations, such a difference is not surprising. The line entitled “First Zacks Recommendation” records the first recommendation in the database for a given analyst–company pair. Consistent with McNichols and O’Brien ~1998!, the first recommendation is usually a buy ~1 or 2!, less often a hold, and rarely a sell ~4 or 5!. This again reflects the reluctance of analysts to issue sell recommendations. This observation is also consistent with the numbers in the last two lines of the table. Of all the recommendations in the database, 47.1 percent are buys whereas only 5.7 percent are sells. Excluding observations with a rating of 6, buys constitute 54.1 percent of the total, whereas sells make up only 6.5 percent. 6 The year 1985 has, by far, the smallest number of covered firms, brokerage houses, and analysts, likely because it is the first year that Zacks began tracking recommendations. Because the 1985 data is so sparse, we do not include the investment returns from that year in our analysis. 538 The Journal of Finance
Security Analyst Recommendations and Stock Returns 539 Table II Transition Matrix of Analyst Recommendations (Number,Median Calendar Days),1985 to 1996 This table shows the number and the median calendar days between changes in or reiterations of recommendations.The first row reports all changes from a recommendation of 1 ("strong buy")to 1,2("buy"),3("hold"),4("sell"),5("strong sell")or discontinuation of coverage,and the total across the columns.The sixth and seventh rows identify recommendations for firms that were previously dropped from coverage and for firms for which coverage was initiated in the database.Fractional recommendations are rounded to the nearest whole value. To Recommendation of: From Recommendation of: 2 ￥ 5 Dropped Total 34,939 15,269 16,887 538 805 9,802 78,240 293 109 128 140 135 121 2 14,010 21.936 17.581 1,349 468 8,177 63,521 95 299 115 106 111 121 3 12.945 14,492 52,813 3,971 2,958 15,332 102,511 113 112 291 114 116 123 4 480 1,180 3,913 2,936 668 1,097 10,274 132 103 98 245 98 135 5 396 316 2,739 439 1.409 1,143 6,442 95 105 94 90 301 99 Dropped 4,951 3,507 5,999 546 400 5,013 20,416 73 65 92 102 110 59 First Zacks 26,053 19,817 24,458 2.392 1,531 5.965 80.216 recommendation Total 93,774 76,517 124,390 12,171 8,239 46,529 361.620 of total 25.9 21.2 34.4 3.4 2.3 12.9 of non-drops 29.8 24.3 39.5 3.9 2.6 We also compute the average three-day announcement period return for changes in or initiations of analyst recommendations.These returns are pre- sented in Table III.Similar to the results of Stickel (1995)and Womack (1996),we find that the compound(size-adjusted)return for the three-day period centered on the day a rating change is announced is,in general,sig- nificantly positive for upgrades and significantly negative for downgrades.7 7 Using the First Call database,Womack(1996)reports three-day returns that are much higher in magnitude than those documented here.This is consistent with his assertion that there are occasional delays in the recording of some of the recommendations in the Zacks data- base.(The difference may also be because Womack's sample consists only of large brokerage house recommendations.If these recommendations are accorded more publicity,this could lead to the market reaction being larger in the few days around their announcement.)As we report in footnote 20,though,supplementary tests using First Call data suggest that any timing issues surrounding Zacks do not have a significant effect on our main results
We also compute the average three-day announcement period return for changes in or initiations of analyst recommendations. These returns are presented in Table III. Similar to the results of Stickel ~1995! and Womack ~1996!, we find that the compound ~size-adjusted! return for the three-day period centered on the day a rating change is announced is, in general, significantly positive for upgrades and significantly negative for downgrades.7 7 Using the First Call database, Womack ~1996! reports three-day returns that are much higher in magnitude than those documented here. This is consistent with his assertion that there are occasional delays in the recording of some of the recommendations in the Zacks database. ~The difference may also be because Womack’s sample consists only of large brokerage house recommendations. If these recommendations are accorded more publicity, this could lead to the market reaction being larger in the few days around their announcement.! As we report in footnote 20, though, supplementary tests using First Call data suggest that any timing issues surrounding Zacks do not have a significant effect on our main results. Table II Transition Matrix of Analyst Recommendations (Number, Median Calendar Days), 1985 to 1996 This table shows the number and the median calendar days between changes in or reiterations of recommendations. The first row reports all changes from a recommendation of 1 ~“strong buy”! to 1, 2 ~“buy”!, 3 ~“hold”!, 4 ~“sell”!, 5 ~“strong sell”! or discontinuation of coverage, and the total across the columns. The sixth and seventh rows identify recommendations for firms that were previously dropped from coverage and for firms for which coverage was initiated in the database. Fractional recommendations are rounded to the nearest whole value. To Recommendation of: From Recommendation of: 1 2 3 4 5 Dropped Total 1 34,939 15,269 16,887 538 805 9,802 78,240 293 109 128 140 135 121 2 14,010 21,936 17,581 1,349 468 8,177 63,521 95 299 115 106 111 121 3 12,945 14,492 52,813 3,971 2,958 15,332 102,511 113 112 291 114 116 123 4 480 1,180 3,913 2,936 668 1,097 10,274 132 103 98 245 98 135 5 396 316 2,739 439 1,409 1,143 6,442 95 105 94 90 301 99 Dropped 4,951 3,507 5,999 546 400 5,013 20,416 73 65 92 102 110 59 First Zacks recommendation 26,053 19,817 24,458 2,392 1,531 5,965 80,216 Total 93,774 76,517 124,390 12,171 8,239 46,529 361,620 % of total 25.9 21.2 34.4 3.4 2.3 12.9 % of non-drops 29.8 24.3 39.5 3.9 2.6 Security Analyst Recommendations and Stock Returns 539
540 The Journal of Finance Table II Three-day Percentage Market-adjusted Returns Associated with Announcements of Changes in and Reiterations of Analyst Recommendations,1985 to 1996 This table shows the percentage market-adjusted returns measured for the day before,the day of,and the day following changes in and reiterations of analyst recommendations.For example, the first row reports the returns associated with all changes from a recommendation of 1 (strong buy)to 1,2(buy),3(hold),4(sell),5(strong sell),or discontinuation of coverage.Re- turns are measured as the three-day buy and hold return less the return on a value-weighted NYSE/AMEX/Nasdaq index.The sixth and seventh rows show the returns associated with recommendations for firms that were previously dropped from coverage,and for firms for which coverage was initiated,respectively.Fractional recommendations are rounded to the nearest whole value.t-statistics,estimated using cross-sectional standard errors,are shown below the returns.Each t-statistic pertains to the hypothesis that the mean size-adjusted abnormal re- turn is zero.(The number of observations in each cell is shown in Table II.) To Recommendation of: From Recommendation of: 1 2 4 分 Dropped 1 0.177 -0.889 -2.192 -1.305 -3.021 -0.020 7.525 -17.448 -32.841 -4.129 -6.792 -0.364 1.059 0.114 -1.415 -0.638 -0.999 0.115 21.565 3.809 -25.876 -3.154 -2.187 2.135 1.488 1.066 0.015 -1.054 -0.976 0.112 27.895 22.877 0.788 -10.195 -5.926 2.630 4 0.723 0.610 0.610 -0.130 -0.336 0.393 3.388 4.105 6.908 -1.399 -1.226 2.347 0.607 1.296 0.400 -0.283 -0.005 0.207 2.113 4.384 3.487 -0.964 -0.032 0.999 Dropped 0.637 0.301 0.051 -1.168 -0.474 8.586 3.533 -0.810 -4.728 -1.463 First Zacks 1.093 0.479 -0.149 -0.209 -0.650 recommendation 29.445 13.150 -4.736 -2.135 -4.384 Furthermore,for the set of initial analyst-company recommendations in the database,a buy rating (1 or 2)is accompanied by a significantly positive return,as expected,whereas a hold or sell rating (3,4,or 5)is associated with a significantly negative return. II.Research Design A.Portfolio Construction To determine whether investors can profit from analysts'consensus rec- ommendations,we construct calendar-time portfolios based on the consen- sus rating of each covered firm.The average analyst rating,Ar,for firm i on date-1 is found by summing the individual ratings,Ar,of the
Furthermore, for the set of initial analyst–company recommendations in the database, a buy rating ~1 or 2! is accompanied by a significantly positive return, as expected, whereas a hold or sell rating ~3, 4, or 5! is associated with a significantly negative return. II. Research Design A. Portfolio Construction To determine whether investors can profit from analysts’ consensus recommendations, we construct calendar-time portfolios based on the consensus rating of each covered firm. The average analyst rating, AN it21, for firm i on date t 2 1 is found by summing the individual ratings, Aijt21, of the Table III Three-day Percentage Market-adjusted Returns Associated with Announcements of Changes in and Reiterations of Analyst Recommendations, 1985 to 1996 This table shows the percentage market-adjusted returns measured for the day before, the day of, and the day following changes in and reiterations of analyst recommendations. For example, the first row reports the returns associated with all changes from a recommendation of 1 ~strong buy! to 1, 2 ~buy!, 3 ~hold!, 4 ~sell!, 5 ~strong sell!, or discontinuation of coverage. Returns are measured as the three-day buy and hold return less the return on a value-weighted NYSE0AMEX0Nasdaq index. The sixth and seventh rows show the returns associated with recommendations for firms that were previously dropped from coverage, and for firms for which coverage was initiated, respectively. Fractional recommendations are rounded to the nearest whole value. t-statistics, estimated using cross-sectional standard errors, are shown below the returns. Each t-statistic pertains to the hypothesis that the mean size-adjusted abnormal return is zero. ~The number of observations in each cell is shown in Table II.! To Recommendation of: From Recommendation of: 1 2 3 4 5 Dropped 1 0.177 20.889 22.192 21.305 23.021 20.020 7.525 217.448 232.841 24.129 26.792 20.364 2 1.059 0.114 21.415 20.638 20.999 0.115 21.565 3.809 225.876 23.154 22.187 2.135 3 1.488 1.066 0.015 21.054 20.976 0.112 27.895 22.877 0.788 210.195 25.926 2.630 4 0.723 0.610 0.610 20.130 20.336 0.393 3.388 4.105 6.908 21.399 21.226 2.347 5 0.607 1.296 0.400 20.283 20.005 0.207 2.113 4.384 3.487 20.964 20.032 0.999 Dropped 0.637 0.301 0.051 21.168 20.474 8.586 3.533 20.810 24.728 21.463 First Zacks recommendation 1.093 0.479 20.149 20.209 20.650 29.445 13.150 24.736 22.135 24.384 540 The Journal of Finance