[MUSIC] So in this video we're going to talk about, is there any way to maybe get an informational edge using kind of modern technology, search engines, websites. You might think it's harder and harder to get an edge with all the information out there. In the 1800s and early 1900s, you could think of monitoring the number of fully-packed trains passing through a depot to get a heads up as to whether the current economic conditions. Like, hey, there seems to be a lot of corn going this way or a lot of coal going this way or a lot of empty cars, maybe that's signifying what the economic conditions are ahead of government reports. Certainly the pace of information transmission clearly has increased a lot over time with the age of the Internet. So are there any sources of information that others may overlook that could be useful in forecasting trends with the kind of hint on the word trends there at the end. Let's see what people are thinking about and how can you do it unless if you're a mind reader, that's easy. But if you're not, you can almost be a mind reader by looking at things such as Google Trends, okay? And here we're looking at, I access this in June 2012. So we see NBA draft 2012, Tom Cruise, Ann Curry, basically getting fired or demoted at the today show or all kinds of hot topics. But I just thought there's really a lot of interesting information here to kind of look at. One of the searches I put in was kind of something near and dear to like all of our hearts here at this point or maybe it's like a familiarity breeds contempt. And we don't want to hear this word anymore. But regardless capm, C-A-P-E. And when you look here at capm you actually see some interesting variation and search over time. There always seems to be this pick up in the spring and searches for capm, then a precipitous decline, then a pick up at the end of the, and then a sharp decline right at the end of the calendar year. So what's going on? What's going on with that? Can you think, what is kind of happening with capm to lead to these searches, okay? Well, let's actually formalizes. I so excited to talk about these results, I kind of stole the thunder from my buddy here. What explains the robust seasonal pattern and Google searches for CAPM? Let's think about that, and then I'll give you my take. So I asked this question to executive MBA class, MSF class, some will give this response right off the bat I'm about to show you. Others will give complicated stories that somehow searches for CAPM is related to the filing of taxes in April, therefore you have this kind of peak around April. And the search for CAPM, turns out it's just a very simple answer. Look at the Google search for the word final exam, okay? You also see a peak in around April and then you see a very strong peak at the end of the year with a precipitous decline. So people are searching for CAPM, at the same time they're searching for final exam, okay. So to me, this gives a little credibility check for the Google Trends data that's provided here, okay? So could we get an early sense of economic activity and this is just an example of one? Don't ask me why I seem to know a lot about coach purse and refer to it in a lot of examples. It's a long and sad story here. But believe me, I know more about coach purses that I care to let on here. So here we're looking at coach purse, and we're just seeing over time, how does their search volume compare? And you can see the high point seems to be the end of 2006, end of 2007. So you see this precipitous increase in searches for coach purse at the end of the year with a decline. So I don't think this is associated with CAPM or final exam. A natural hypothesis is, this is people looking at, what type of person do I want to buy right during the holiday season. Christmas, Hanukkah searching, Hey, what coach person I get? Look at 2008, how much the search volume dropped off from December 2007 in December 2006 and how much it dropped off further December 2009. Maybe there's some information in here that would be useful to investors and coach in terms of forecasting what their fourth quarter earnings will be in 2008. Maybe there's some useful information in this search volume that isn't all ready reflected in the stock price and is yet to come out in terms of the earnings report, which would be released sometime in the next year. So maybe there's some kind of nuggets here in seeing what people are searches, particularly in this focus search coach purse, right? It's higher at the end of the year, but it seems to be less high at the end of the years in 2008 and 2009, maybe forecasting lower sales, lower earnings during those holiday seasons, okay? Just another here, looking at the Google Trends housing bubble, okay? And here on the bottom, this is accessed in 2012 when Google Trends will just show you the amount of news coverage, news references to housing bubble. So you can see the housing bubble that's coming from just people searching on Google. This seems to be spiking in 2005, housing prices actually reached their peak in 2006. The news coverage using the term housing bubble only happened after housing prices are collapsing in 2008 into 2009. So it's interesting that the Google searches for housing bubble actually seemed to happen pretty close to the high price in prices and seemed then to be predictive of the housing decline, much more so than newspaper coverage of housing bubble. So you can see that news coverage happens clearly after the housing bubble already popped. The Google searches seem to proceed the housing bubble. So at least in this one case some predictability there from the searches on Google. Now, it's always useful to do a placebo test. Think of something where there might be no information and let's do a Google Trends search on that and see if nothing comes up. So I'm sorry, I was a little vain here. Let's search for Scott Weisbenner. And I have to be honest with you guys, we built this strong relationship or these two courses do not have enough search volume to show graph. So, kind of very sad ending here. So I would just say, can we please fix this travesty. Core Syrians unite start to search for my name, hashtag get Scott trendy. I don't have a twitter account but still we could create the hashtag get Scott trending. Let's fix this travesty if possible. So we did a few fund searches on Google Trends, we looked at CAPM final exam coach purse and then humiliating examination of Scott Weisbenner but we're going to fix that, right? And #getscotttrendy, I look forward to seeing that happen. Da Engelbert and Gao did a more systematic analysis looking at Google Trends and examined whether it predicts markets. So and particularly they looked at the frequency of searches in Google using Google Trends for stock tickers like MSFT or AAPL. And they looked at the search volume of these stock tickers, okay? Now why are we using stock tickers? Apple is a great example. People could be searching for apples because, hey I want to make apple pie for thanksgiving. So it's better to look at the stock ticker AAPL is indicating like people may be interested in buying the stock, okay? So the authors look at the search volume for a stock ticker in a given week and then they compare that to the median search volume for this ticker over the prior eight weeks. And then they calculate this Abnormal Search Volume Index or ASVI. So they're looking at searches this week and then the benchmark it relative to the medium of what the searches had been over the prior eight weeks, okay? So then the authors examined, does this ASVI, this Abnormal Search Volume Index predict future stock returns, okay? So you could think of two hypotheses here. So maybe we see these searches on Google for a stock's sticker, maybe it's predictive of short term buying for that stock, okay? And that's results in short-term upward price pressure, but it's not really based on any good information. So if we see a bunch of searches on Google, but it's not based on good information, we would predict that this liquidity effect should have a short term effect on prices, but not a long term effect, okay? On the other hand, are a bunch of searches on Google for a stock's ticker motivated by good information, some investors may have about that stock. Well, if that's the case, this revelation of good information that comes out in the future should predict both short and long term returns. We shouldn't see the return reversal, okay? So let's look at the results here that the authors have in this study. They're looking at the Abnormal Search Volume Index from a Google search on a stock ticker and then they're looking at kind of future returns, okay? So here we're looking at the sample period, January 2004 to June of 2008, okay? The dependent variable is the stock return for the firm relative to this Daniel Greenblatt, Chipman Warmers benchmark. So we haven't really talked about that, but just think if we're looking at the returns risk adjusted with this kind of benchmark here, okay? And the returns are reported in basis point. So 20 basis points would be 0.2%. The regression coefficients can be interpreted is the impact of a one standard deviation change, okay? So remember, we're looking at search volume on a given week, what does it predict about returns one week out, two weeks out, three weeks out, four weeks out up to weeks 5 to 52 here, okay? So are going what the search volume? A big increase in search in a particular week, predict about the stock's performance of that firm going up to a full year, okay? So first, let's look in the short term. So if we have a one standard deviation in abnormal search for a stock ticker, the stock for that firm increases in value 19 plus 15 here. We're talking about 33, 34 basis points, about 0.3% increase in the stock price over the next two weeks. So there is some short term increase in the price. Now the question is does that sustain or does that gradually reverse over time? If you add up the these coefficients here. These effects on return from a one standard deviation increase in search volume, 1915, that's roughly 34 kind of 38. But then minus 30 here when you add these two and if you have these coefficients across all the columns, the net coefficients about zero. So basically a year later the stock is at the same place where it started. So these searches for stock tickers are predictive of short term increase in the stock price. In weeks one and two after the searches consist of this liquidity effect, but not a long term increase in the stock price. Suggesting these searches on Google by a stock ticker aren't predictive of long term prospects of the firm, aren't predictive of good information, okay? Da Engelbert and Gao they also study how searches are related to the first day return and subsequent performance of Initial Public Offerings IPOs. So they're looking at the frequency of searches in Google using the company name if the firm hasn't had an IPO yet, it doesn't have a stock ticker. So they have to look at the company's name, like Facebook before Facebook had its IPO or Twitter before Twitter had its IPO so they can look at this search volume to how the IPO performs, okay? And we're going to look at what's the first day return, the stock trades that after its IPO And then we're also going to look at benchmark adjusted longer term returns. So for example returns 4 to 12 months after the IPO to see if there's any evidence of return reversals happen, okay? So what do we see when we look at the data? So let's break IPOs firms into two categories. Those that had low pre IPO searches, those that had high pre IPO searches. So those where there's a lot of searches on Google before the IPO have the bars on the left, bars are the right, they had a lot of searches before the IPO. So the black bars here are the AVS bridge IPO, first day return the gray bars are the median. For those firms where there wasn't much searching on google the company name before the IPO. They have lower first day returns than the companies where there was a lot of search prior to the IPO on the company name. Suggesting this kind of interest that was shown in Google searching for the name transferred over to the public markets leading to bidding up of the stock, okay? So this was looking at the Pre-IPO search volume and relating it to the average first day return. Now let's look at this pre IPO search volume and related to cumulative returns 4 to 12 months after the IPO. So we're looking at did this initial search that caused an increase in the price of the firm right after they had their IPO and were publicly traded. Was that increase in price sustained or did it lead to Was it reversed once we go 4 to 12 months after the IPO. So we have two firms here again. The black is the average return. The gray is the median return group one. They had a high first day return on their IPO But they had low search. Group two also had high first day returns but they had high search volume before the IPO. So maybe the thought processes of the authors is if we look at the second group on the right, maybe one of the reasons they had this high first day return is they had all these individual investors searching for the company. They were very aggressive in buying the stock that drove up the price temporarily because of the liquidity effect. But since it didn't reflect firm fundamentals, we should see that price ultimately reverse and see negative returns over the period for two month 12. Once we adjust here for the the industry industry performance here, this is exactly what we find. Those firms that had high first day return accompanied by this high search volume on Google before the IPO. Those are firms that did particularly poorly in months 4 to 12. As this initial boy up of the price due to all these individuals buying the stock. We saw they were interested in it because they searched for, they bid aggressively drove the stock up the first day it was public traded. But that enthusiasm wanes, that liquidity effect gradually gets reversed long term. These type of IPOS underperform their industry benchmarks. So again no predictability of the individual investors searches on Google does affect prices in the short term. But those short term effects are ultimately reversed in the long term. So we just looked at some research whether searches on Google particularly searches for stock ticker predict returns in the short term in the long term. And the short term there seems to be some evidence, that you see a bunch of Google searches for a particular stock ticker, there's maybe a buying effect, kind of some liquidy effect that drives up prices in the next week or two. But then that's reversed over the long term, and we saw that in the IPO market as well. Google is one thing we could look at, where's another place we can look at, you know maybe to get a sense of what our customers liking, does the customer know best. Is there any way to get a sense of what products customers like and which they don't before this information is fully reflected in stock prices. So obviously we know what products customers like after the fact exposed like news flash. A lot of people seem to like iPhone, okay? But the key thing is, can we identify people are liking these iPhone early on and invest in Apple stock before the market is a whole, fully processes this information, okay? So a colleague of mine, Huang, actually examines whether Amazon product ratings provide useful information. Do changes in Amazon review ratings, predict future stock returns and revenue earnings surprises. So this could be the point where I kind of go over to the tablet and talk about his results, but I had an idea kind of an epiphany here, why don't we make this a faculty focus episode? I love seeing this logo here, but this will be the last time we show it, this will be our fourth and final faculty focus episode. I'll be interviewing here, talking about his Amazon related research. It's an important interview when you see the rare purple shirt and purple tie combination. You only bring that out when it's going to be the final faculty focus episode. And we'll talk about in this, we'll have a conversation with how useful are changes in Amazon product ratings in predicting future stock returns and fundamental information about the firm. So stay tuned for that, I'm sure you'll enjoy it.
