Google has been making some waves recently over reports that it intends to enter the asset management industry following recent suitability studies the firm has conducted. Facebook has recently explored entering the financial services industry as well, as the company is reportedly seeking to set up an international monetary transactions service for Facebook users.
However, what really sets Google apart in its foray into the financial services industry are its advanced analytics capabilities from Google keyword search volume to advanced satellite imagery that could be leveraged to predict macroeconomic and company trends, which could ultimately be applied to developing highly advanced investment strategies.
Google Headquarters in Mountain View, California
Using “Google Trends” search volume for keywords can help predict economic variables like unemployment claims
Google Chief Economist Hal Varian (Photo Credit: Peter DaSilva/The New York Times)
Hal Varian, Google’s chief economist, co-authored a paper shortly after the financial crisis, demonstrating how Google Trends, which aggregates search volume for specific words or phrases, can accurately forecast unemployment claims, a key economic indicator which is released weekly by the Department of Labor. Being able to accurately forecast such a figure could certainly help an investment strategy to time the market on weekly unemployment claims releases and produce excess returns for clients. Certainly, this type of approach could be expanded to predicting other economically relevant variables ahead of market-moving government data releases.
Google Trends aggregate search volume for “Unemployment Claims” versus official Initial Jobless Claims figures
Several other firms have been using this approach to gain an informational advantage for quite some time using what are commonly known as “web scrapers” which are programs that scrape and aggregate data off of the internet. Hedge funds like Bridgewater Associates with $87.1 bn in hedge fund assets and AQR Capital Management with $29.9 bn as of January 1, 2014 are both well-known for their advanced web-scraping capabilities significantly contributing their excess returns and their rapid growth in assets over the past 10 years, putting them among the top 10 largest hedge funds in the world.
There is little doubt that Google could follow suit and push the envelope even further by leveraging their search tools and technology infrastructure to identify and capture various information asymmetries that exist in the financial marketplace.
Google could use satellite imagery from its newly acquired Skybox to predict company earnings
Another type of information extraction that Google could employ in investment strategies comes from Neil Currie, a stock analyst at UBS, who famously forecasted sales figures for Walmart by using satellite photos to observe parking lot volume across Walmart stores.
Google’s interest in entering the asset management industry is remarkably well-timed with Google’s recent acquisition of Skybox Imaging, a company specializing in high quality satellite imaging that has produced the first HD video of earth from space ever. Of course, this type of satellite imaging could be applied to analyzing and extracting all sorts of company data. For instance, satellite imagery could be expanded to determine inventory for car manufacturers or the amount of traffic coming through seaports. It could be used to analyze the progress of mining operations or infrastructure development. It could be used to study the supply chains of multinational companies like AppleApple. The list of applications of satellite imaging is really endless.
Google acquisition Skybox Imaging captures world’s first high-resolution, HD video of Earth from space
Regulatory concerns and a technological arms race for asset managers
There are some concerns about hurdles that would be in front of Google entering the asset management industry. Catherine Tillotson, managing partner at Scorpio Partnership, a wealth management consultancy commented to the FT that “Entering a highly regulated industry is not something you do lightly. There probably is a subsection of investors who would have confidence in Google, but I think the vast majority of investors want a relationship with an entity which can supply them with high quality information, market knowledge and a view on that market. I think it is unlikely they would turn to Google for those qualities.”
On the other hand, many see Google with its strong technology infrastructure as standing out among an industry where there is a technological arms race happening as a recent PwC white paper released in January, entitled “Asset Management 2020: A Brave New World”, asserted that some fund managers’ failure to keep up with technological change will create opportunities for groups like Google, Apple, Twitter, or Amazon to break into the market.
Google’s experience with venture capital and investing its own money
While Google is in talks to begin an asset management arm that would manage client assets, in 2010 the internet group launched a California-headquartered trading operation (which also has a bank-style trading floor) to manage its own cash better. Not to mention, Google’s venture capital arm, Google Ventures, has invested in more than 189 companies including Uber, a taxi app, and Kensho, a financial analytics firm.
With the technology and analytics infrastructure that Google has combined with the small finance infrastructure it has already built out to manage its own capital, entering the asset management space really seems like a no-brainer for Google.
By Mark Skousen “A balanced Input-Output framework…provides a more accurate and consistent picture of the U. S. economy.”
Starting in spring 2014, the Bureau of Economic Analysis will release a breakthrough new economic statistic on a quarterly basis. It’s called Gross Output, a measure of total sales volume at all stages of production. GO is almost twice the size of GDP, the standard yardstick for measuring final goods and services produced in a year.
This is the first new economic aggregate since Gross Domestic Product (GDP) was introduced over fifty years ago.
It’s about time. Starting with my work The Structure of Production in 1990 and Economics on Trial in 1991, I have made the case that we needed a new statistic beyond GDP that measures spending throughout the entire production process, not just final output. GO is a move in that direction – a personal triumph 25 years in the making.
GO attempts to measure total sales from the production of raw materials through intermediate producers to final retail. Based on my research, GO is a better indicator of the business cycle, and most consistent with economic growth theory.
GO is a measure of the “make” economy, while GDP represents the “use” economy. Both are essential to understanding how the economy works.
While GDP is a good measure of national economic performance, it has a major flaw: In limiting itself to final output, GDP largely ignores or downplays the “make” economy, that is, the supply chain and intermediate stages of production needed to produce all those finished goods and services. This narrow focus of GDP has created much mischief in the media, government policy, and boardroom decision-making. For example, journalists are constantly overemphasizing consumer and government spending as the driving force behind the economy, rather than saving, business investment, and technological advances. Since consumer spending represents 70% or more of GDP, followed by 20% by government, the media naively concludes that any slowdown in retail sales or government stimulus is necessarily bad for the economy. (Private investment comes in a poor third at 13%.)
For instance, the New York Times recently reported, “Consumer spending makes up more than 70% of the economy, and it usually drives growth during economic recoveries.” (“Consumers Give Boost to Economy,” New York Times, May 1, 2010, p. B1) Or as the Wall Street Journal stated a few years ago, “The housing bust has chilled consumer spending — the largest single driver of the U. S. economy…” (“Home Forecast Calls for Pain,” Wall Street Journal, September 21, 2011, p. A1.)
Or take this report during the economic recovery:
“Friday’s estimates of second-quarter gross domestic product [1.3%, well below consensus forecasts] provided a sobering look at how a decline in public spending and investment can restrain growth….The astonishingly slow growth rate from April through June was due in large part to sluggish consumer spending and an increase in imports, which subtract from growth numbers. But dwindling government spending also held back growth.” (“The Role of Government Spending,” New York Times, July 29, 2011.)
In short, by focusing only on final output, GDP underestimates the money spent and economic activity generated at earlier stages in the production process. It’s as though the manufacturers and shippers and designers aren’t fully acknowledged in their contribution to overall growth or decline.
Gross Output exposes these misconceptions. In my own research, I’ve discovered many benefits of GO statistics. First, Gross Output provides a more accurate picture of what drives the economy. Using GO as a more comprehensive measure of economic activity, spending by consumers turns out to represent around 40% of total yearly sales, not 70% as commonly reported. Spending by business (private investment plus intermediate inputs) is substantially bigger, representing over 50% of economic activity. That’s more consistent with economic growth theory, which emphasizes productive saving and investment in technology on the producer side as the drivers of economic growth. Consumer spending is largely the effect, not the cause, of prosperity.
Second, GO is significantly more sensitive to the business cycle. During the 2008-09 Great Recession, nominal GDP fell only 2% (due largely to countercyclical increases in government), but GO collapsed by over 7%, and intermediate inputs by 10%. Since 2009, nominal GDP has increased 3-4% a year, but GO has climbed more than 5% a year. GO acts like the end of a waving fan. (See chart below.)
I believe that Gross Output fills in a big piece of the macroeconomic puzzle. It establishes the proper balance between production and consumption, between the “make” and the “use” economy, and it is more consistent with growth theory. As Steve Landefeld, director of the BEA, and co-editors Dale Jorgenson and William Nordhaus state in their work, A New Architecture for the U. S. National Accounts (University of Chicago Press, 2006), “Gross output [GO] is the natural measure of the production sector, while net output [GDP] is appropriate as a measure of welfare. Both are required in a complete system of accounts.”
The history of these two economic statistics goes back to several pioneers. Two economists in particular had much in common — they were both Russian Americans who taught at Harvard University, and both won the Nobel Prize. Simon Kuznets did breakthrough work on GDP statistics in the 1930s. Following the Bretton Woods Agreement in 1946, GDP became the standard measure of economic growth. A few years later, Wassily Leontief developed the first input-output tables, which he regarded as a better measure of the whole economy. I-O accounts require examining the “intervening steps” between inputs and outputs in the production process, “a complex series of transactions…among real people.”
I-O data created the first estimates of Gross Output. However, GO was not emphasized as an important macroeconomic tool until my own work, The Structure of Production, was published in 1990 by New York University Press. In chapters 6 and 9, I created a universal four stage model of the economy (see the diagram below) demonstrating the relationship between total spending in the economy and final output.
In chapter 6, I made the point that GDP was not a complete picture of economic activity, and compared it to GO for the first time, contending that GO was more comprehensive and more accurately revealed that business investment was far bigger than consumption in the economy.
Since writing Structure, I discovered that the BEA’s Gross Output does not include all sales at the wholesale and retail level. The BEA only includes value-added data for commodities after they become finished products. Gross sales are ignored at the final two stages of production. David Wasshausen, a BEA staff researcher, offers this rationale: since “there is no further transformation of these goods…to the production process, they are excluded from wholesale/retail trade output.”
Therefore, in the 2nd edition of Structure, published in 2007, I created my own aggregate statistic, Gross Domestic Expenditures (GDE), which includes gross sales at the wholesale and retail level and is therefore significantly larger (more than double GDP). For a comparison between GDE, GO and GDP, see my working paper.
The BEA has been compiling GO statistics from input-output data for years, but the media have largely ignored these figures because they came out only every five years (known as benchmark I-O tables). Since the early 1990s, the BEA has been estimating industry accounts annually. Even so, the data was never up-to-date like GDP. (The latest input-output industry accounts are for 2011).
That has gradually changed. Under the leadership of BEA director Steve Landefeld, the BEA now has the budget to report the input-output data, including Gross Output, on a quarterly basis, and has already begun publishing quarterly data prior to 2012. This is a major breakthrough involving the cooperation of the Bureau of the Census, Bureau of Labor Statistics, the Federal Reserve Board, and other government agencies.
Controversies Over This New Statistic
Several objections have been made over the years to the use of GO and GDE. Economists are especially fixated over the perceived problem of “double counting” with GO and GDE. I am the first to note that GO and GDE involve double counting. A commodity is often sold repeatedly as it goes through the resource, production, wholesale and retail stages. Why not just measure the value added at each stage rather than double or triple count? they ask. GDP eliminates double counting and measures only the value added at each stage.
There are several reasons why double counting should not be ignored and is actually a necessary feature to understanding the overall economy. As accountants and financiers know, double counting is essential in business. No company can operate or expand on the basis of value added or profits only. They must raise the capital necessary to cover the gross expenses of the company — wages and salaries, rents, interest, capital tools and equipment, supplies and goods-in-process. GO and GDE reflect this vital business decision making at each stage of production. Can publicly-traded firms ignore sales/revenues and only focus on earnings when they release their quarterly reports? Wall Street would object. Aggregate sales/revenues are important to measure on an individual firm and national basis.
In my own research, I find it interesting that GO and GDE are far more volatile than GDP during the business cycle. As noted in the chart above, sales/revenues rise faster than GDP during an expansion, and collapse during a contraction (wholesale trade fell 20% in 2009; retail trade dropped over 7%).
Economists need to explore the meaning of this cyclical behavior in order to make accurate forecasts and policy recommendations. Double counting counts.
Another objection involves outsourcing and merger/acquisitions. Companies that start outsourcing their products will cause an increase in GO or GDE, while companies that merge with another company will show a sudden decrease, even though there is essentially no change in final output (GDP).
That’s a legitimate concern. Similar problems occur with GDP. When a homeowner marries the maid, the maid may no longer be paid and therefore her services may no longer be included in GDP. Black market activities often fail to show up in GDP data as well. Certainly if a significant trend develops in outsourcing or merger & acquisition activity, it will be reflected in GO or GDE statistics, but not necessarly in GDP. It bears further investigation to see how serious this issue is. No aggregate statistic is perfect, but GO and GDE offer forecasters an improved macro picture of the economy.
In conclusion, GO or GDE should be the starting point for measuring aggregate spending in the economy, as it measures both the “make” economy (intermediate production), and the “use” economy (final output). It complements GDP and can easily be incorporated in standard
national income accounting and macroeconomic analysis. To see how, take a look at the 4th edition of my textbook, Economic Logic (Capital Press, 2014), available in paperback and Kindle.
Mark Skousen is editor of Forecasts & Strategies and a Presidential Fellow at Chapman University in 2014. He is the author of The Structure of Production (New York University Press, 1990, 2007), which introduced the concept of Gross Output as an essential macroeconomic tool.