A broad segment of the industry invests based on established aspects resembling value, momentum and low risk. In this post we share the important thing findings of our study study of out-of-sample aspects over a big and economically essential sample period. Using the longest sample period thus far – 1866 to 2020s – we allay concerns about data mining and underperformance of equity aspects. We find that out-of-sample equity aspects are robust and have been a pervasive phenomenon in financial markets for greater than 150 years.
Concerns about data mining are real
Why did we conduct this study? First, more research on factor premiums is required, particularly using out-of-sample data. Most practitioner studies of equity aspects use samples dating back to the Nineteen Eighties or Nineties, covering roughly 40 to 50 years. From a statistical perspective, this just isn’t a big amount of knowledge. In addition, these years were unique, characterised by few recessions, the longest expansion and bull market in history, and only minimal episodes of inflation until 2021. Academic studies of stock aspects often use longer samples, typically starting in 1963 using the US Center for Research in Security Prices (CRSP) database from the University of Chicago. But imagine if we could double that sample size using a comprehensive dataset of stock prices. Long before the twentieth century, stock markets were crucial to economic growth and financing innovation.
Second, scientists have discovered lots of of things – sometimes called the “factor zoo.” Recent scientific research suggests that lots of these aspects could also be as a consequence of data corruption or statistical coincidences attributable to extensive testing by each scientists and industry researchers. A single test typically has a 95% confidence level, meaning that about one in 20 tests will “discover” a false factor. This problem is exacerbated when multiple tests are performed. This is crucial as tens of millions of tests have been carried out within the financial markets. This is a serious concern for investors as factor investing has grow to be mainstream worldwide. Imagine if the aspects driving lots of of billions of dollars of investment were the results of statistical noise and subsequently unlikely to provide returns in the long run.
Figure 1 illustrates one in every of the motives of our study. It shows the test statistics for size, value, momentum, and low-risk portfolios over the in-sample and out-of-sample periods inside the CRSP era (post-1926). Consistent with previous studies, most aspects show significance throughout the in-sample period. However, the outcomes look significantly different in subsequent out-of-sample periods as several aspects lose their significance at traditional confidence levels. This decline in equity factor performance could be attributed to several reasons, including limited data samples as discussed within the literature. Regardless, it highlights the necessity for independent out-of-sample tests of justice aspects in a sufficiently large sample. In our research Paperwe address this challenge by testing out-of-sample equity aspects in a sample that has never been touched before by expanding the CRSP data set to incorporate 61 years of knowledge.
Figure 1.
Source: Global Financial Data, Kenneth French website, Erasmus University Rotterdam
Stock markets within the nineteenth centuryTh century
Before diving into the important thing findings, let’s provide an summary of the US stock market within the nineteenth century. In our work, we collect information from all major stocks listed on US stock exchanges between 1866 and 1926 (the beginning date of the CRSP dataset). This period was characterised by strong economic growth and rapid industrial development, which laid the inspiration for the United States to grow to be the world’s leading economic power. Stock markets played a critical role in economic growth and funding innovation, with market capitalization growing greater than 50-fold in 60 years – consistent with U.S. nominal GDP growth over the identical period.
In some ways, the markets of the nineteenth and twentieth centuries were similar. Stocks could easily be bought or sold on exchanges through dealer firms, traded through derivatives and options, bought on margin, and shorted from known short sellers. Major 19Th Technical innovations of the century resembling the telegraph (1844), the transatlantic cable (1866), the introduction of the ticker tape (1867), the provision of local telephone lines (1878) and direct telephone connections via cable enabled a fluid and lively secondary marketplace for stocks, extensive brokerage and market making activities, rapid arbitrage between prices, rapid price reactions to information and large-scale trading activities. Price quotations were immediately known from coast to coast and even across the Atlantic. Similar to today, investors had access to a big selection of reputable sources of data, while a large industry of monetary analysts provided market assessments and investment advice.
Furthermore, trading costs within the nineteenth century weren’t significantly different from those within the twentieth centuryTh century costs. Market information and academic studies show that transaction costs for higher volume and well-arbitrated NYSE stocks are around 0.50%, however the minimum tick is 1/8Th in each centuries. Furthermore, the median quoted spread on the NYSE in the last decade before World War I used to be 86 basis points and 1 / 4 of trades were made with spreads of lower than 36 basis points. Additionally, stock turnover in NYSE stocks was higher between 1900 and 1926 than it was in 2000. Overall, U.S. stock markets have been a vibrant and economically essential source of trading for the reason that nineteenth centuryTh Century and represents a vital and reliable out-of-sample test field for factor premiums.
The pre-CRSP stock dataset
Creating this dataset was a significant effort. Our sample includes stock returns and characteristics of all major stocks since 1866. Why 1866? It is the beginning date of , a key source also utilized by the CRSP database. You could also be wondering why CRSP starts in 1926. While the precise reason stays speculative, it appears arbitrary and warrants the inclusion of some data from before the 1929 stock market crash.
In our work, we recorded all market capitalizations by hand – extremely relevant for examining factor premiums and stock prices. Additionally, we hand-validated samples of price and dividend data Global financial data — an information provider specializing in historical price data. Unlike CRSP, we focused data collection on all major stocks traded on major exchanges. These include not only the NYSE, but additionally the NY Curb (which later became the American Stock Exchange, AMEX) and several other regional exchanges. You can imagine how much work this took and the way much time the research assistants at Erasmus University Rotterdam took up. But the outcomes were well worth the effort. The result’s a high-quality dataset of U.S. stock prices from 1866 to 1926, covering roughly 1,500 publicly traded stocks.
The performance of out-of-sample aspects is everlasting
So what are the out-of-sample results from the pre-CRSP period from 1866 to 1926? Before we discuss this, we would love to remind you that this era has not been studied extensively and subsequently allows us to conduct a real out-of-sample test of equity factor premiums.
Figure 2 summarizes the important thing results of our research. It shows the alpha of the determined equity factor premiums over the longest possible CRSP sample (in gray) and the out-of-sample period before CRSP (in black). Interestingly, the out-of-sample alphas for value, momentum and low-risk aspects are very just like those of the CRSP sample. In fact, the differences between the 2 samples are statistically insignificant. More than 150 years of evidence on factor premiums (the black bars) confirms this conclusion and shows attractive premiums which might be each economically and statistically significant. Overall, the independent sample confirms the validity of essential equity factor premiums resembling value, momentum and low risk.
Figure 2.
Source: Global Financial Data, Kenneth French website, Erasmus University Rotterdam
These results suggest several strong conclusions. First and most significantly, factor premiums are an everlasting feature of monetary markets. They are usually not artifacts of researchers’ efforts or specific economic conditions, but have existed for the reason that emergence of monetary markets and have existed for greater than 150 years. Second, factor premiums don’t decline outside the sample but are likely to remain stable. Third, factor premiums offer significant investment opportunities as a consequence of their durability. These results should give investors greater confidence within the robustness of factor premia and strengthen their usefulness in developing effective investment strategies.
