How meticulous work results in insights from 150 years of data

In recent years, the search for variables, or characteristics of various companies, that would result in continuous and predictable greater returns has devolved into an intellectual arms race. Many of these vanish when the variable is required to live outside of the sample. One of the main issues is that the majority of academics are mining a small number of well-known datasets. These databases date all the way back to before the Great Depression of 1929, but not much beyond.

Piecing together all available information

Baltussen has compiled a database of individual US security trades dating all the way back to 1866, when the nation was emerging from its civil war. The researchers gathered a group of passionate students willing to put up with the meticulous task of scanning newspapers and other data from the 1860s, long before the Wall Street Journal was accessible, in order to rebuild daily pricing. The intention is to make the whole collection accessible to academics and other scholars. Baltussen’s team developed both cap-weighted and equal-weighted index versions of its results, which seem to be consistent with past historical efforts to piece together returns from that timeframe.

Uses of the data

It is impossible to rebuild all of the variables that now intrigue academics. For example, accounting rules were not sufficiently reliable at the time. However, they had enough data on market capitalizations, which enabled them to examine the size effect on firms, and they also had adequate data on dividends. And, since they know how various companies' prices have changed over time, they can quantify the momentum effect by seeing which equities gained and lost during a certain period and determining whether the momentum carried over into the subsequent period. According to professor Baltussen, volatility was lower during this era, despite the fact that the United States lacked a central bank for the majority of the period.