Wages and Human Capital in the U.S. Finance Industry: 19092006 ONLINE APPENDIX

(1941) gives estimates of net income, wages and salaries, and the number of employees separately for banking, insurance, and real estate over the period 19191938. The banking category, however, covers only commercial banks, savings banks, and federal reserve banks; brokerage, investment banking, and other \u85nancial activities are not included. As a result, the size of the industry is smaller than that implied by BEA data. Fortunately, there is large overlap of 10 years with the BEA data over which the correlation between the two series is 96.6%. It seems therefore quite safe to impute values for the period 19191928 using Kuznetsdata. Martin (1939) provides data for \u85nance, insurance, and real estate, but not for \u85nance and insurance only. For the period 19091929 the estimates are based on data collected from banking, insurance, and real estate. For the period 18991908, however, the 1909 estimate is projected to 1899 on the basis of other data indicating a probable trend for this period.We \u85nd this procedure questionable, and therefore truncate our sample in 1909. We collected additional data from Mitchell (1921) for the banking sector in 19091919. The implied banking wage calculated from Mitchell (1921) is quite similar to the implied wage from Martin (1939) and the census data to measure the number of employees, except that it grows slightly faster. As mentioned, the data from Martin (1939) includes real estate. This does not appear to be a problem for the long-run trends. Using BEA data for the period 19292005, we \u85nd a correlation of 0.993 between the relative wag