War and pestilence as labor market shocks: U.S. manufacturing wage growth 1914-1919.

Garrett, Thomas A.

there is not a single expert machinist in the labor market today. If such a one exists he will be able to command his own wages...

The Arkansas Gazette (October 22, 1918).1

I. INTRODUCTION

The possibility of a worldwide influenza pandemic in the near future is of growing concern for many countries around the globe. The World Bank estimates that a global influenza pandemic would cost the world economy $800 kill tens of millions of people. (2) Researchers at the U.S. Centers for Disease Control and Prevention calculate that deaths in the United States could reach 207,000 and the initial cost to the economy could approach $166 billion, or roughly 1.5% of GDP. (3) The U.S. Department of Health and Human Services paints a more dire picture--up to 1.9 million dead in the United States and initial economic costs near $200 billion. (4)

While researchers and public officials can only speculate on the likelihood of a global influenza pandemic, many of the worst-case scenario predictions for a current pandemic are based on the global influenza pandemic of 1918. This global pandemic (often termed the "Spanish Flu") killed 675,000 people in the United States (nearly 0.8% of the 1910 population) and nearly 40 million people worldwide. (5) Only the Black Death of 13481351 is estimated to have killed more people (roughly 60 million) over a similar time period. (6) The global magnitude and spread of the 1918 influenza pandemic was exacerbated by World War I, which itself is estimated to have killed roughly 10 million civilians and 9 million troops (Nicolson 1980). Not only did the mass movement of troops from around the world lead to the spread of the disease, tens of thousands of Allied and Central Power troops died as a result of the influenza rather than combat]

Although combat deaths in World War I did increase the mortality rates for participating countries, civilian mortality rates from the 1918 influenza were typically much higher. For the United States, estimates of combat-related troop mortalities are about one-tenth that of civilian mortalities from the 1918 influenza.

Given the magnitudes and the concurrence of both the 1918 influenza pandemic and World War I, one would expect volumes of research on the economic effects of each event. Although there does exist a significant literature on the economic consequences of World War I (Rockoff 2004), the scope of published research on the economic effects of the 1918 influenza pandemic is scant.8 Most research that has been done has focused on the health and economic outcomes of decedents of pandemic survivors (Almond 2006; Keyfits and Flieger 1968) and mortality differences across socioeconomic classes (Mamelund 2006; Noymer and Garenne 2000).

This paper contributes to the literature on the economic effects of the 1918 influenza pandemic and World War I by exploring the influence of mortalities from these concurrent events on the growth of manufacturing wages in U.S. states and cities over the period 1914-1919. (9) The general conceptual foundation for the paper is that, ceteris paribus, both of these events resulted in a large number of deaths, which constituted a significant negative shock to manufacturing labor supply, and thereby would have increased wages in the manufacturing sector immediately following both events. A similar conceptual framework has been used in studies that explored the economic effects of the Black Death (Campbell 1997; Karakacili 2004) and the effects of immigration on labor markets (Borjas 1987; Card 1990; Greenwood and McDowell 1986).

The analysis presented here is the first to separate the potential manufacturing labor market effects of World War I and the 1918 influenza. This is an interesting exercise, given that mortalities from the 1918 influenza were nearly ten times greater than World War I combat mortalities, but World War I combat mortalities were more likely men aged roughly 18-44, prime ages for manufacturing workers. The results reveal that states and cities having had higher influenza and combat mortalities experienced greater growth in manufacturing wages over the period 1914-1919, and influenza mortalities had a greater overall effect on manufacturing wage growth from 1914 to 19l9 than did World War I combat mortalities.

The paper is organized as follows: The next section presents state-level data on mortalities from the influenza pandemic and World War I and discusses differences in mortality rates across subpopulations. The third section of the paper discusses in more detail the conceptual framework for the main hypothesis. The data and empirical methodology are discussed in the fourth section. The fifth section presents the empirical results from both state-level and city-level analyses and highlights the economic significance of the regression estimates. The final section of the paper concludes.

II. WORLD WAR I AND THE PANDEMIC ACROSS THE STATES

The influenza pandemic of 1918 began in early 1918 and lasted throughout the spring of 1919, with the vast majority of deaths occurring during the fall of 1918. (10) Unlike a typical influenza that predominately kills the youngest and oldest members of a population, Brainerd and Siegler (2003) report that the 1918 pandemic disproportionately killed people aged 15-44, with over 1% of the male population in the United States aged 25-34 dying from the disease. Data from the U.S. Bureau of the Census (1922) reveal that general population mortality rates in Maryland, Pennsylvania, and Colorado were the highest in the nation, with over 1% of the states' populations killed by the pandemic. During most influenza seasons male and female mortality rates are similar; however, Brainerd and Siegler (2003) report that during the 1918 influenza male mortality rate was at least 50% higher than the female mortality rate.

Evidence also suggests that influenza mortality rates had no relationship with state economic conditions, climate, or geography (Brainerd and Siegler, 2003; Crosby, 2003). After providing a survey of anecdotal evidence and conducting statistical analyses, Brainerd and Siegler (2003, p. 7) conclude that "The statistical evidence also supports the notion of influenza mortality as an exogenous shock to the population." Table 1 lists the states having had the five highest and five lowest influenza mortality rates in 1916 1917 (non-pandemic years) and 1918, along with the average mortality rate for all other states.

American troops were deployed to Europe starting in the early summer of 1918 and remained there until the signing of the Armistice on November 11, 1918. The United States deployed over 4 million troops, and approximately 3% were killed in action (corn oat or otherwise). Table 1 lists the states having had the five highest and five lowest combat mortality rates, as well as the average combat mortality rate for all other states. World War I combat mortality rates shown in Table 1 do not include death from diseases, such as the influenza pandemic. World War I combat mortality rates ranged from a low of about 19 per 100,000 population in Florida to over 161 per 100,000 population in Montana, with an average of about 52 per 100,000 population for all states.

III. THE CONCEPTUAL FRAMEWORK BEHIND THE HYPOTHESIS

The testable hypothesis of this paper is that influenza and World War I mortalities had a direct impact on wage rates in the manufacturing sector immediately following both events. A neoclassical analysis of the manufacturing labor market serves as the conceptual basis for this hypothesis. A decrease in the supply of manufacturing workers resulting from influenza and World War I mortalities would have had the initial effect of reducing manufacturing labor supply, increasing the marginal product of labor, and increasing real wages. Capital per worker would also have initially increased. In terms of the Solow (1956) growth model and the growth model of Romer (1986), this initial increase in capital per worker would have resulted in an increase in output per worker and an increase in wages. (11) In the short term, labor immobility across cities or states is likely to have prevented wage equalization, and a substitution away from relatively more expensive labor to capital is unlikely to have occurred. (12)

The mortality rates and subpopulations most affected by each event are suggestive of the likely effect on wages. Mortalities from both events were disproportionately males aged 18-44. It is this subpopulation that constituted the common demographic of a manufacturing sector employee. For example, using 1910 data on the male population between the ages of 18 and 44, the average combat mortality rate from World War I was about 202 per 100,000 men aged 18-44, or about four times the total population mortality rate. The 1918 influenza pandemic also disproportionately killed men aged 25-34 compared to typical cases of influenza. Since state-level mortality rates from the influenza pandemic were greater than the mortality rates from World War I, it seems reasonable that influenza mortality rates would have had a greater overall effect on manufacturing wage growth than World War I combat mortalities.

IV. DATA AND EMPIRICAL METHODOLOGY

To accurately assess the influence of influenza and World War I combat mortalities on manufacturing wage growth, it is important when building the empirical model to consider others factors that are likely to have influenced the manufacturing labor market, and thus wage changes, over the period 1914-1919. For example, wartime production is likely to have increased wage rates as a result of an increase in labor demand. Also, general growth in manufacturing, and thus changes in labor supply and labor demand, from 1914 prior to World War I and the influenza pandemic, should be considered. Finally, the capital stock available to manufacturing workers and the productivity of manufacturing labor are likely to have had effects on manufacturing wage growth. Descriptive statistics for all variables used in the analysis, which are discussed in detail below, are shown in Table 2.

A. Data

The dependent variable in the empirical model is the percentage change in real manufacturing wages per worker from 1914 to 1919. Data on the number of manufacturing wage earners and the total manufacturing wage bill for 1914 and 1919 were obtained from the 1919 U.S. Census of Manufactures (U.S. Bureau of the Census 1923). The 1919 estimates for the annual average wage per worker have been converted to 1914 prices using the national Consumer Price Index (CPI).13

Influenza Mortalities 1918-1919. State-level influenza mortality rates for 1918 and 1918 were obtained from Mortality Statistics 1920 from the U.S. Bureau of the Census (1922, 30). These data were then used to compute a single influenza mortality rate for 1918-1919 for each state. (14) The mortality rates used in this study (and in that of Brainerd and Siegler 2003) consider deaths from both influenza and pneumonia because "it is not believed to be best to study separately influenza and the various forms of pneumonia.... for doubtless many cases were returned as influenza when the deaths were caused by pneumonia and vice versa." (15)

Although the census compilation is the most cited source for reliable influenza pandemic mortality estimates, the data do have several limitations, many of which can be remedied. First, only 30 states reported influenza and pneumonia deaths for 1918 and 1919. (16) Second, influenza and pneumonia were common diseases prior to the 1918 pandemic, so an arguably better measure of how the pandemic affected mortality rates would be to compare mortality rates in non-pandemic years with the mortality rates in 1918 and 1919. One could then calculate how many more deaths than usual were from influenza and pneumonia in 1918 and 1919 and use this as an explanatory variable. Unfortunately, although Mortality Statistics contains annual data for a 10-yr period, these data are incomplete for many states for many years. For the 30 states that do report influenza and pneumonia deaths for 1918 and 1919, a portion of these states do not report mortality rates for other years, thus making a comparison between pandemic and non-pandemic years impossible. As a result, the state-level influenza and pneumonia mortality rate variable will be capturing "natural" influenza and pneumonia mortalities that occurred in 1918 and 1919 as well as the much higher number of deaths caused by the pandemic. Table 1 allows a rough comparison of the "natural" influenza and pneumonia mortality rate (1916-1917) and the rates from the 1918 pandemic. Mortality rates from the 1918 pandemic were roughly 2.5 times higher than in the non-pandemic years of 1916 and 1917.

A third issue is that the mortality death rates are based on annual state-level population values that were computed using linear interpolation between the 1910 and 1920 decennial census values. (17) By construct, this methodology imposes a constant growth rate in each state's population between 1910 and 1920. To correct for this fact, the mortality rates use 1914 state population estimates based on U.S. Bureau of the Census (1956) revisions of earlier estimates of state population. (18,19)

One final issue is the matching of influenza mortality to manufacturing employment. The mortality data used here are for the general population rather than influenza mortalities for manufacturing workers since data for the latter do not exist, so admittedly the link between influenza mortalities and manufacturing wage growth is imperfect. However, the majority of influenza mortalities were young to middle-aged men, which was a prime age for a manufacturing worker. Crosby (2003, pp. 319-22) reports that out of 272,500 male influenza deaths in 1918, nearly 49% were aged 20-39, whereas only 18% were under age 5 and 13% were over age 50. Two different sets of mortality rates are employed for the state-level analysis to best link between influenza mortalities and manufacturing employment. The two mortality rates are (1) total population mortality rates (men and women of all ages), and (2) mortality rates for men and women aged 2049.

Worm War I Combat Mortalities. World War I servicemen mortality data are from the Adjutant General of the Army 1919 (1920). These data are quite detailed, listing the number of servicemen that died from various causes, such as combat, accident, drowning, suicide, or homicide. Because the 1918 influenza pandemic affected troops during the war, the World War I combat mortality figures used in the empirical analysis are total serviceman mortalities minus the number of serviceman mortalities from disease. (20) This is done to avoid double counting because soldiers who died from influenza are included in the mortality figures discussed earlier. Like influenza mortalities, the number of serviceman combat mortalities from each state is normalized by 1914 state population.

One issue is the labor supply effect from the deployment of servicemen throughout 1918 to fight in World War I. (21) The absence of these troops from the manufacturing labor market would have immediately reduced the supply of labor and increased wages in that sector. (22) However, a great majority of deployed U.S. troops returned home after the signing of the Armistice in November of 1918. Under the assumption that most of the servicemen who returned home from the war in late 1918 were re-employed (especially in a booming economy), then the labor supply shock lasting throughout 1919 would have been from combat mortalities. The fact that the 1919 U.S. Census of Manufactures was taken in 1920 and covered the 1919 calendar year suggests a reasonable time period for the re-employment of returning troops and a reduction in labor supply resulting solely from combat mortalities.

Labor Market Control Variables. Because labor productivity influences wage rates, it is important to control for this influence in the empirical models. Capital per worker and value added per worker are included in the models to capture potential productivity differences across states. To avoid possible endogeneity problems, the 1914 value of both variables are included in the model rather than the percentage change in each variable from 1914 to 1919. Capital per worker in 1914 is computed using 1914 data on capital and the number of wage earners from the Census of Manufactures 1919. Similarly, value added in manufacturing in 1914 is divided by the number of wage earners in 1914 to get valued added per worker.

The effect of these two variables on the percentage change in wages is unclear. Theory suggests that states having had greater value added in manufacturing and greater manufacturing capital per worker were likely to have a higher level of wages. In fact, there is a high positive correlation between 1914 annual wages per manufacturing worker and 1914 value added per worker ([rho] = .81) and 1914 capital per worker ([rho] = .69). In terms of each variable's effect on the percentage change in wages, however, a story of convergence from Solow's (1956) growth model would suggest that a state having had a lower initial level of productivity in 1914 would have had a greater percentage increase in productivity through 1919 and thus a higher rate of wage growth. Convergence would thus suggest a negative relationship between the two productivity variables and the percentage change in real wages per worker--higher levels of initial productivity (i.e., value added per worker and capital per worker) resulted in lower rates of wage growth. As support for the possibility of convergence in manufacturing wages, Barro and Sala-i-Martin (2004) provide evidence on the convergence of state personal income over decennial periods beginning in 1920 and 20-year intervals prior to 1920. (23)

The United States did not officially enter World War I until President Wilson asked for and received a declaration of war on Germany from the U.S. Congress in April 1917. Prior to this time, however, the United States did supply its European allies fighting the war with aircraft and weaponry. Shipbuilding and aircraft production were the largest components of wartime production (Venzon 1999). The New York Shipbuilding Company located in Camden, New Jersey, was the predominant producer of World War I naval battleships and aircraft carriers. (24) The largest contractors for military aircraft were located in Ohio, New York, and Michigan. (25) It is reasonable to suspect that wartime production resulted in an increase in labor demand and thus wages in these states. To control for wartime production in the empirical models, a dummy variable is included that has the value of '1' if the state was involved in wartime production (New York, Ohio, Michigan, and New Jersey) and a value of '0' otherwise. It is expected that the coefficient on wartime production is positive, reflecting the increase in labor demand and thus wage growth in war production states. (26)

The general growth of the manufacturing sector between 1914 and 1919 is another potential influence on manufacturing wage growth. A variable designed to capture the growth in a state's manufacturing sector is included in the models. Specifically, the variable is the percentage change in a state's manufacturing wage earner workforce between 1914 and 1919. This variable reflects the growth in a state's manufacturing sector from increases in labor demand and/or labor supply caused by factors such as population growth, immigration, labor force demographic change, technological change, etc. (27) If the percentage increase in the number of workers was predominately a result of increases in labor demand (supply), it is then expected that wage growth would have been higher (lower). (28)

B. Empirical Methodology

To assess the influence of mortalities from the 1918 influenza and World War I on initial wage growth in the manufacturing sector, the percentage change in real manufacturing wages per worker from 1914 to 1919 is regressed on 1918-1919 influenza mortalities per capita, World War I mortalities per capita, value added per worker in 1914, capital per worker in 1914, the wartime production dummy variable, and the growth in manufacturing sector variable. In addition to the variables listed above, a set of three regional dummy variables based on U.S. Census regions are also included to capture unobserved differences in manufacturing wage growth, such as union activity. (29) Several regressions are run including or excluding variables as evidence for robustness across specifications. (30)

V. EMPIRICAL RESULTS

The state-level results from eight different regression specifications are shown in Table 3. Influenza mortalities and World War I combat mortalities are included individually and together. The aggregate effect of mortalities from both events is also considered in specifications (7) and (8). Because of a relatively high level of correlation between value added per worker and capital per worker ([rho] = .76), regressions were estimated with and without both variables included as independent variables. To complement the empirical results in Table 3, evidence on the economic significance of each variable is shown in Table 4. The impact on wage growth from a 10% increase from the mean of each continuous independent variable is presented. The effect of each binary variable on manufacturing wage growth is also presented in Table 4.

The estimates shown in Table 3 reveal that most of the coefficients on the labor market control variables are statistically significant. The states involved in wartime production experienced a greater growth in manufacturing wages compared to non-production states (about 10 percentage points, on average, across specifications). Also, states with manufacturing sectors that grew at a faster rate over 1914-1919 experienced greater manufacturing wage growth. Specifically, the data in Table 4 reveal that growth in the manufacturing sector increased real manufacturing wage growth by 0.70 percentage points, on average. This suggests that overall increases in manufacturing labor demand were greater than overall increases in manufacturing labor supply. States with higher value added per worker in 1914 experienced a slower rate of real wage growth from 1914 to 1919--a 10% increase from the mean value is shown to have decreased manufacturing wage growth by 2.90 percentage points, on average. This finding supports the concept of convergence derived from the Solow (1956) growth model. Finally, the coefficients on the regional dummy variables reveal that the average growth in manufacturing wages was higher, on average, in southern and mid-western states (about 20 percentage points and 9 percentage points, respectively, on average) compared to states located in the west and northeast.

Turning to the variables of interest--influenza pandemic and World War I combat mortalities the coefficient estimates on influenza and World War I combat mortalities reported in Table 3 are positive and significant, thus supporting the hypothesis that the reduction in manufacturing labor supply resulted in an initial increase in real wage growth in the manufacturing sector. (31) The coefficients on aggregated mortalities (specifications (7) and (8)) are also positive and statistically significant. The coefficient estimates on influenza and World War I combat mortalities are also relatively stable across specifications.

The economic significance of the effects of influenza and World War I combat mortalities on manufacturing wage growth are shown in Table 4. Averaging across the predicted changes shown in Table 4, mortalities from the influenza pandemic and World War I resulted in a 1.93 and a 0.91 percentage point increase in real manufacturing wage growth, respectively. (32) The aggregate effect of influenza pandemic and World War I mortalities, shown in the fourth data column of Table 4, reveals a 2.49 percentage point increase in real manufacturing wage growth.

A. Influenza Mortalities Aged 20-49

Empirical models are now estimated that consider influenza deaths of those aged 20 49 to better link influenza mortalities and manufacturing employment (and thus wage growth). (33) The regression results are shown in Table 5. The first specification reveals that influenza deaths of those aged 20-49 had a positive and significant effect on manufacturing wage growth. The coefficient on influenza deaths is not significant in the second specification that also considers World War I mortalities, however. This is a likely result of the relatively high correlation between World War I mortalities and age 20-49 influenza mortalities ([rho] = .60), which is supported by the large difference between the influenza mortality coefficients reported in specification (1) and (2). The total effect of World War I mortalities and age 20-49 influenza mortalities is positive and significant as seen in the third specification in Table 5, again suggesting that states having had more World War I and influenza mortalities experienced greater manufacturing wage growth.

The economic significance of World War I mortalities and age 20-49 influenza mortalities is demonstrated in Table 6 that considers the impact on wage growth from a 10% increase from the mean of the World War I mortality variable and the mean of the age 20-49 influenza mortality variable. Age 20-49 mortalities from the influenza pandemic and World War I mortalities resulted in a 1.21 and a 0.92 percentage point increase in real manufacturing wage growth, respectively. The aggregate effect of age 20-49 influenza mortalities and World War I mortalities is a 1.41 percentage point increase in real manufacturing wage growth.

B. City Influenza Mortalities and Wage Growth

In order to test the robustness of the state-level results, a similar analysis using city-level data is also conducted. The city-level analysis is absent any information about World War I mortalities since combat deaths by city are not available. All city labor market variables were obtained from the 1914 and 1919 U.S. Census of Manufactures and are identical to those used in the state-level analysis. City-level mortality data are obtained from a U.S. Public Health Service (1930) report on influenza mortality rates in 50 cities in the United States. This report provides monthly data on excess mortalities from the influenza computed on an annual basis. (34) For the purposes here, monthly city mortality rates were summed over the period March 1918 to April 1919 and then divided by 12 to get a city influenza death rate (per 100,000 population) during the pandemic. Excess influenza mortality rates by city during the pandemic are shown in Table 7.

The percentage change in city manufacturing wages per worker over the period 1914-1919 is regressed on excess influenza mortalities per capita (as described earlier), value added per worker in 1914, capital per worker in 1914, the wartime production dummy variable, and the growth in manufacturing sector variable. State dummy variables are included to capture any unobserved heterogeneity. The regression results are shown in Table 8.

The city-level regression results are somewhat different than the state-level results presented earlier. Although the coefficients on capital per worker and value added per worker (both highly correlated) retain their negative signs, the coefficients are not statistically significant. War production remains positive and significant, and its effect on wages is higher than in the state-level regressions--manufacturing wage growth was about 23 percentage points higher in war production cities (based on specifications 3 and 4). Also, the coefficient on the growth in manufacturing variable is positive and significant. The lower adjusted [R.sup.2] and a fewer number of significant coefficients compared to the state-level models suggests there is much more unexplained variation in wage growth across U.S. cities than across states.

Focusing on influenza mortalities, the regression results reveal that cities having had higher excess influenza death rates experienced greater growth in manufacturing wages over the period 1914 to 1919. For each city used in the analysis (Table 7), the percentage point increase in wages resulting from a 10% increase in the influenza mortality rate was calculated using the coefficient on influenza mortalities from the third specification of Table 8. These estimates are shown in Table 9 along with the actual increase in city manufacturing wages from 1914 to 1919 to get an idea of the impact of influenza mortalities on wage growth relative to overall wage growth. On average, a 10% increase in city-level mortality rates resulted in a 3.1 percentage point increase in city manufacturing wages (which had an average growth rate of 84.2%). Most manufacturing activity was located in urban areas and influenza mortality rates were higher in urban areas as well, so it seems reasonable that influenza mortalities in a city would have had a greater impact on city manufacturing wages compared to more rural areas (which are included in the state-level analysis).

VI. CONCLUSION

The concurrence of the 1918 influenza pandemic and World War I resulted in a significant number of mortalities over a brief period in history. Mortalities from both events were disproportionately higher for males of prime working age, thus making it likely that the manufacturing labor market was influenced by mortalities from the influenza pandemic and World War I. The hypothesis of this paper was that mortalities from both events resulted in a decrease in the supply of manufacturing labor, which initially increased both the marginal product of labor and capital per worker and thus resulted in an immediate increase in wage rates. The empirical results obtained from a sample of U.S. states and cities support the theoretical predictions. Specifically, states and cities that had more influenza mortalities and states having had more World War I combat mortalities experienced greater growth in manufacturing wages over the period 1914-1919. The total effect of influenza mortalities on wage growth was greater than that from World War I combat mortalities.

Certainly the 1918 influenza pandemic had serious negative effects on local economies. While specific estimates are prohibited by the lack of detailed economic data of the era, the deaths of primary breadwinners left many families without sources of income, commerce was hindered as a result of quarantines and mortalities, and a general fear by the public seriously hampered economic activity (Crosby 2003). The findings here suggest that one benefit of the influenza pandemic was an immediate increase in wages, at least in the manufacturing sector, for those surviving the pandemic. Of course, no reasonable argument can be made that this benefit outweighed the costs from the tremendous loss of life and overall economic activity. Nonetheless, the results obtained here suggest that some labor markets today, especially those for hourly, lower-skilled employees without contracts, may provide workers with higher wages during and in the immediate aftermath of a modern-day pandemic.

doi: 10.1111/j.1465-7295.2008.00137.x

REFERENCES

Adjutant General of the Army. Summary of Casualties among Members of the American Expeditionary Forces during the Worm War. Government Printing Office, Washington, D.C, 1920.

Allen, S. "Changes in the Cyclical Sensitivity of Wages in the United States, 1891-1987." American Economic Review, 82, 1992, 122-40.

Almond, D. "Is the 1918 Influenza Pandemic Over? Long-Term Effects of In Utero Influenza Exposures in the Post-1940 U.S. Population." Journal of Political Economy, 114, 2006, 672-712.

Ayres, L. The War With Germany." A Statistical Summary. Washington D.C.: Government Printing Office, 1919.

Barro, R., and X. Sala-i-Martin. Economic Growth. Cambridge, MA: The MIT Press, 2004.

Bloom, D., and A. Mahal. "AIDS, Flu, and the Black Death: Impacts on Economic Growth and Well-Being," in The Economics of HIV and AIDS: The Case of South and Southeast Asia, edited by D.

Bloom and P. Godwin. Delhi: Oxford University Press, 1997, 22-52.

Borjas, G. "Immigrants, Minorities, and Labor Market Competition." Industrial and Labor Relations Review, 40, 1987, 382-92.

Brahmbhatt, M. Avian Influenza. Economic and Social Impacts. Washington, D.C.: The World Bank, 2005.

Brainerd, E., and M. Siegler "The Economic Effect of the 1918 Influenza Epidemic." Discussion Paper 3791, Centre for Economic Policy Research, 2003.

Campbell, B. "Matching Supply to Demand: Crop Production and Disposal by English Demesnes in the Century of the Black Death." Journal of Economic History, 57, 1997, 827-58.

Card, D. "The Impact of the Mariel Boatlift on the Miami Labor Market." Industrial and Labor Relations Review, 43, 1990, 245-57.

Crosby, A. America's Forgotten Pandemic. The Influenza of 1918. Cambridge: Cambridge University Press, 2003.

Gordon, R. "A Century of Evidence on Wage and Price Stickiness in the United States, the United Kingdom, and Japan," in Macroeconomics, Prices, and Quantities, edited by J. Tobin. Washington, D.C.: Brookings Institution, 1983, 85-121.

Greenwood, M., and J. McDowell. "The Factor Market Consequences of U.S. Immigration." Journal of Economic Literature, 24, 1986, 1738-72.

Karakacili, E. "English Agrarian Labor Productivity Rates before the Black Death: A Case Study." Journal of Economic History, 64, 2004, 24-60.

Keyfits, N., and W. Flieger. World Population: An Analysis of Vital Data. Chicago: University of Chicago Press, 1968.

Maloney, T. "African American Migration to the North: New Evidence for the 1910s." Economic Inquiry, 40, 2002, 1-11.

Mamelund, S. "A Socially Neutral Disease? Individual Social Class, Household Wealth and Mortality from Spanish Influenza in Two Socially Contrasting Parishes in Kristiania 1918-19." Social Science and Medicine, 62, 2006, 923-40.

Meltzer, M., N. Cox, and K. Fukunda. "The Economic Impact of Pandemic Influenza in the United States: Priorities for Intervention." Emerging Infectious Diseases, 5(5), 1999, 659-71.

Minns, C. "Income, Cohort Effects, and Occupational Mobility: A New Look at Immigration to the United States at the Turn of the Century." Explorations in Economic History, 37, 2000, 326-50.

Mitchener, K., and I. McLean. "U.S. Regional Growth and Convergence, 1880-1980." Journal of Economic History, 59, 1999, 1016-42.

--. "The Productivity of U.S. States since 1880." Journal of Economic Growth, 8, 2003, 73-114.

Nicolson, C. The Longman Companion to the First Worm War. New York: Longman, 1980.

Noymer, A., and M. Garenne. "The 1918 Influenza Epidemic's Effects on Sex Differentials in Mortality in the United States." Population and Development Review, 26, 2000, 565-81.

Potter, C. "A History of Influenza." Journal of Applied Microbiology, 91, 2001, 572-79.

Rockoff, H. "Until It's Over, Over There: The U.S. Economy in World War I." Working Paper No. 10580, National Bureau of Economic Research, Cambridge, MA, 2004.

Romer, P. "Increasing Returns and Long Run Growth." Journal of Political Economy, 94, 1986, 1002-37.

Schultze, C. "The Cyclical Flexibility of Wages." American Economic Review, 76, 1986, 1152-3.

Solow, R. "A Contribution to the Theory of Economic Growth." Quarterly Journal of Economics, 70, 1956, 65-94.

Sundstrom, W. "Rigid Wages or Small Equilibrium Adjustments? Evidence from the Contraction of 1893." Explorations in Economic History, 29, 1992, 430-55.

U.S. Bureau of the Census. Mortality Statistics 1919. Washington D.C.: Government Printing Office, 1921.

U.S. Bureau of the Census. Mortality Statistics 1920. Washington D.C.: Government Printing Office, 1922.

U.S. Bureau of the Census. 14th Census of the United States 1920. Volume 9. Manufactures 1919. Washington, D.C.: Government Printing Office, 1923.

U.S. Bureau of the Census. Current Population Reports Series P25-139. Washington, D.C.: Government Printing Office, 1956.

U.S. Department of Health and Human Services. HHS Pandemic Influenza Plan. Washington, D.C.: U.S. Department of Health and Human Services, 2005.

U.S. Public Health Service. Mortality From Influenza and Pneumonia in 50 Large Cities of the United States 1910-1929. Washington, D.C.: Government Printing Office, 1930.

Venzon, A. The United States in the First Worm War. An Encyclopedia. Oxford, UK: Routledge, 1999.

(1.) "Solving Problem of Labor with Dilution," The Arkansas Gazette, Little Rock, Arkansas, October 22, 1918, p. 6.

(2.) Brahmbhatt (2005).

(3.) Meltzer, Cox, and Fukunda (1999).

(4.) U.S. Department of Health and Human Services (2005).

(5.) See Potter (2001). Although 40 million is the commonly accepted number of worldwide deaths from the pandemic, it is likely an underestimate given the lack of adequate recordkeeping in many parts of the world. The 1918 influenza pandemic was termed the "Spanish Flu" by the Allies of World War I since Spain had one of the worst early outbreaks of the disease, with nearly 8 million people infected by early 1918.

(6.) Bloom and Mahal (1997).

(7.) Ayres (1919).

(8.) In their unpublished manuscript, Brainerd and Siegler (2003) conduct an analysis of the economic effects of the 1918 influenza in the decade following the pandemic.

(9.) The manufacturing sector was chosen over other major employment sectors, such as agriculture, for several reasons. First, the Census of Agriculture prior to 1977 was conducted on a decennial basis (with the U.S. Census), thus only providing data for 1910 and 1920. It was thought that 1910 was too prior to the concurrence of the influenza pandemic and World War I to get reliable estimates of the influence of each event on agricultural wage growth. The coverage dates of the Census of Manufactures (1914 and 1919) are more proximal to both events. Second, the U.S. Census of Agriculture only reported wage data for farm labor, which is only one of many classifications of agricultural employment. Third, many of the labor market control variables, which are discussed later, such as value added and capital, are not available or, because of the limitations on the agriculture employment data, would provide misleading values for capital per worker and value added per worker.

(10.) See Crosby (2003) for a detailed discussion of the 1918-1919 influenza pandemic in the United States.

(11.) This framework assumes some degree of labor market flexibility during the late 1910s. There is no evidence that real or nominal wages were completely rigid during this period, although there is debate in the literature over wage flexibility (in response to the business cycle) at the beginning of the twentieth century compared to the end of twentieth century. See Gordon (1983), Schultze (1986), Allen (1992), and Sundstrom (1992).

(12.) The empirical models discussed later will capture the short run effect of labor mobility. The long-run effect of influenza and war mortalities on manufacturing wage growth is less clear. The Solow (1956) growth model suggests that capital per worker will eventually fall (due to diminishing returns to capital) and therefore decrease wages, whereas Romer's (1986) growth model predicts capital per worker will continue to rise over time as a result of non-diminishing returns to capital, thereby increasing wages. It is also possible that the war and the pandemic decreased consumer confidence, investment and savings, and long-term income growth of households due to the death of households' primary breadwinners. These factors would result in lower aggregate output and production, thereby decreasing the demand for labor and placing downward pressure on manufacturing wages.

(13.) The only effect of using real 1919 wages rather than nominal 1919 wages is that, with real 1919 wages, each coefficient estimate will differ from its value if nominal wages were used by a factor of [theta], where [theta] is equal to the ratio of the 1914 and 1919 national CPI.

(14.) The influenza mortality rate for each state was computed as follows: the 1918 and 1919 influenza mortality rates from the U.S. Bureau of the Census (1922) were first multiplied by state population to arrive

at the number of influenza deaths in 1918 and 1919. The number of influenza deaths for these 2 years was then summed for each state and then normalized by state population.

(15.) U.S. Bureau of the Census (1921, 28).

(16.) The states excluded from the analysis are Alabama, Arkansas, Arizona, District of Columbia, Delaware, Florida, Georgia, Iowa, Idaho, Mississippi, North Dakota, Nebraska, New Mexico, Nevada, Oklahoma, South Dakota, Texas, West Virginia, and Wyoming.

(17.) See U.S. Bureau of the Census (1922, p. 9).

(18.) The mortality rates were recomputed as follows: First, the linearly interpolated population estimates were computed using the 1910 and 1920 decennial population estimates. The mortality rates for each of the 30 states in the sample were then each multiplied by the respective interpolated population value to get the number of deaths from influenza and pneumonia. The number of deaths was then divided by the revised U.S. Census (1956) estimates of 1914 population. The U.S. Census (1956) used statistical sampling rather than linear interpolation to reestimate state population estimates for earlier years.

(19.) State population in 1914 is used to normalize influenza and World War I mortalities because many of the labor market variables discussed later are normalized by the number of manufacturing workers in 1914. Normalizing influenza and World War I mortalities by state population in other years did not significantly change the empirical results. Differences in the mortality rates published in Mortality Statistics and the reestimated mortality rates using the revised population estimates for 1914 were less than 1%.

(20.) Ideally, one would like to have data on the number of combat troops that worked in manufacturing prior to World War I in order to better link World War I combat mortalities to manufacturing wage growth.

(21.) Contacts at the various branches of the armed services and the National Archives report that no comprehensive, official list of World War I servicemen by state exists. The population survey for the 1930 U.S. Census did ask the question of veteran status, but contacts at the Census Bureau report that this information would likely be a misleading indicator of the number of servicemen during World War I due to mobility across states and the deaths of servicemen that occurred after the war but prior to the 1930 Census.

(22.) The labor supply effect would have been less if servicemen were enlisted before the war began.

(23.) Also see Mitchener and McLean (1999; 2003).

(24.) See www.history.navy.mil.

(25.) See the U.S. Centennial of Flight Commission at www.centennialofflight.gov/essay/Aerospace/WWi/ Aero5.htm.

(26.) Certainly numerous states in addition to those listed here were involved in some form of wartime production. One cannot build ships and aircraft without steel and one cannot make steel without iron ore and coal, etc. The coefficient on the wartime dummy variable can thus be interpreted as wartime production of final goods.

(27.) The number of manufacturing workers increased from 1914 to 1919 for all states in the sample.

(28.) Hausman tests reveal that the percentage change in manufacturing workers is exogenous to the percentage change in manufacturing wages. State population in 1914 and the percentage of children aged 5-18 that were enrolled in school in 1912 were used as instruments along with all exogenous variables listed in Table 2. The average p value on the Hausman test statistics (t statistic) was .71. The finding that the percentage change in manufacturing workers is exogenous is not too surprising given the time span used in this study, that is, it is unlikely that a 5-year period is long enough to capture the possible effect of higher wages inducing a substitution away from other sectors of the economy to the manufacturing sector.

(29.) The dummy variable for the western region is omitted to avoid collinearity problems. The regional dummy variables may also capture the effects of African American emigration from the South as well as European emigration (Maloney 2002: Minns 2000).

(30.) State population density was considered in all empirical models m order to test whether higher density states experienced greater growth in the manufacturing sector. Also, a measure of educational attainment (the percentage of children aged 5-18 enrolled in public school in 1912) was considered. The coefficients were not statistically significant and the variables were thus omitted from the final regressions.

(31.) The marginal impact of an influenza death and a combat death on 1914-1919 wage growth was (statistically) the same. An F test on the equality of the influenza mortality and the World War I combat mortality coefficients in specification (6) of Table 3 reveals that the null hypothesis of coefficient equality cannot be rejected at conventional levels (F = 2.18).

(32.) Recall that the influenza mortality rate variable also contains a 'natural' rate of influenza mortality that would have occurred in the absence of a pandemic. As seen in Table 1, mortality rates from the influenza pandemic were about 2.5 times higher than the mortality rate in earlier non-pandemic years. The city-level analysis that follows uses data on influenza mortalities that are in excess of the normal number of influenza mortalities expected in a year.

(33.) This variable was computed by summing influenza deaths in each state for those aged 20-29, 30-39, and 40-49. In the sample of states used here, influenza mortalities of those aged 20-49 averaged 46.1% of all influenza mortalities. This variable has a standard deviation of 6.4%, a minimum value of 35%, and a maximum value of 63%. The variable was normalized by 1914 state population.

(34.) 'Excess' deaths refers to those deaths above (or below) some historical median rate. The monthly rates were converted to an annual number by multiplying by 12. See U.S. Public Health Service (1930, p. 27) for a specific discussion on how the rates were calculated.

THOMAS A. GARRETT, The author would like to thank David Wheelock and two anonymous referees for helpful comments and suggestions. Seminar participants at West Virginia University, Eastern Illinois University, and Washington State University also provided useful comments. Thanks also to Tim Syzek at the National Archives and Records Administration for obtaining the World War I mortality data used in this study. Lesli Ott provided valuable research assistance. All remaining errors are my own. The views expressed here are those of the author and not necessarily those of the Federal Reserve Bank of St. Louis or the Federal Reserve System.

Garrett: Assistant Vice President and Economist, Federal Reserve Bank of St. Louis, St. Louis, MO 63166-0442.

Phone 1-314-444-8601, Fax 1-314-444-8731, E-mail garrett@stls.frb.org

TABLE 1
Influenza and WWI Combat Mortality Rankings, By State

Influenza Mortalities Per
100,000 Population

Top 5 States 1916-1917

Connecticut 455.5
Pennsylvania 413.3
Vermont 410.2
Maryland 407.7
Rhode Island 407.7

Bottom 5 States 1916-1917

Utah 252.2
North Carolina 251.7
California 228.4
Minnesota 209.8
Washington 157.7
Average of other states 318.2

Top 5 States 1918-1919

Pennsylvania 1,119.6
Maryland 1,042
Colorado 1,020
New Jersey 995.9
Connecticut 992.2

Bottom 5 States 1918-1919

Washington 599.4
Wisconsin 584.1
Michigan 581.5
Minnesota 557.4
Oregon 523.6
Average of other states 802.7

WWI Combat Mortalities
Per 100,000 Population

Top 5 States

Montana 161.1
Wyoming 92.6
North Dakota 84.8
Connecticut 83.1
Idaho 77.1

Bottom 5 States

Georgia 26.5
Delaware 23.1
Louisiana 22.3
Mississippi 18.7
Florida 18.7
Average of other states 50.9

Notes: Influenza mortalities include death from pneumonia. See
text for explanation. Data are from U.S. Bureau of the Census
(1922, p. 30). Twenty-six states reported influenza and pneumonia
death rates in 1916 and 1917 and 30 states reported influenza and
pneumonia mortality rates in 1918 and 1919. Influenza pandemic
and World War I mortalities shown above are normalized by 1914
state population. Combat mortality data are from the Adjutant
General of the Army (1919). Combat mortalities refer to all
deaths in combat from any cause other than disease. See text for
a further description of this variable. WWI, World War I.

TABLE 2
Descriptive Statistics

 Standard
Variable Mean Deviation

Percentage change 11.49 12.64
in real manufacturing
wages per worker 1914-1919

1918-1919 Influenza 829.28 171.64
mortalities per 100,000 population

World War I combat 56.10 23.81
mortalities per 100,000 population

Influenza and WW1 mortalities 885.38 181.88
per 100,000 population

Capital per manufacturing 3,507.00 1,109.40
worker 1914 ($)

Value added 1,427.43 414.40
per manufacturing worker 1914 ($)

Wartime production 0.1333 0.3457

Growth in manufacturing 32.64 24.96
sector 1914-1919
(percentage change in workers)

Variable Minimum Maximum

Percentage change -15.85 38.05
in real manufacturing
wages per worker 1914-1919

1918-1919 Influenza 544.41 1,154.14
mortalities per 100,000 population

World War I combat 22.30 161.06
mortalities per 100,000 population

Influenza and WW1 mortalities 592.10 1,294.02
per 100,000 population

Capital per manufacturing 1,854.97 6,663.48
worker 1914 ($)

Value added 666.66 2,751.18
per manufacturing worker 1914 ($)
Wartime production 0 1

Growth in manufacturing 2.41 103.10
sector 1914-1919
(percentage change in workers)

Note: Number of observations = 30. All data for U.S. States.
See text for data sources and descriptions.

TABLE 3
Influenza and WWI Mortalities on Manufacturing Wage Growth 1914-1919

Variable (1) (2)

Constant 0.216 (1.51) 0.218 *** (2.87)
WWI mortalities 183.67 *** (3.21) 183.41 *** (3.45)
 per capita
Influenza mortalities -- --
 per capita
WWI and influenza -- --
 mortalities per capita
Capital per worker-1914 (a) 0.003 (0.02) --
Value added per -0.239 *** (5.08) -0.239 *** (5.54)
 worker-1914 (a)
War production dummy 0.114 *** (5.70) 0.114 *** (6.21)
Growth in manufacturing 0.156 ** (2.78) 0.156 *** (3.36)
 sector
Northeast dummy 0.030 (0.39) 0.030 (0.65)
South dummy 0.193 ** (2.23) 0.193 *** (3.50)
Midwest dummy 0.063 (1.33) 0.062 ** (2.25)
Adjusted [R.sup.2] 0.696 0.709

Variable (3) (4)

Constant 0.014 (0.08) -0.079 (0.53)
WWI mortalities -- --
 per capita
Influenza mortalities 31.39 ** (2.71) 27.30 ** (2.63)
 per capita
WWI and influenza -- --
 mortalities per capita
Capital per worker-1914 (a) -0.036 (1.54) --
Value added per -0.136 *** (5.53) -0.167 *** (5.37)
 worker-1914 (a)
War production dummy 0.077 ** (2.55) 0.100 *** (4.51)
Growth in manufacturing 0.234 *** (3.22) 0.236 *** (3.32)
 sector
Northeast dummy 0.022 (0.30) 0.088 * (1.85)
South dummy 0.174 ** (2.21) 0.239 *** (4.20)
Midwest dummy 0.099 * (1.95) 0.134 *** (3.01)
Adjusted [R.sup.2] 0.705 0.698

Variable (5) (6)

Constant 0.052 (0.30) 0.015 (0.09)
WWI mortalities 120.54 ** (2.25) 141.01 ** (2.56)
 per capita
Influenza mortalities 22.59 * (1.85) 19.22 * (1.75)
 per capita
WWI and influenza -- --
 mortalities per capita
Capital per worker-1914 (a) -0.019 (0.85) --
Value added per -0.201 *** (4.53) -0.227 *** (4.87)
 worker-1914 (a)
War production dummy 0.087 *** (2.92) 0.099 *** (4.37)
Growth in manufacturing 0.227 *** (3.33) 0.227 *** (3.41)
 sector
Northeast dummy 0.022 (0.33) 0.053 (1.11)
South dummy 0.193 ** (2.54) 0.226 *** (4.14)
Midwest dummy 0.100 * (2.06) 0.116 ** (2.60)
Adjusted [R.sup.2] 0.719 0.727

Variable (7) (8)

Constant 0.009 (0.06) -0.086 (0.64)
WWI mortalities -- --
 per capita
Influenza mortalities -- --
 per capita
WWI and influenza 30.872 *** (2.99) 28.109 *** (3.03)
 mortalities per capita
Capital per worker-1914 (a) -0.032 (1.47) --
Value added per -0.152 *** (6.99) -0.179 *** (6.16)
 worker-1914 (a)
War production dummy 0.077 ** (2.58) 0.097 *** (4.37)
Growth in manufacturing 0.239 *** (3.41) 0.244 *** (3.60)
 sector
Northeast dummy 0.021 (0.30) 0.083 * (1.87)
South dummy 0.180 ** (2.33) 0.241 *** (4.43)
Midwest dummy 0.103 ** (2.12) 0.137 *** (3.32)
Adjusted [R.sup.2] 0.717 0.712

Notes: *** denotes significance at 1%, ** at 5%,and * at 10'%. t
statistics are in parentheses and are based on White's
heteroscedasticity-corrected standard errors. The dependent variable
is the percentage change in state-level real manufacturing wages from
1914 to 1919. Number of observations = 30. See text for a complete
description of each variable. WW I, World War I.

(a) Coefficients multiplied by 1,000.

TABLE 4

Estimated Impact of Variables on Manufacturing Wage Growth 1914-1919

Variable Change

WWI mortalities Mean + 10%
 per capita
Influenza mortalities Mean + 10%
 per capita
WWI and influenza Mean + 10%
 mortalities per capita
Value added Mean + 10%
 per worker-1914
War production dummy 0 to 1
Growth in manufacturing sector Mean + 10%
Northeast dummy 0 to 1
South dummy 0 to 1
Midwest dummy 0 to 1

 Impact of Change
 on Wage Growth
 (Percentage Point Increase/Decrease)
 From Select Specifications in Table 3

Variable (2) (4) (6) (8)

WWI mortalities 1.03# -- 0.79# --
 per capita
Influenza mortalities -- 2.26# 1.59# --
 per capita
WWI and influenza -- -- -- 2.49#
 mortalities per capita
Value added -3.41# -2.38# -3.24# -2.56#
 per worker-1914
War production dummy 11.4# 10.00# 9.90# 9.70#
Growth in manufacturing sector 0.51# 0.77# 0.74# 0.80#
Northeast dummy 3.00 8.80# 5.30 8.30#
South dummy 19.30# 23.90# 22.60# 24.10#
Midwest dummy 6.20# 13.40# 11.60# 13.70#

Note: Coefficient estimates and mean values from specifications (2),
(4), (6), and (8) in Table 3 are used to estimate percentage point
increases-decreases. Bold-faced values reflect impacts based on
statistically significant coefficient estimates in Table 3. WWI, World
War I.

Note: Bold-faced values reflect impacts based on statistically
significant coefficient estimates in Table 3, indicated with #.

TABLE 5
Influenza Mortalities Aged 209 and Wage Growth 1914-1919

Variable (1) (2)

Constant 0.176 (1.03) 0.200 (1.25)
WWI mortalities per capita -- 163.73 ** (2.30)
Influenza mortalities aged 31.384 * (1.87) 8.910 (0.44)
 20-49 per capita
WWI and influenza mortalities -- --
 per capita
Capital per worker-1914 (a) -0.030 (1.24) -0.005 (0.21)
Value added per -0.154 *** (5.33) -0.232 *** (4.76)
 worker-1914 (a)
War production dummy 0.104 *** (4.47) 0.111 *** (5.12)
Growth in manufacturing sector 0.177 ** (2.44) 0.169 ** (2.61)
Northeast dummy 0.029 (0.35) 0.029 (0.38)
South dummy 0.185 * (1.95) 0.197 ** (2.28)
Midwest dummy 0.082 (1.35) 0.073 (1.30)
Adjusted [R.sup.2] 0.654 0.683

Variable (3)

Constant 0.165 (1.02)
WWI mortalities per capita --
Influenza mortalities aged --
 20-49 per capita
WWI and influenza mortalities 31.926 ** (2.28)
 per capita
Capital per worker-1914 (a) -0.027 (l.15)
Value added per -0.172 *** (5.52)
 worker-1914 (a)
War production dummy 0.103 *** (4.35)
Growth in manufacturing sector 0.185 ** (2.65)
Northeast dummy 0.028 (0.35)
South dummy 0.192 ** (2.08)
Midwest dummy 0.088 (1.53)
Adjusted [R.sup.2] 0.669

Notes. *** denotes significance at 1%, ** at 5%, and * at 10%.
t statistics are in parentheses and are based on White's
heteroscedasticity-corrected standard errors. The dependent variable
is the percentage change in state-level real manufacturing wages from
1914 to 1919. Number of observations = 30. See text for a complete
description of each variable. WWI, World War I.

(a) Coefficients multiplied by 1,000.

TABLE 6
Effect of Age 20-49 Mortalities on Wage Growth:
Impact of a Mean + 10% Change on Wage Growth

 Percentage Point
 Increase from Table 5
 specifications

Variable (1) (2) (3)

WWI mortalities -- 0.92# --
 per capita
Influenza mortalities 1.21# 0.34 --
 aged 20-49 per capita
WWI and influenza -- -- 1.41#
 mortalities per capita

Note: Coefficient estimates from specifications (l), (2), and (3) in
Table 5 are used to estimate change. Bold-faced value, reflect impacts
based on statistically significant coefficient estimates in Table 5.
WWI, World War I.

Note: Bold-faced value, reflect impacts based on statistically
significant coefficient estimates in Table 5 is indicated with #.

TABLE 7
(Excess) Influenza Pandemic Mortality Rates (per 100,000)
For 50 U.S. Cities

Albany 583.3
Atlanta 368.6
Baltimore 616.0
Birmingham 712.8
Boston 680.9
Bridgeport 748.0
Buffalo 539.2
Cambridge 503.1
Chicago 361.4
Cincinnati 552.6
Cleveland 583.3
Columbus 365.3
Dayton 429.3
Denver 688.6
Detroit 354.3
Fall River 641.1
Grand Rapids 226.1
Indianapolis 419.9
Jersey City 698.7
Kansas City, MO 694.8
Los Angeles 496.8
Louisville 637.3
Lowell 555.4
Memphis 580.8
Milwaukee 396.4
Minneapolis 336.8
Nashville 781.9
New Haven 567.2
New Orleans 678.3
New York 466.6
Newark 582.8
Oakland 531.5
Omaha 567.9
Paterson 609.7
Philadelphia 792.7
Pittsburgh 1104.7
Portland 538.9
Providence 625.0
Richmond 532.7
Rochester 406.3
San Francisco 656.9
Scranton 792.5
Seattle 433.8
Spokane 494.1
St. Louis 415.5
St. Paul 409.1
Syracuse 485.0
Toledo 354.6
Washington, D.C. 579.2
Worcester 617.7

Notes: U.S. Public Health Service (1930) provides monthly excess
mortality statistics on an annual basis. To compute the data shown
in the table, the monthly excess mortalities from U.S. Public Health
Service (1930) from March 1918 to April 1919 were summed and then
divided by 12. 'Excess' number of deaths refers to those deaths above
(or below) a historical median rate. See U.S. Public Health Service
(1930, p. 27) for more information.

TABLE 8
City Influenza Mortalities and Wage Growth, 1914-1919

Variable (1) (2)

Constant 1.187 *** (7.91) 0.636 *** (4.02)
Influenza mortalities per 6.843 (0.58) 37.688 * (1.77)
 capita
Capital per worker-1914 (a) -- --
Value added per -0.279 *** (3.31) --
 worker-1914 (a)
War production dummy 0.217 *** (2.88) 0.193 *** (3.16)
Growth in manufacturing sector 0.095 * (1.81) --
State dummy variables No Yes
Adjusted [R.sup.2] 0.273 0.209

Variable (3) (4)

Constant 0.702 *** (3.13) 0.703 *** (3.13)
Influenza mortalities per 56.407 ** (2.81) 56.844 ** (2.73)
 capita
Capital per worker-1914 (a) -- -0.004 (0.13)
Value added per -0.178 (1.61) -0.173 (1.37)
 worker-1914 (a)
War production dummy 0.232 *** (3.48) 0.233 *** (3.51)
Growth in manufacturing sector 0.233 *** (5.19) 0.232 *** (5.19)
State dummy variables Yes Yes
Adjusted [R.sup.2] 0.425 0.395

Notes: *** denotes significance at 1%, ** at 5%, and * at
10%. t statistics are in parentheses and are based on White's
heteroscedasticity-corrected standard errors. The dependent
variable is the percentage change in real city manufacturing
wages from 1914 to 1919. Number of observations = 50. The mean
value of influenza mortalities is 0.00556 (556 per 100,000).
See text for a complete description of each variable.

(a) Coefficients multiplied by 1,000.

TABLE 9
Total City Wage Growth Versus Wage Growth From
Influenza Deaths

 Percentage Percentage Point
 Change Increase in Wages
 in Wages From 10% Increase
City 1914-1919 in Flu Deaths

Albany 61.2 3.29
Atlanta 93.3 2.08
Baltimore 119.0 3.47
Birmingham 92.4 4.02
Boston 73.3 3.84
Bridgeport 114.5 4.22
Buffalo 97.1 3.04
Cambridge 66.1 2.84
Chicago 84.5 2.04
Cincinnati 77.9 3.12
Cleveland 105.4 3.29
Columbus 78.6 2.06
Dayton 117.9 2.42
Denver 62.0 3.88
Detroit 110.8 2.00
Fall River 88.2 3.62
Grand Rapids 88.1 1.28
Indianapolis 70.6 2.37
Jersey City 102.5 3.94
Kansas City, MO 65.9 3.92
Los Angeles 51.9 2.80
Louisville 76.5 3.60
Lowell 111.8 3.13
Memphis 50.0 3.28
Milwaukee 82.0 2.24
Minneapolis 62.0 1.90
Nashville 74.1 4.41
New Haven 76.0 3.20
New Orleans 86.7 3.83
New York 106.5 2.63
Newark 107.1 3.29
Oakland 74.7 3.00
Omaha 96.3 3.20
Paterson 98.0 3.44
Philadelphia 111.3 4.47
Pittsburgh 103.8 6.23
Portland 90.8 3.04
Providence 79.8 3.53
Richmond 86.7 3.00
Rochester 80.0 2.29
San Francisco 54.4 3.71
Scranton 72.2 4.47
Seattle 94.2 2.45
Spokane 39.7 2.79
St. Louis 67.3 2.34
St. Paul 64.4 2.31
Syracuse 80.0 2.74
Toledo 89.4 2.00
Washington, D.C. 84.0 3.27
Worcester 91.9 3.48

Note. The percentage point increase in computed by
increasing influenza death rates (Table 7) by 10%
and multiplying by the coefficient on influenza
deaths in Table 8, specification 3 (56.407).