Abstract
From the 1960s, Minsky argued that implementing a decentralized job-guarantee policy funded by the federal government was a relevant way to promote full employment and price stability, and to alleviate poverty. This policy aims at providing a job to anybody willing to work and to pay a living wage. Over the past fifteen years, this idea has been subject to greater scrutiny and this paper contributes to that literature by estimating the gross cost of implementing a job-guarantee policy (JG). In order to calculate this cost, the paper uses the data available from the 1930s work programs. These work programs provide some interesting insights because enough data are available to determine the cost of JG under widely different rates of unemployment. The paper shows that JG would have been quite expensive during the early part of the 1930s when the unemployment rate was at 20 percent or more. Once unemployment receded to a usual level, the gross cost of JG would have been low.
1
From the 1960s, Minsky (1965, 1966) argued that implementing an employer of last resort policy funded by the federal government was a relevant way to promote full employment and price stability, and to alleviate poverty. While this policy agenda did not gain much attention while Minsky was alive, it has been studied more carefully over the past fifteen years and has taken other names like job guarantee (JG) or buffer stock employment. Wray (1998) and Mitchell and Muysken (2008) have developed the theoretical framework behind JG, and Harvey (1989, 2000); Fullwiler (2007, 2013); Kaboub (2008, 2012, 2013); Tcherneva and Wray (2007); Gordon (1997); Watts and Mitchell (1997, 2000); Kitson, Michie, and Sutherland (1997); and Majewski (2004) have provided some empirical analysis of the cost, benefits, and socio-economic impacts of JG for several countries. There are some differences among authors in terms of the structure of JG and its implementation (wage paid, types of activities, mandatory or not), but the broad point is that the federal government should stand ready to employ anybody willing and able to work at the wage paid by JG. Estimates of the cost of JG vary between 1 and 5 percent of GDP depending on the assumptions, the country’s average unemployment level, the elements included in costs (labor and/or non-labor costs), and the measurement of cost in gross terms (i.e. total budget) or net terms (total budget less savings from unemployment insurance and other ills created by unemployment).
This paper contributes to the empirical side of JG by studying the 1930s work programs. We are fortunate to have several government reports and books that provide a wealth of information about the programs. While far from being a JG in the way envisaged by Minsky and his followers, the 1930s work programs are probably the closest policy experiment in the United States. The goal of this paper is to estimate the gross cost of JG by taking the data provided by these books and reports. This is done by studying what the cost of federal work programs would have been if all unemployed had been enrolled in JG, and if they had been paid a living wage. The paper shows that while the cost of JG would have been quite substantial during the early part of the 1930s when the unemployment rate was over 20 percent, the cost would have been quite low by the 1940s when the unemployment rate was at a usual level.
The first part of the paper provides an overview of the work programs, which helps to understand how similar or different they were from JG. The second part estimates the actual cost of the different programs by measuring expenses relative to GNP on a monthly basis. The third part of the paper provides estimates of the cost of JG on the basis of the New Deal programs. The fourth part discusses the limits of the estimates.
2. The 1930s Work Programs
The Great Depression that started in August 1929 continuously worsened until 1933; by that time the U.S. economy had reached rock bottom and social unrest was rapidly growing. One of the economic data that illustrates this state of affairs is the unemployment rate. Figure 1 shows that the unemployment rate reached 27 percent in March 1933 and did not show any sign of receding. Millions of people fell into poverty and lost their skills, health, and moral, which compounded existing problems by depressing further aggregate demand, by lowering productive capacities, and by decreasing the employability of workers.
Unemployment Rate: 1929-1943. Seasonally adjusted.
For the first two years of the recession, the response of the federal government was sparse. The dominant policy prescription of the time called for limited federal involvement and a balanced budget. In addition, the relief system prevailing in the United States was modeled on the English poor laws that considered poverty a local problem best managed by municipalities and private charities. The Great Depression overwhelmed this relief system and by 1931 states had to intervene. By 1932, states themselves were overwhelmed and the Hoover administration allowed the Reconstruction Finance Corporation (RFC) to provide some financial support to states for the purpose of relief.
Most of the relief system consisted of in-kind and monetary aid. Work relief had always been marginal in terms of scope and achievements, and had more to do with “work-test” activities that were supposed to show readiness to work and so justified assistance. They involved working on a woodpile, leaf raking, digging holes and filling them up, and other miscellaneous and mostly irrelevant activities. This minimalist approach to work relief started to change in the middle of 1932 when states and localities took the initiative to organize more meaningful job opportunities. By the first trimester of 1933, State Emergency Relief Administrations (SERAs) supervised the employment of 1.5 to 2 million individuals through General Relief work programs (GR) (Figure 2). From July 1933, the federal government helped SERAs fund their job initiatives through the Federal Emergency Relief Administration (FERA). As long as New Deal programs were nonexistent or too small, FERA usually allocated 30 to 40 percent of its relief funds to work relief (Federal Emergency Relief Administration 1942: Table 10). Still, the quality and meaningfulness of the work done was uneven because of the lack of organization and funding (Schwartz 1984: 32). In addition, individuals on work relief were treated in a way that made them feel inferior and a burden to society. They were subject to routine supervisions by social workers as well as monthly investigations of their budgetary qualifications for relief, and were paid through weekly handouts of food and subsistence checks.
Employment in the Federal and State Work Programs: January 1933-June 1943.
With Roosevelt’s New Deal, the federal government progressively took over the task of selecting, funding, and supervising work programs. States and localities mostly acted as sponsors by submitting projects for approval to the federal government, and by committing to cover some of the expenses (mostly administrative, supervisory, and non-labor costs). This framework allowed projects to be tailored to the needs of localities and states, while also giving some oversight to the federal government to make sure that its funds were spent in a relevant way and in compliance with federal labor laws.
The first New Deal program was the Civilian Conservation Corps (CCC) (formerly named the Emergency Conservation Work) that operated from March 1933 until July 1942. 1 The CCC employed youth (16 to 25 years old) mostly in environmental activities (reforestation, soil conservation, fire and flood prevention, and wildlife protection, among others). The National Youth Administration (NYA) operated from June 1935 to June 1943 and complemented the CCC by employing young adults in white-collar activities, manufacturing, and crafting. The NYA also allowed some young individuals to get a formal education.
The first attempt by the federal government to employ individuals of all ages seeking work was the Public Works Administration (PWA) (legally known as the Federal Emergency Administration of Public Works) that operated from June 1933 to June 1939. The PWA worked through private enterprises. It let them hire individuals but paid the wage bill. Because private companies only hired the most qualified unemployed and because the PWA insisted that the prevailing wage be paid to PWA workers, the PWA had a limited success in terms of enrollment. This led to a succession of programs to accommodate more unemployed individuals. The Civil Works Administration (CWA) had a short life from November 1933 to February 1934 because of political oppositions to its rapid expansion. From April 1934 to July 1935, the federal government transferred the responsibility of helping unemployed adults back to states and localities. The management of the CWA enrollees was transferred to the Emergency Work Relief program (EWR) funded by FERA (Federal Emergency Relief Administration 1936). Finally, from July 1935 to June 1943, the federal government returned to the task of providing work for all adults via the Work Projects Administration (WPA) (formerly named the Works Progress Administration). The building of transportation infrastructures was a significant part of the activities of the PWA, CWA, EWR, and WPA. The construction and improvement of public buildings, water-related infrastructures (sanitation, irrigation, flood prevention, among others), and recreational facilities was also quite frequent.
The three main goals of the federal work programs were to help the recovery, to provide relief, and to provide goods and services that improve the long-term well-being of the nation. In order to achieve these three goals, the federal government established several principles in the implementation of work programs. One of the main points was to achieve these goals while breaking away from the minimalist and demoralizing work relief that mostly prevailed before 1933. Individuals applying for a job with the New Deal programs ought to be treated as usual job seekers looking for relevant work and decent pay, rather than untrustworthy beggars requiring routine supervision by social workers.
The achievement of these goals required a systematic organization of federal work programs through planning in order to determine how to allocate enrollees, to determine the completion time of projects and how to deal with fluctuations in the pool of enrollees, and to determine how to prioritize some projects over others while also having a view of future needs. Another important principle of federal work programs was that projects ought to be “self-liquidating, 2 non-competitive with private business, and [paying] a ‘security wage,’ higher than the dole, but not so high as to become a lucrative alternative to private employment” (Salmond 1967: 57).
All this had to be done while being aware of the structure of the labor force involved in work programs. Indeed, except for the PWA and the CWA, work programs readily accepted people whom the private sector did not find appealing to employ: long-term unemployed, old and young, and individuals with limited skills. These characteristics meant that enrollees had a relatively lower productivity, and the focus on labor-intensive techniques of production further hindered productivity gains. It was important to be aware of this fact in order to organize projects in such a way that the quality of projects was satisfying, and the overall cost of production was contained. The issue of productivity was all the more important since there was a willingness to pay a wage that was high enough to make sure that subsistence was met.
The PWA dealt with the productivity issue in several ways. First, it hired engineers to make sure that the projects submitted were not too big, met safety standards, and, among others, were able to meet population growth. Second, it required that competitive bidding be performed by potential private builders in order to ensure that a project could be completed at a reasonable price. Third, the PWA asked for a detailed schedule of accomplishments and was very diligent at making sure that the schedule was followed. Fourth, it demanded that private builders cover any cost incurred beyond the scheduled completion time. Finally, it performed some quality control in the form of questionnaires sent to the sponsors of the projects. The questionnaires revealed that, compared to similar projects that were not funded by PWA, costs of production were higher in half of the projects but quality was higher and maintenance costs were lower. The higher costs of production were mostly due to a PWA requirement to pay the prevailing wage: “the rate earned by a majority of employees in a given occupation in a given locality” (Federal Works Agency 1947: 24). The CWA, after an early period of rushed implementation, also required a detailed presentation of projects, put in place work-safety standards, and improved greatly the auditing of its expenditures (Schwartz 1984: 57ff.). The WPA followed the same procedures.
The state and federal work programs employed on average 2.7 million individuals each month with a peak employment level over 4.5 million (Figure 2). Through the decade, over 10 million individuals obtained work with 8 million employed by the WPA. While these numbers seem impressive, they are quite small relative to the size of the labor force. In terms of unemployment, the maximum proportion of unemployed enrolled in the work programs varied from 50 to 70 percent depending on the unemployment estimate, but on average only 34 percent of the unemployed were enrolled (Figure 3). In terms of employment, the proportion of enrollees represented on average 4 percent of total employment with a maximum around 10 percent during the CWA (Figure 4).
Proportion of Unemployed Enrolled in Work Programs, NSA.
Proportion of Enrollees Relative to Total Employment, NSA.
While the New Deal work programs did improve the relevance of tasks, they were always conceived as emergency programs in the sense that it was thought that the unemployment problem would deal with itself in a short period of time. This state of mind led to a lack of planning, even though planning was already greatly improved compared to the work relief programs of SERAs and FERA. This mindset also led to a lack of accountability with comprehensive reports about the achievements and challenges of the programs not written until the 1940s. As a consequence, there was sometimes a failure to detect problems early and to learn from successes (Macmahon et al. 1941).
Despite some issues in terms of organization and administration, the achievements of the work programs were significant. Beyond providing work, they helped the economy recover and provided solid foundations for future economic growth in the United States. Prior to the Great Depression, the average growth rate of real GNP during an expansion was 6.13 percent, and the 1933 and 1938 expansions recorded a respective average growth rate of 8.77 and 10.63 percent (6.4 percent prior to 1941). While government expenditures contributed to this growth, private domestic investment and consumption were the main contributors to economic growth until 1941. The official unemployment rate dropped by 15 percentage points in four years, from 27 percent in 1933 to 12 percent in 1937. The employment gains were all the more significant if one includes enrollees among the employed; in this case, the unemployment rate almost dropped to 6 percent in 1937 (Figure 1).
Beyond the immediate recovery, the work programs also contributed to the long-term performance of the economy through the construction of infrastructure (Fields 2011), the maintenance and improvement of the health and productivity of the labor force, and the tackling of the environmental issues of the time. Looking through the different reports, one can conclude that transportation was massively improved with about a million miles of roads, streets, and highways; hundreds of thousands of bridges and viaducts; and thousands of airports constructed or improved. Millions of individuals gained access to water supply, sanitation, electricity, and recreational activities. Millions of families received help in terms of housekeeping, provision of clothing and food, and other welfare benefits without which the effect of poverty would have been far worse. Through environmental preservation and restoration that involved the plantation of hundreds of millions of trees, the improvement and construction of thousands of miles of irrigation systems, and, among others, thousands of acres of eradication of insects and invasive plants, the New Deal contributed to the tackling of the environmental issues of the time.
This brief review of the 1930s work programs shows that they shared similar goals to the JG in terms of promoting economic recovery, providing work, and supplying goods and services that are beneficial to society. However, the programs were also not conceived as permanent and were of limited scope, whereas JG is supposed to be a structural program with no financial and political constraint on its capacity to hire unemployed individuals. As shown later, another major difference concerns the wage paid to enrollees.
3. Estimation of the Actual Cost of the 1930s State and Federal Work Programs
In order to employ on average almost 3 million individuals every month, large sums of money had to be spent on administration, capital equipment and raw materials, and, above all, earnings and benefits of enrollees (shelters, clothes, entertainment, education, and others). As noted in the previous section, there were two types of work programs: one type run as work relief by SERAs with financial aid from the federal government (RFC and FERA), another type run and funded mostly by the federal government (New Deal). The goal of this section is to determine the actual cost of the state and federal work programs relative to the size of the economy.
For the General Relief work programs run by the states, FERA provides monthly data about total expenses and labor cost so the estimation is straightforward. For the New Deal programs, the cost estimation is done by first studying the federal side of work programs and then adding expenses incurred by states and localities sponsoring projects. On the federal side, earnings data for enrollees are available on a monthly basis but administrative and other non-labor costs usually need to be estimated either from quarterly or annual data. Data for sponsors’ expenses are not available on a monthly basis but existing data can help to get a reliable estimate.
Figure 5 shows that earnings of enrollees in General Relief and New Deal work programs were around 1.6 percent of GNP with a peak at almost 5 percent during the CWA and 3 percent during the WPA. The WPA was the main employer and so recorded the highest labor cost of all programs.
Earnings of Enrollees Relative to GNP.
Beyond earnings of enrollees, Appendix 1 explains how the paper estimates the total federal expenses and total sponsors’ expenses in federal work programs, and Figure 6 shows the outcome of that estimation. The main difference between earnings and total expenses comes from the EWR, WPA, and CWA programs. The total cost reached on average 2.2 percent of GNP for the period 1933-1943, with a maximum of 6 percent of GNP during the time of the CWA and almost 4 percent of GNP during the time of the WPA.
Cost of State and Federal Work Programs Relative to GNP.
What this exercise tells us is that enrollees’ earnings in Figure 5 provide a reliable estimate of the cost of the work programs.
4. Estimated Cost of JG: Lessons from the Federal Work Programs
A central goal of JG is to provide a job to anybody willing to work, and to do so while paying a living wage. One may wonder how much it would cost to achieve these two goals. The New Deal work programs provide some interesting insights because enough data are available to answer this question for widely different rates of unemployment. The following limits itself to the estimation of earnings of enrollees that have been shown to be the main expenses. These expenses are mark-up to account for other costs.
The first estimate calculates the cost of employing all the unemployed at the wages paid at the time, and the second estimate pays all enrollees a living wage. In order to perform these estimates, one needs to determine the monthly unemployment level to know how many people wanted a job, to get an estimation of the structure of unemployment in order to allocate the unemployed among the different programs, to select a living wage and to know the percentage of enrollees who were not paid a living wage. A more detailed explanation is provided in Appendix 2.
If one focuses on measuring the living wage, one can approach this issue by using the standard of living measures developed by the WPA. In order to check if the WPA monthly wage was appropriate, the WPA calculated two budgets for different cities: a “basic maintenance” budget and an “emergency” budget that was on average 72 percent of the maintenance budget (National Resources Planning Board 1942: 162, Appendix 15). 3 The emergency budget represents a bare minimum standard of living, and if a family lives on that budget for an extended period of time family members will have health issues. Sadly, the National Resources Planning Board (NRPB) notes that a large proportion of U.S. families could not afford the emergency budget, and that General Relief grants were usually far below the emergency budget (ibid.: 202ff., 453ff.). The data show that, in June 1940, about 81 percent of WPA employees did not earn enough to afford the emergency budget, and that on average the monthly WPA wage was 74 percent of the emergency budget (ibid.: Table 27).
The paper uses the data about the emergency and maintenance budgets to construct two estimates of living-wage earnings for non-youth programs (see Appendix 2 for details). Youth programs—focused on relief, education, and apprenticeship—keep their actual wage because they were not designed to provide gainful employment but rather to provide financial support, training, and to sustain moral. They also provided room and board, so this paper assumes that the wage paid to young enrollees was adequate. Table A2.3 shows that the CCC average wage was not that far from the emergency living wage while the NYA average wage was far below it. Only the PWA wage paid on average more than the emergency living wage after 1935. The PWA paid enough to meet the maintenance living wage only the last two years of its existence.
Proceeding according to Appendix 2 yields Figure 7 and Tables 1a and 1b. When the unemployment rate was above 20 percent, the cost of JG would have represented on average 20 percent of GNP. As the unemployment rate declined, the cost would have declined rapidly. Once the unemployment rate reached 6 to 9 percent—roughly the post-WWII average—a JG would have cost about 2 to 4 percent of GNP depending on the living wage paid.
Cost of JG Relative to GNP and Unemployment Rate.
Average of the Cost of JG for Four Different Scenarios, Percent.
Note: FE means all unemployed are enrolled in the New Deal programs, FE + LW also assumes that all JG enrollees are paid a living wage.
Note: Total cost includes all other expenses besides earnings of enrollees (administration, raw materials, earnings of non-enrollees, etc.). Estimates assume a non-labor cost that represents 25 percent of total cost, which is in line with the average for the WPA.
Standard Deviation of the Cost of JG under Four Different Scenarios, Percent.
5. Sources of Under- and Over-estimations
There are several limitations to the previous calculations that lead to over-estimations and underestimations of the cost of JG. If one focuses first on the assumptions that lead to an over-estimation of the cost, they relate to the number of enrollees, and the impact on GNP of paying a living wage. In terms of the number of enrollees, the unemployment data contain people who are over 13 years old, but work programs only accepted individuals who were at least 16 years old. In addition, the estimates assume that all unemployed would be enrolled in the JG, but some individuals probably preferred to look for a non-JG job. Finally, the allocation of enrollees among the different programs affects the cost and it is doubtful that the proportion that existed during the 1930s would have prevailed if all unemployed had been provided work. The allocation performed in Appendix 2 between youth programs and other programs under-estimates youth unemployment in the early 1930s.
Another source of over-estimation of the cost of JG comes from the ignorance of the positive impact of work programs on GNP. First, the work programs generate a demand for goods and services and increase the productivity of the labor force. Second, employing individuals and paying them a higher income than the dole boosts aggregate demand and so raises economic activity and the demand for labor in the private sector. This is even more true if a living wage is paid. The PWA noted that “contractors found that paying good wages brings good men who do a better job quicker, and therefore cost them less” (Public Works Administration 1939: 85). Thus paying a living wage may raise labor productivity even further than the mere employment of individuals seeking work.
In terms of the source of underestimation of the cost of work programs, there are at least two issues that relate to the estimation of the living wage and the ignorance of indirect costs of work programs. In terms of the living wage, the estimates assumed that the emergency budget provided a sufficient lower bound for the living wage, but the minimum living wage probably needed to be higher. Indeed, the emergency budget was calculated on the basis of a family of four, but 47 percent of WPA enrollees had more than four members in their family, so paying them a wage equal to the emergency budget would not have allowed them to meet their emergency needs (National Resources Planning Board 1942: 178). The cost estimate based on the maintenance wage partly accounts for that problem.
In terms of cost, the estimate focuses purely on the expenses incurred by work programs. There were, however, indirect costs in terms of the adverse impacts on the private and public sectors, which may have led to higher unemployment (and so higher work program expenses) and lower GNP. One of the central goals of the New Deal work programs was to employ individuals with minimal adverse disruptions on the rest of the economy. This was not always possible and work programs did perform activities that government and the private sector performed at a higher wage. In most cases, however, the NRPB notes that problems concerned competition with existing public services, because of restrictive provisions that prevented federal work programs to compete with the private sector. For example, some localities did fire some of their employees and rehired them through the work programs at a lower wage (ibid.: 344).
Instead of competing in terms of economic activities with the private sector, a more common issue for the private sector seems to have been in terms of the decline in available labor supply. Some work program enrollees seem to have rejected private job offers, either because they paid less, or because work programs provided more stability in terms of length of employment. The wage issue was especially a problem in the rural South and for unskilled persons who were paid more than the prevailing wage (ibid.: 352). However, private employers sometimes created an artificial shortage by being unwilling to hire blacks, or people who had been on relief, even though they were qualified for the job.
Finally, if the government is willing to provide a job to anybody willing and able, it is highly probable that discouraged individuals and others in and out of the labor force will apply for a JG job. Given that the sum of the “Persons Marginally Attached” and “Persons at Work Part Time for Economic Reasons” is usually quite significant (since 1994 the sum represented on average 78 percent of the unemployed in the United States), one must expect the actual number of people applying for a JG job to be significantly higher than the level of unemployment. The New Deal work programs did set up some limits in terms of physical capability, earnings, and other criteria to avoid some of the issues that may arise, and probably something similar would have to be implemented with JG. In terms of the previous cost estimates, the impact would depend on the level and structure of the marginally attached and underemployed. Unfortunately, there is no data for the period of estimation. In addition, one also would have to account for the positive impacts on GNP of employing these people to get a fair estimate of the cost.
6. Conclusion
The work programs implemented during the 1930s employed millions of individuals every month in order to alleviate the socio-economic problems induced by the Great Depression. The cost of the programs was rather small but they employed only a fraction of the unemployed and did not pay them enough to meet subsistence. Once these elements are included, the work programs turn out to be quite expensive (around 20 percent of GNP) during the early part of the 1930s when the unemployment rate was at 20 percent or more. Once unemployment receded to a usual level (between 6 and 9 percent of the labor force) the gross cost of JG is quite low (between 2 and 4 percent of GNP).
Getting an estimate of the gross cost of running a JG is important because it helps to get an idea of the potential inflationary pressures (Tcherneva 2012), the potential disruptive impacts on the rest of the economy, and the budgetary impact of JG. As much as possible, this cost should be compared to the actual benefits of JG even though it might be difficult to do that well. Here again the 1930s work programs provide some insights and show that the benefits were substantial and durable. If one focuses purely on the economic benefits, they ranged from helping the recovery, to improving the natural environment, improving the productivity of workers, and providing access to basic infrastructure to millions of individuals.
One may wonder what the impact on government spending would be if JG was implemented. This paper has not answered this question, but the impact, if positive, is not fully positive. JG would raise government spending but government expenses on unemployment insurance, healthcare, and the penal system, among others, would decline with the implementation of JG. If one applies this logic to the most recent economic recession when the official unemployment rate peaked at 10 percent, the gross cost of JG would have been around 5 percent of GNP at the maximum. At the same time, JG would have allowed some individuals to keep their home, would have sustained communities, would have used unemployed to repair and expand U.S. infrastructure, would have reduced unemployment insurance, and among others would have kept purchasing power in the hands of households. All of this means that government spending would have increased by less than 5 percent of GNP given everything else.
Footnotes
Appendix 1: Estimation of the Actual Cost of Work Programs
In order to determine the cost of federal work programs, administrative costs and other non-labor costs incurred by the federal government must be added to the cost of enrollees presented in Figure 5. Unfortunately, data about overall administrative expenses are limited except for the WPA that provides quarterly data about administrative earnings and annual data about other administrative costs. Data about administrative cost are not available for the PWA and CCC, and they are an insignificant proportion of the GNP for the NYA (Federal Security Agency 1944: 243). For the CWA, administrative expenses were on the order of $37 million for the nine months of operation of the CWA, which is about 0.71 percent of GNP and about 4 percent of CWA expenses over nine months (Work Progress Administration 1939: Table 12).
Data about other non-labor costs are scarce but it is possible to get an estimate of the total cost. The WPA and CWA provide data about monthly total expenses for the period of their existence, which provides a reliable check point for estimates because they were the biggest New Deal work programs. If detailed data of each component are not available, an alternative way of calculating the total cost of federal work programs is provided by Table A1.1. It shows the average annual cost per individual of running the WPA, CCC, and NYA. For the CCC, the value of $1,004 per man-year (taken from Salmond 1967) is used before 1936 and from 1941. For the NYA, an average of the 1936-1940 values is used from 1941. The average difference between earnings to GNP and cost-per-man-year to GNP is 0.27, 0.08, and 0.007 percentage points for the WPA, CCC, and the NYA respectively.
In order to have an overall estimate of the cost of the work programs, one also needs to estimate the cost incurred by states and localities that sponsored the projects implemented by the New Deal programs. These costs existed for the CWA, the WPA, the PWA, and the NYA. For the CWA, data are available on a monthly basis and sponsors’ expenses amounted to $91 million over the time of its operation, which is about 9.6 percent of total CWA expenses and 1.74 percent of GNP spread over 9 months (Work Progress Administration 1939: Table 9). Data for the NYA available from the NRPB (Table A1.2) show an insignificant difference between total expenses and expenses of states and localities. Most of sponsors’ expenses for the NYA came from the out-of-school program and represented on average 13 percent of total NYA out-of-school expenses (Federal Security Agency 1944: Table 27). Figure A1.1 shows that, on average, 77 percent of WPA expenses came from the federal government, and non-federal expenses averaged 0.42 percent of GNP per year from 1936 to 1943. The figures are consistent with the NRPB data in Table A1.1 that stop in 1940. The data show that sponsors’ expenditures grew in proportion to overall expenses but mostly stayed below 0.5 percent of GNP.
Appendix 2: Estimation of the Cost of JG
Data about the unemployment level are available on a monthly basis for the period 1929-1943 from four different sources: the National Industrial Conference Board (1929-1940), the Work Projects Administration (1935-1941), the Bureau of Labor Statistics (1940-1943), and the American Federation of Labor (1929-1941). Nixon and Samuelson (1940) analyzed the different estimates and concluded that the AFL data are the most reliable, but the paper uses all of them by calculating their average.
The NYA and the Census Bureau provide some data about the proportion of unemployed by age groups for the years 1930, 1937, 1940-1943. All other years are estimated through a linear interpolation of available data in order to get Table A2.1. 4
The linear interpolation underestimates the proportion of youth unemployment in the early 1930s.
Table A2.1 is combined with the number of unemployed to allocate them by age groups. This helps to allocate the unemployed among New Deal work programs for the youth (CCC and NYA) and others (WPA, PWA, CWA, and others). This allocation is important for the cost estimate because each program paid a different average monthly wage. The allocation within each category of work program was done on a proportional basis. For example, there were an estimated 10,241,370 unemployed individuals in January 1936, with 3,466,850 of them between the age of 14/15 and 24 years. New Deal data show that the NYA employed 44.3 percent of the young enrollees at the same date, so 3,466,850 is multiplied by 44.3 percent. Thus, 1,535,646 of the young unemployed would have been enrolled with the NYA and 1,931,204 would have been employed by the CCC. Table A2.2 shows the results of proceeding in similar fashion with the other programs. In the early part of the 1930s, the CWA, the PWA, and the CCC would have been the three main programs. From the mid-1930s, the WPA, the NYA, and the CCC would have employed most of the unemployed.
For part of 1933 and 1943, New Deal work programs did not exist, so not all unemployed individuals could have been enrolled in them. This explains why the difference between total unemployment and the sum of all enrollees is not zero for these two years.
Once one knows the estimated number of unemployed individuals and their allocation amongst the different work programs, it is possible to determine the amount of earnings that would have to be paid relative to GNP. In order to do that, the official monthly earnings data for each program are used and are divided by the official monthly number of enrollees in each program. This provides the average monthly wage per person for each program (Table A2.3). This average wage is multiplied by the number of enrollees obtained in Table A2.2 in order to calculate the earnings cost of JG when all unemployed are enrolled in the work programs.
In order to determine the monthly living wage, the paper uses the emergency budget and maintenance budget data of the WPA. In June 1940, the average monthly wage in the WPA was $57.86 and, following WPA data, it is assumed that this only covers 74 percent of the emergency budget. This yields an emergency living wage of $78.19 for the month of June 1940 that is adjusted for inflation for the decade. 5 Regarding the maintenance budget, the NRPB states that in March 1935, it was $1,261 per year, i.e. $105.08 per month. This is adjusted for inflation to get the value of the maintenance living wage from 1933 to 1943.
Except for the PWA, none of the work programs paid enough to meet the emergency budget.
If one notes W the monthly wage paid to enrollees, the following formula is used to determine earnings at full employment that meet the emergency and maintenance budget for each work program for adults:
With:
81 percent of enrollees earn the living wage while 19 percent (individuals employed in skilled activities, professionals, and technicians) earn 12 percent above the living wage. This reflects the wage structure that prevailed in June 1940 in the WPA, and it is assumed to apply to the CWA, the PWA, and other programs. Once this is done, earnings estimates are assumed to represent 75 percent of total cost (grossly the cost structure of the WPA), which allows to calculate total cost.
Appendix 3: Monthly Unemployment Data
Unemployment Rate with Work Programs, Seasonally Adjusted, Percent.
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Annual | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
1.6 | 1.9 | 2.1 | 2.5 | 3.2 | 3.4 | 2.7 | 1.5 | 2.4 | 2.4 | 2.6 | 3.1 | 2.5 |
|
|
4.1 | 4.9 | 5.7 | 6.1 | 6.1 | 6.7 | 7.8 | 8.5 | 9.4 | 10.3 | 11.3 | 12.1 | 7.8 |
|
|
13.0 | 13.5 | 13.7 | 14.0 | 14.4 | 14.8 | 15.4 | 16.0 | 16.8 | 17.8 | 18.7 | 19.5 | 15.6 |
|
|
20.4 | 21.2 | 21.9 | 22.9 | 23.9 | 24.6 | 25.0 | 25.4 | 25.0 | 25.0 | 25.2 | 25.7 | 23.9 |
|
|
23.8 | 23.9 | 24.0 | 23.4 | 22.5 | 21.5 | 20.3 | 19.0 | 18.0 | 18.0 | 15.0 | 15.3 | 20.4 |
|
|
13.7 | 13.7 | 14.2 | 16.3 | 15.5 | 15.2 | 15.3 | 15.2 | 16.2 | 16.3 | 16.6 | 16.0 | 15.4 |
|
|
15.3 | 15.0 | 14.9 | 14.3 | 14.1 | 13.9 | 13.7 | 13.7 | 13.8 | 13.3 | 11.7 | 10.4 | 13.7 |
|
|
9.8 | 9.4 | 8.9 | 8.6 | 8.4 | 8.3 | 8.1 | 7.5 | 6.8 | 6.5 | 6.6 | 6.9 | 8.0 |
|
|
7.3 | 7.2 | 7.0 | 6.8 | 6.2 | 6.6 | 6.8 | 7.4 | 7.9 | 8.4 | 10.3 | 11.9 | 7.8 |
|
|
12.8 | 13.2 | 13.4 | 13.1 | 13.3 | 12.9 | 12.1 | 11.8 | 11.3 | 10.8 | 10.5 | 10.3 | 12.1 |
|
|
10.6 | 10.7 | 10.9 | 11.0 | 10.9 | 10.3 | 10.6 | 10.8 | 11.0 | 10.5 | 10.6 | 10.4 | 10.7 |
|
|
10.7 | 10.7 | 10.5 | 10.6 | 10.7 | 10.7 | 10.8 | 10.2 | 9.2 | 9.2 | 8.9 | 8.3 | 10.0 |
|
|
7.6 | 6.8 | 7.2 | 7.2 | 6.7 | 7.0 | 6.6 | 6.4 | 6.2 | 4.9 | 4.3 | 4.0 | 6.2 |
|
|
3.7 | 3.6 | 3.6 | 3.3 | 3.2 | 3.9 | 2.8 | 2.3 | 2.5 | 2.8 | 2.7 | 2.5 | 3.1 |
|
|
0.8 | 1.2 | 1.5 | 1.6 | 1.9 | 2.9 | 1.3 | 1.4 | 1.9 | 2.1 | 2.0 | 2.0 | 1.7 |
Sources: See Figure 1.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
1
These dates reflect the dates of creation and order of liquidation of each program. Programs may not have hired immediately, and they may have continued to employ individuals after the order of liquidation, but they no longer hired new individuals.
2
The term “self-liquidating” comes from the Emergency Relief and Construction Act of 1932 and meant that a project had to pay for itself by generating revenues large enough to cover its costs. This was subject to interpretation as many projects could only indirectly pay for themselves via increases in states’ tax revenues. It turned out that the self-liquidation principle was the hardest to achieve as many projects were not self-liquidating under any kind of definition; this is to be expected given the type of activities performed by work programs (
: 19).
3
4
Only annual averages are presented in the Tables. Monthly data are available on request.