Rent seeking for madness: the political economy of mental asylums in the United States, 1870 to 1910

We are not fond of the terms “insane” and “feeble-minded”. However, those are the terms used in the historical sources that we rely on herein. We will use the historical terms for the sake of simplicity and without any derogatory intent.

Administrative costs represented roughly 25% of the budgets of almshouses, while the early government welfare programs expended 2–3% of their budgets on administration (Lindert, 2004, p. 35).

This antagonism is most evident in the literature on the origins of consumer regulations and antitrust (Libecap, 1992; McChesney & Shughart, 1995; DeLorme et al., 1997; Olmstead & Rhode, 2015; Newman, 2019a; Geloso, 2020).

Examining boards acted essentially as policing entities that enforced codes of ethics, degree requirements, reserved acts and entry requirements (Hamowy, 1979, p. 77).

As documented in the vocabulary of the time. For example, the 1880 and 1890 US Census specifically lists insane and feeble-minded individuals as part of the “degenerate classes” or “defective classes”. The Council on Mental Hygiene, when issuing reports on the number of “mental defectives” housed in asylums, called for more confinement (outside of family houses) of the feeble-minded to advance American interests (Rothman, 2002, p. 322).

Most of that attention is on the quality of care provided to patients in asylums, e.g., overcrowding, death rates (Grob, 1992; Noll, 1995; Pressman, 2002), the ethical behavior of asylum physicians, notably regarding the use of questionable procedures such as lobotomies (March and Geloso, 2020) and the forcible commitment of otherwise sane individuals (Lombardo, 2008).

The AMA’s first successful attempt to seek rents occurred in 1877 when the Illinois state legislature allowed the State Board of Health (enacted at the same time) to refuse to grant medical licenses based on the perceived quality of physicians’ medical degrees. The number of physicians in Illinois declined dramatically. As Starr (1982, p. 104) notes, “within a decade [of passing legislation], 3000 practitioners were said to have been put out of business.” That represented a decline of nearly 40% in Illinois’s physician labor force (Hamowy, 1979, p. 82). After successfully passing legislation in Illinois, Burnham (2015, p. 311) notes that “state after state rushed to set up examining boards.”

By 1844, superintendents had formed the Association of American Superintendents of American Institutions for the Insane, three years before the broader medical profession established the American Medical Association, which pursued similar political goals (Burnham, 2015; Starr, 1982).

State governments also benefited from the collaboration. Sutton (1991, p. 667) speaks of “career contingencies in the commitment process” providing benefits for political figures. By funding large-scale public asylums, state legislators earned patronage from physicians caring the mentally ill, contractors who built asylums, and from families of mentally ill or disabled individuals who proved too difficult to manage at home (Rothman, 1971; Sutton, 1991).

Involuntary commitment also was facilitated by stretching psychiatric terminology to incorporate more diagnostic criteria (Luchins, 1988, p. 477).

The main problem with sources of financial data is that the definitions change between census documents. For some years, but not all, current expenditures are blended with capital expenditures. For some other years, only one type of expenditure is reported. Comparisons across time therefore are difficult. More problematic is the fact that not all asylums reported financial data. While we have more accurate numbers on patient populations, neither expenditures nor appropriations are reported consistently enough to be relied on herein.

Registration laws required physicians (and other healthcare providers in some instances) to register with a county medical society or some other county official in order to practice medicine in the state. However, Hamowy did not report information on registration laws for all states and we had to turn to Baker, (1984) for observations on that variable.

Examining boards largely were used to exclude individuals who did not hold medical school diplomas; the boards also could operate as barriers to entry in conjunction with a code of ethics.

The passing of a licensing examination was required of all candidates seeking to practice medicine in the state whether medical school graduates or not.

The code of ethics did not differ dramatically by state because the empowering of examining boards to revoke, or refuse, certification “effectively” legislated “the code of ethics of the American Medical Association” (Hamowy, 1979, p. 113).

Some territories, such as Dakota, which eventually became two US states, had at least one asylum before statehood.

Otherwise, we would observe infinitely large jumps in care levels from a false zero base.

The years are census years. See online supplement for more details on the special census of 1904.

In the online supplement, we also asked if similar results are found when using rules restricting the practice of midwifery provided by Anderson et al. (2020). While they supported our overall intuition, most of the midwifery rules were adopted after 1910 (12 states had adopted a rule against midwifery before 1910 and 8 before 1900), which limits its reliability. As such, we report the findings as supplementary results in the supplement.

Sutton (1991) also considered two other explanatory variables that we ended up dropping. The first was the number of Civil War pensioners estimated from federal documents, but he overlooked the fact that Confederate veterans were not eligible for pensions (see more below). He also included a dummy variable for traditional party organization. It was defined as states in which political machines flourished, whether in a one- party or competitive environment. The variable was time-invariant and had a value of 1 for thirteen states. However, our fixed-effect estimator captures the effect of such a time-invariant control and we did not include it.

The empirical literature has thrown into contention the link between insanity and urbanization (Sutton, 1991, p. 671). However, it has been entered in previous work as a control variable because it is related to “lower social tolerance for aberrant behavior” as well as the heightened “status of medical and mental health professionals” (Sutton, 1991, p. 671). Furthermore, it is included in the work of Sutton (1991) because the data do not allow for an easy and consistent distinction between city and state institutions (see online supplement). Thus, “a measure of urbanization is a rough proxy for a municipal asylum system” whereby larger urban populations also meant more asylums regardless of whether or not they are mandated by state or city governments (Sutton, 1991, p. 672).

Only 0.2% of asylum patients in the 1920s were younger than age 15 (US Bureau of the Census 1926, p. 27).

When institutions that cared for black populations were operating, they were less well-funded than institutions that cared for white populations (Grob, 1983, p. 26).

We relied on the report on benevolent institutions for 1910 to identify the agencies (US Census Office, 1911, p. 14). We then gathered numerous legal documents, legislative records, session papers and state histories to identify their dates of creation. However, state welfare agencies are coded as dummy variables and they do not address degrees of welfare support. A better variable for public interest would have been pensions to veterans because it captures support for the group most likely to access asylums in the absence of taxpayer-financed support. Sutton (1991) relied on federal pensions. However, that choice was problematic because veterans of the Confederacy’s armed forces, who were not eligible for federal pensions, benefited from state level pensions (Eli and Salisbury, 2016), which were not covered by his variable. Normally we could simply add information about state-level pensions to the federal ones to get a more complete picture. However, some southern states provided no relevant information. In the online supplement, we report a robustness check includes veterans receiving pensions with and without adding the state-level pensions to confederate veterans as collected by Eli and Salisbury (2016). Our results are unchanged, but we take them with a grain of salt because of the incompleteness of confederate pensions data and the differences in generosity between programs for which we cannot account.

We have defined dependents as the sum of individuals below 15 and above 65 years of age.

We relied on the same sources as Sutton (1991).

We also estimated our regressions weighted by population and the results are very similar: examining boards are significant in all specifications.

For these estimates, the year-fixed effects are coded as i.year in Stata, thereby avoiding assuming a linear time trend (each year is unique). However, we also tried using the c.year command to capture the fact that, because of the special census of 1904, the census years are not spaced evenly. When c.year is substituted fort i.year to capture national trends, the coefficients for registration laws always are significant at the 5% level; examining boards always are significant at the 1% level. Some other laws, such as the exclusion of substandard medical schools, become significant at the 10% level and increase modestly the level of institutionalization. The same applies when we used c.year with the population housed in feeble-minded institutions.

In the online supplement, we estimated the effect of regulations against midwives using the data of Anderson et al. (2020). Eight states adopted such prohibitions during the nineteenth century; most states that adopted them did so in the twentieth century, after our period of interest. Thus, we were unwilling to report those results in the main body of the article even if they reinforce our claim because the marginal effects were significant in two out of four specifications and always positive.

However, a note of caution must be raised regarding the observations for 1870 that we obtained from the work of Turner et al. (2007). To conduct their own study, Turner et al. (2007) had to construct estimates of state-level incomes for 1870 to complement other existing sources. To do so, they had to make some assumptions based on later, or earlier, census years. Thus, it could be that observations on the income levels in 1870 are approximative. However, for our purposes, that possible flaw does not affect the coefficients or significances of the medical laws.

Assuming that a state switched from 0 to 1 (i.e., not having an examining board to having one), the percentage impact of the law on institutionalization (because they are logged) is 100*[exp(\(\beta_{1}\)) − 1] as proposed by Halvorsen and Palmquist (1980).

We thank Gregory Niemesh for this useful suggestion.

We thank Allison Shertzer for this insightful comment.

For instance, Illinois created its examining board in 1877. The “fake” is coded as 1 for the observations for 1870 and 0 for 1880, 1890, 1904, and 1910.

We attempted to create such a measure using the special reports in the censuses of 1880 and 1890. However, the published financial measures are not consistent over time and do not encompass a methodologically consistent set of hospitals. Moreover, little to no information about the financial conditions of private asylums is provided. Even less information exists about special institutions for the feeble-minded.

Confederate veterans were not allowed to receive federal pensions. Whatever pensions they received were funded by state legislatures and were considerably less generous than for Union veterans (Eli and Salisbury, 2016).