Contracting for sex in the Pacific War

The protracted political dispute between South Korea and Japan over the wartime brothels called c̈omfort stationsöbscures the contractual dynamics involved. These dynamics reflected the straightforward logic of the c̈redible commitmentss̈o basic to elementary game theory. The brothel owners and potential prostitutes faced a problem: the brothel needed credibly to commit to a contractual structure (i) generous enough to offset the dangers and reputational damage to the prostitute that the job entailed, while (ii) giving the prostitute an incentive to exert effort while working at a harsh job in an unobservable


Introduction
As the Japanese army advanced and retreated across East Asia in the 1930s and 40 s, it encouraged private entrepreneurs to establish semi-official brothels next to its bases. Venereal disease had wreaked havoc on its Siberian expedition in 1918, and it needed that risk under control. From the cooperating brothel owners, it demanded that the prostitutes undergo regular medical examinations. In exchange, it promised to forbid its men from patronizing brothels anywhere else.
To staff their brothels, the cooperating entrepreneurs hired women primarily from Japan and Korea. All else equal, the soldiers preferred Japanese women. Among the non-Japanese staff, they preferred the Koreans. Korea was part of the Japanese nation, after all (Japan had annexed the peninsula in 1910), and most Korean women spoke at least some Japanese. The army called the coop-E-mail address: ramseyer@law.harvard.edu 1 Mitsubishi Professor of Japanese Legal Studies, Harvard University. I gratefully acknowledge the generous and helpful comments and suggestions of Elizabeth Berry, Yoshitaka Fukui, Il-Young Jung, Mitsuhiko Kimura, Yoshiro Miwa, Jason Morgan, Minoru Nakazato, Gregory Noble, Geoffrey Ramseyer, Jennifer Ramseyer, Frances Rosenbluth, Richard Samuels, Henry Smith, Frank Upham, and the referees and editors of this journal. erating brothelscomfort stations(ianjo). It called the prostitutes comfort women(ianfu).
Consider the contracting problem. To staff these brothels, entrepreneurs needed to recruit young women. They obviously needed to promise the women extremely high pay. Even in the best of circumstances, prostitution is harsh and dangerous work and exacts a major reputational penalty. Women will take the job only if they expect to earn income both high enough to offset these costs, and significantly higher than their next-best alternative.
For work in distant wartime locations, the entrepreneurs needed to promise pay substantially higher even than in the Tokyo or Seoul brothels. To the standard travails of prostitution, the comfort stations added the risk of warfare. They added the cost of living in a foreign environment. The women would be away from friends, allies, people to whom they could turn for help should the brothel try to cheat them. And they raised the difficulty of absconding should the brothel try to cheat, and the price of returning home if they stayed their term and retired at the end.
Although they needed to promise the women high pay, the entrepreneurs could not just offer a high monthly wage. They were hiring women to perform unpleasant work in impossible-tomonitor surroundings. If they promised a fixed monthly wage, they gave each woman an incentive to be sufficiently unpleasant that no one asked for her at the front desk. Necessarily, they needed a wage contract that rewarded effort. https://doi.org/10.1016/j.irle.2020.105971 0144-8188/© 2020 Elsevier Inc. All rights reserved. Somehow, the entrepreneurs needed to make their promises of high pay from this incentive-based wage contract credible. To entice a woman to take this contract, an entrepreneur needed to convince her that she would earn high wages. Yet she knew that he had every incentive to exaggerate her potential earnings, and he knew that she knew. Some women would have doubted their own capacity to generate high earnings. In some jobs, a woman might have experimented for a short period to learn how much revenue she could earn. Given the reputational hit she took simply for entering the occupation, she could not do that here.
To recruit women to brothels on the military front, the entrepreneurs and women faced contractual problems harder by orders of magnitude. Most obviously, the women faced all the dangers of war -fighting, bombing, and rampant disease. The women also faced far more serious risks of brothel non-performance. Should a brothel owner in Tokyo try to cheat on the contract, a prostitute could complain to the police. They would not all be sympathetic, but some would. She could sue the brothel owner in court for nonperformance. Some did exactly that, and won. She could walk out of the brothel, and disappear into the anonymity of the Tokyo metropolis. In a distant foreign country, she faced the risk that she might not be able to do any of this.
The entrepreneurs and the women addressed these problems with a multi-year indenture agreement that bundled an up-front advance, additional cash compensation, a maximum term, and the right to quit early if a woman generated enough revenue. In the article that follows, I detail the economic logic to these contracts. I compare the sexual service contracts the entrepreneurs and women negotiated (i) for the comfort stations, with the contracts they negotiated (ii) for domestic Japanese brothels, (iii) for domestic Korean brothels, and (iv) for the non-comfort-station war-time brothels across Japanese-governed East Asia.
I begin by outlining the contracts used in domestic Japanese brothels (Sec. 2.2). I compare them to the contracts used in Korea, and to the contracts used in the unofficial (non-comfort-station) brothels elsewhere within the Japanese empire (Secs. 2.3, 2.4). Finally, I turn to the contracts used by the comfort stations themselves (Sec. 3).

Introduction
The comfort stations operated as the overseas military analogue to the private brothels in Japan and Korea. Whether in Japan or in Korea, brothels hired, and women looked for work. The work at stake in these transactions involved sexual services, but the economic logic to the arrangements that the two parties -brothel and prostitute -negotiated reflected the resources and alternative opportunities that both sides understood each other to hold. Recruiters and brothels could lie, but prostitutes could shirk or take the money and run. The women understood that the recruiters and brothels could lie, and understood too that they could shirk or disappear. The brothels could replace them with other women, but the women could find other work too, however low-paying. Yes indeed, parents did sometimes sell their daughters and brothels did sometimes trap women or keep them virtually imprisoned. But the economic logic (detailed below) to the contractual arrangements reflects the fact that brothels could not -and did not -trap or imprison all or even most of the women.
The contracts themselves reflect the intelligence and resourcefulness of the women involved. They were people with few attractive alternative economic opportunities, but they had some and the terms of the contracts suggest they knew they had some. They chose prostitution over those alternative opportu-nities because they believed prostitution offered them a better outcome. Recruiters could lie. Brothels owners could cheat. Parents could abuse their children and steal the advance payment that the women earned. But the contracts suggest that the women knew that recruiters could lie, knew that brothel owners could cheat, and did not quietly defer to abusive parents.

Japan
1. Licensed prostitutes. -(a) The contracts as stated. Prostitution had been a licensed industry in pre-war Japan (see generally Ramseyer, 1991). In 1924, 50,100 licensed prostitutes (shogi) worked out of 11,500 licensed brothels in Japan (Fukumi 1928: 50-56, 178;Kusama, 1930: 14-26). Most commonly, the licensed prostitutes worked under multi-year indenture contracts. 2 (a) The brothel paid the woman (or her parents) a given amount upfront, and in exchange she agreed to work for the shorter of (i) the time it took her to pay off the loan or (ii) the stated contractual term (b) The mean upfront amount in the mid-1920s ranged from about 1000 to 1200 yen. The brothel did not charge interest. (c) The most common (70-80 percent of the contracts) term was six years. (d) Under the typical contract, the brothel took the first 2/3 to 3/4 of the revenue a prostitute generated. It applied 60 percent of the remainder toward the loan repayment, and let the prostitute keep the rest.
I know of no source detailing how often the upfront payment went to the woman herself, how often it went to her parents who kept it on her behalf, and how often it went to abusive parents who kept it for themselves. Note, however, that prostitutes were not prisoners. In cities like Tokyo, they could easily leave their brothels and disappear into the anonymous urban environment. If they did, the brothel would then sue their parents on the cash advance (a prostitute's father typically signed the contract as guarantor). That this only happened occasionally suggests (obviously does not prove) that most prostitutes probably chose the job themselves; they probably did so because -in their minds -it made the best of a bad situation.
(b) The contracts as applied. In practice, the prostitutes repaid their loans in about three years and quit. Surely, historians sometimes insist, the brothels must have manipulated the charges for food and clothing to keep prostitutes mired in perpetual debt. At least on a large scale, however, they did not do this. Probably, the brothels -established institutions with a large capital investment -realized that cheating on their initial contract would raise their future recruitment costs. Not only did the brothels specifically promise a woman she could quit debt-free at the end of six years regardless of the revenue she generated, they generally kept their promise.
If brothels manipulated charges or otherwise cheated on their terms to keep prostitutes locked in debt, the number of licensed prostitutes should have stayed reasonably constant at least up to age 30. The minimum age for licensed prostitutes was 18. In 1925, there were 737 licensed Tokyo prostitutes aged 21, and 632 aged 22. There were only 515 aged 24, however, 423 age 25, and 254 age 27 (Fukumi, 1928: 58-59).
Similarly, if brothels were keeping prostitutes locked indebt slavery,ẗhe number of years in the industry should have stayed constant beyond six. Yet of 42,400 licensed prostitutes surveyed, 38 percent were in their second or third year, 25 percent were in their fourth or fifth, and only 7 percent were in their sixth or seventh (Ito, 1931: 208-11;Kusama, 1930: 281). On a workforce of about 50,000 licensed prostitutes, 18,800 women registered as new licensed prostitutes in 1922 and 18,300 de-registered (Yamamoto, 1983: 388;Ito, 1931: 211-13). Consistent with a general tenure of about three years, in other words, one third of the work force replaced itself every year (Keishi, 1933: 96-98;Kusama, 1930: 227-28).
2. The contractual logic. -(a) Credible commitments. This indenture contract in the licensed sector reflected the straightforward game theoretic logic ofcredible commitments (Ramseyer, 1991). Young women understood that prostitution was dangerous and harsh, and imposed a large upfront hit to their reputations. What is more, they understood that they incurred that reputational hit even if they quit after a very short period. Recruiters promised them very high wages, but they understood that recruiters had every incentive to exaggerate. Wholly apart from the recruiters' incentives to exaggerate, some women would simply have doubted their own ability to generate high revenue levels.
As a result, before a young woman agreed to work at a brothel she needed credible assurance that she would earn wages high enough to compensate her for the negative characteristics associated with the job. Were there no reputational hit to entering the industry, she could try the job for a few months to see how much she could earn. Given that she incurred the reputational cost even from a short stint, however, she could not readily verify the recruiters' claims.
The women forced recruiters to overcome this problem of promissory credibility by paying each prostitute a large fraction of her earnings upfront, and capping the number of years she would have to work. If the brothel paid her 1000 yen in advance and set the maximum term to six years, she knew the minimum she would earn. She also knew that if she repaid it earlier (as most prostitutes did), she would earn even higher effective monthly wages.
In turn, the brothel needed a way to create an incentive for its prostitutes to please their customers. The women performed harsh work in impossible-to-monitor environments. If brothels paid them a fixed wage (like an initial 1000-yen payment on a fixed six-year term), they had little incentive to try to please their customers. If a prostitute were sufficiently unpleasant that guests seldom requested her by name, so much the better.
By coupling a maximum six-year term with the ability to quit early, the brothel gave a prostitute an incentive to please her customers. The more customers requested her, the more revenue she generated. The more she generated, the sooner she could quit.
(b) Loans. Obviously, through these contracts the brothel extended to the woman or her parents a loan. If she or her parents needed that cash advance, the employment contract offered it. Young European men in the 19th century needed cash to pay their passage to North America; redemptioner contracts (a varia-tion on indentures) offered that advance. So too here: the woman's promise to work facilitated a credit extension.
Two aspects of this labor market, however, suggest that the demand for a loans does not explain the use of these contracts in the sexual services market. First, very few other labor contracts included a loan with the contract. Suppose parents needed a cash advance. If a daughter could obtain a cash loan from a brothel, a son could have obtained a cash loan from a factory. Yet sons and daughters rarely took large cash advances upon signing employment contracts. Although some other employers did sometimes lend money to new hires, they did so only haphazardly, and only for relatively small amounts.
Second, the licensed brothels paid the cash advance to all new hires. Although some prospective prostitutes and some parents would have wanted 1200-yen cash loans, many would not. The money did not come free. The brothels did not charge a stated interest, but they obviously discounted the woman's earnings to present value. Were the brothels paying the large cash advances only in response to a demand in the credit market, they would have paid the indentures to some of their hires and not paid to others. The fact that they coupled the massive cash advances with all of their labor contracts suggests that some other contractual dynamic was at play.
3. Unlicensed prostitutes. -Below the licensed prostitutes in this sexual services market worked the independent, unlicensed prostitutes. Given the choice between the two sectors, most prostitutes preferred the licensed. From 1920 to 1927, of all the women who applied for work as licensed prostitutes in Tokyo, only 62 percent obtained jobs (Chuo, 1926: 381-82;Kusama 1930: 27-30, 36). Far from being job that no one wanted, positions in the licensed brothels were jobs with half again as many applicants as the brothels wanted to hire. Many of the unlicensed prostitutes were those women whom the licensed brothels had refused to hire (Kusama, 1930: 37). Historical records contain no reliable censuses of the unlicensed workers, but otherwise trustworthy observers put their number at about 50,000 in the mid-1920s (Fukumi 1928: 26-28, 32, 50-56, 178).
Because the unlicensed prostitutes nominally violated the law, they lacked the option of working for an established brothel. Brothels developed reputations. Given that illegal unlicensed prostitutes could not work for a brothel with a reputation for high quality service, the unlicensed prostitutes earned less money. Among female workers from the northern Akita prefecture in 1934, licensed prostitutes earned room & board plus 884 yen per year. Bar maids (shakufu; the general euphemism in this literature for unlicensed prostitutes) made 518 yen, waitresses made 210 yen, and other women workers made 130 yen (Shakai, 1935: 160-61).
The unlicensed sector also presented clients with higher risks. By law, licensed prostitutes underwent weekly medical examinations for venereal disease, and infected women could not return to work until they recovered. In 1932, 3.2 percent of licensed prostitutes in Tokyo had venereal or other infectious disease. The same study found a 9.7 percent rate among unlicensed prostitutes. Other studies confirm a 1-3 percent infection rate among licensed prostitutes, but find rates much higher than 10 percent among the unlicensed. 3 4. Karayuki. -As Japanese businessmen moved abroad for work, young women followed. There in the foreign countries, the women worked as prostitutes for the Japanese clientele.Karayukisan,Japanese called them:ẅomen heading abroad (Nihon, 1920). Given the usual preference among expatriate Japanese men for Japanese women, they earned substantially higher wages than their local competitors. Given the cost of relocating abroad, they earned generally higher wages than they could earn within Japan (Park, 2014: 451).
The expatriate prostitutes tended to come from two discrete communities on or near the southern island of Kyushu: Shimabara and Amakusa. That most of them came from a few small communities is crucial: it makes implausible any notion that they had been tricked by duplicitous recruiters. Trickery works when the target audience does not know what is at stake. When young women (or girls) from small, closed communities leave for several years and then return, they report what happened. Word travels, and others in the community learn what the trip entails.
Author Tomoko Yamazaki (1972) traveled to Amakusa to explore this history. There, she befriended an elderly emigrant prostitute named Osaki. Osaki had indeed worked many years abroad, but hers was not a story either of paternal oppression or of sexual slavery. Osaki had been born in a small village to a family who already had a boy and a girl. A few years after her birth, her father died. Her mother then found a new lover. As he had no interest in her very small children, she abandoned them and married him anyway. The three children survived together in a tiny shack, and scrapped together what they could to eat. Other women in the community had worked as prostitutes abroad, and had returned with substantial sums of money. In time, her older sister left to work abroad as a prostitute herself.
When Osaki turned ten, a recruiter stopped by and offered her 300 yen upfront if she would agree to go abroad. The recruiter did not try to trick her; even at age 10, she knew what the job entailed. She discussed it with her brother, and decided to take the work to help him establish himself in farming. She travelled to Malaysia, and worked as a maid for three years. She was happy, she recalled. Her family fed her white rice and fish every day, which was more than the three abandoned children had been able to scavenge in Amakusa.
At age 13, she began working for the family as a prostitute. Because of the cost of passage and three years of room and board, she now owed 2000 yen. Under the new terms, customers paid 2 yen for a short stay and 10 yen for an overnight visit. The brothel owner kept half the amount, and provided room and board. Out of the remaining half, she paid down her outstanding balance and bought cosmetics and clothing. If she worked hard, she found that she could repay about 100 yen a month.
Before Osaki had finished repaying her loan, her owner died and she found herself transferred to a brothel in Singapore. She disliked her new owner, so one day she and some of the others went down to the harbor and bought a ticket back to Malaysia. The point is important: even overseas, women who disliked their jobs at a brothel could -and did -simply disappear.
Osaki found a new brothel. She liked the couple who owned it (and they negotiated her release from the earlier brothel), and in time took to calling the wifemother.There she stayed until an expatriate Britisher made her his mistress. Later in life, she returned to her home in Amakusa.

Prostitution in Korea
1. The phenomenon. -As Japanese emigrants began to move to Korea, they established in their communities structures akin to the licensed brothels at home. Japan formally annexed Korea in 1910, and the new government imposed uniform licensing rules for brothels across all Korea in 1916. It set the minimum age for prostitution at 17 (not 18 as on the Japanese islands), and required regular medical examinations (Fujinaga, 1998(Fujinaga, , 2004Kim and Kim 2018: 18, 21).
Although both Koreans and Japanese could use the new licensing system, the Japanese did so more readily. By 1929, for example, 1,789 Japanese licensed prostitutes worked in Korea but only 1,262 Koreans. The Japanese prostitutes entertained 450,300 guests, where the Koreans entertained 110,700 (252 guests per year for the Japanese prostitute, 88 for the Korean). By 1935 the number of Japanese licensed prostitutes had fallen to 1,778 but the number of Koreans still had risen only to 1,330 (Kim and Kim 2018: 18, 21;Fujinaga, 2004).
Plenty of Korean women worked as prostitutes, but they simply did not work within the licensing structure. In 1935 Korea, government records report that 414 Japanese women worked as bar maids and 4,320 as cabaret workers (both euphemisms for unlicensed prostitutes). Of Korean women, 1,290 worked as barmaids and 6,553 as cabaret workers. 4 2. The contracts. -(a) Prices. To recruit their licensed prostitutes, Korean brothels used indenture contracts much like those in Japan. Prices, however, reflected the lower standard of living in Korea. Over the economy as a whole, from 1910 to 1940 the ratio of Japanese to Korean wages varied from about 2.5 to 1.5. Korean men in the 1930s earned about 1-2 yen per day (Odaka 1975: 150, 153).
Within this Korean market, Japanese prostitutes charged more than Korean prostitutes. Japanese customers were generally wealthier than Korean customers, after all, and Japanese customers tended to prefer Japanese women. By one account, in 1926 Korean prostitutes charged 3 yen for an assignation; Japanese prostitutes in Korea charged 6-7 yen. Customers spent an average 3.9 yen on a visit to a Korean licensed prostitute; they spent an average of 8 yen on a visit to a Japanese licensed prostitute in Korea (Kim and Kim 2018: 26, 89, 96;Nihon yuran, 1932: 461). In one (apparently poorer) Korean community in 1929, Japanese licensed prostitutes generated annual revenues of 1,052 yen; Korean licensed prostitutes generated 361 yen (Nihon, 1994).
The higher Japanese revenues resulted in higher upfront cash payments to the Japanese prostitutes working in Korea than to the Korean prostitutes. One source (see Kim and Kim, 2018: 96) described Korean licensed prostitutes receiving advances of 250-300 yen (and occasionally 400-500 yen) on three year contracts; Japanese licensed prostitutes received 1,000-3,000 yen (note the higher amounts than in Japan). Another source calculated the average upfront payment to Korean licensed prostitutes at 420 yen, while the Japanese licensed prostitute received 1,730 yen (Nihon, 1994: 63).
(b) Contractual term. Consistent with the experience of prostitutes in Japan quitting within six years, Korean licensed prostitutes left the industry by their mid-20s. In one study, 61 percent of Korean licensed prostitutes were 20-25 years old; only 16 percent were over 25 (Kim and Kim, 2018: 97;see Ito, 1931: 172-94). In another, 680 of the 1,101 licensed prostitutes in the Seoul area were age 20-24, but only 273 were 25−29. Of that group of 1,101, 294 were in their fifth year of service; 65 were in their sixth, and 17 in their seventh. On the base population of 1,101, 317 entered in 1924, and 407 quit (Michiya, 1928).
2. Korean prostitution abroad. -Like the Japanese karayuki, young Korean women too travelled abroad. Crucially, Korean women went abroad to work as prostitutes long before several Shanghai brothels became the first licensed "comfort stations" in 1932. The comfort stations, in other words, did not begin the practice of Korean young women working abroad as prostitutes. The young women had been working abroad as prostitutes for decades before.
Already by the 1920s, Korean women were travelling to Manchuria to work as prostitutes (Fujinaga, 1998). In 1929 Korean women worked in Taiwan as licensed or unlicensed prostitutes (Fujinaga, 2001;Taiwan, 1932), and in 1924 67 Korean women worked in Dailin (Fujinaga, 2000: 219). Presumably, some served a Japanese clientele, some served a Korean clientele, and some served a Chinese clientele.
And long after those first comfort stations, Korean women continued to travel abroad to work as unlicensed prostitutes as well -again, for a wide variety of customers. In 1937, for example, the Tianjin immigrants association reported 81 unlicensed prostitutes from Korea. During one month in 1938, 90 Korean women petitioned the (Japanese-controlled) Korean government for permission to travel to the Chinese city of Jinan to work as unlicensed prostitutes (Kitashina, 1938). And while 12 Korean women worked in comfort stations in Shanghai in 1940, 527 worked as unlicensed prostitutes. 5

Recruitment in Japan and Korea
1. Japan. -Many reformers sought to ban prostitution in prewar Japan, but virtually none complained about recruiters abducting young women into brothels. Young women from poor communities routinely left town to work as prostitutes, but they rarely claimed that any recruiter or brothel had forced them to take the job. Neither did many reformers complain that recruiters tricked young women into working for brothels (Senda, 1973: 89). Instead, when Japanese reformers complained about how women had become prostitutes, they complained about the parents: that parents had effectivelÿ soldẗheir daughters into prostitution. They had not wanted to go, some women reported. But their parents had induced them to agree in order to collect the indenture advance.
For the network of overseas comfort stations, the Japanese government drafted recruiting regulations designed to select only prostitutes already in the industry (Gun'ianjo, 1938;Shina, 1938). The government -the regulations imply -realized the political risks it was running. Reformers within Japan had been fighting for decades to ban prostitution. The last thing it needed were accounts of naive young girls duped by mercenary and dishonest recruiters into a multi-year stint in a Shanghai brothel.
To avoid this morass, the Home Ministry issued clear instructions (Gun'ianjo, 1938;Shina, 1938): (a) For women traveling for the purpose of prostitution, approval shall be granted only to those women heading to North and Central China who are currently working as licensed or effective prostitutes, who are 21 years old or older, and who are free of venereal and other infectious diseases . . . (b) When receiving the identification documents detailed in the preceding section, the women should understand that they should immediately return to Japan upon the conclusion of their provisional contract or when that completion is no longer necessary. (c) Women intending to travel for the purpose of prostitution must apply to the police office for their identification documents in person.
The Ministry told recruiters to hire only women who already worked as prostitutes. To insure that the women knew what they were agreeing to do, it told police not to issue travel documents unless each woman applied in person with her contract. And at the time of the interview, it demanded that the police tell each applicant to return immediately when her contract expired. 5 Takei (2012: tab. 6), Zai Jokai (1938,1937).
2. Korea. -Korea had a problem distinct from any in Japan. It had a large corps of professional labor recruiters, and those recruiters had a history of deceptive tactics. In 1935, Korean police records counted 247 Japanese and 2,720 Korean recruiters. To be sure, these men and women (and they included both men and women) recruited workers for factories as well as brothels (Nihon, 1994: 51;Yamashita, 2006: 675). But throughout the prewar decades, newspapers reported recruiter fraud related to the sex industry.
Back in 1918, the Japanese language daily in Seoul (Keijo nippo, 1918;Senda, 1973: 89) complained ofä massive increase in the number of cases where a delinquent entices a woman to Seoul and, after playing all sorts of tricks on her, sells her off to one of the 'dubious restaurants.'Ïn the late 1930s, Korean newspapers reported a ring of 11 recruiters who attracted over 50 young women into prostitution (Toa, 1937). They reported one astonishingly skillful couple who had deceived over 100. Apparently, the couple promised parents that they would find a job for their daughters in Seoul factories, paid the parents 10 or 20 yen, and then forwarded the daughters to overseas brothels for 100 to 1,300 yen each (Toa, 1939;Yamashita, 2006: 675).
Note, however, what this problem was not. It was not that the government -either the Korean or the Japanese government -forced women into prostitution. It was not that the Japanese army worked with fraudulent recruiters. It was not even that recruiters focused on the army's comfort stations. Instead, the problem involved domestic Korean recruiters who had been tricking young women into working at brothels for decades.

Venereal disease
The reams of Japanese government documents about the comfort stations from the 1930s and early 1940s make clear that the government established the institution to fight venereal disease. To be sure, it had other reasons too. It wanted to reduce rapes. And one strange 1939 army document from North China suggested that comfort stations would help the army fight communism within its ranks (Kitashina, 1939). Primarily, however, the military set up the stations to fight venereal disease: by definition, acomfort stationẅas a brothel that had agreed to follow the military's stringent sanitation and contraceptive procedures.
The Japanese military did not need additional prostitutes; it had plenty. Prostitutes have followed armies everywhere, and they followed the Japanese army in Asia. Instead, the Japanese military needed healthy prostitutes. During the army's Siberian expedition in 1918, the commanders had found large numbers of their soldiers disabled by venereal disease. 6 As the army expanded across China in the 1930s, it found that there too the local prostitutes were heavily infected. If its soldiers were going to patronize brothels, it wanted them patronizing brothels that kept the debilitating diseases in check.
To minimize that risk of disease, the army took several steps. It licensed those brothels that agreed to meet its standards -and named themcomfort stations.Ït required prostitutes at the licensed brothels to undergo weekly medical examinations. If they became infected, it banned them from serving customers until fully recovered. It ordered all customers to use condoms (provided free either by the army or by the brothel), and forbad prostitutes from serving anyone who refused to do so. It required all prostitutes and customers to wash with disinfectants immediately after sex. And it banned its soldiers from patronizing any brothels other than the licensed institutions. 7

Contract duration
Although the comfort stations hired their prostitutes on contracts that resembled those used by the Japanese licensed brothels on some dimensions, the differences were important. To leave the countryside for work at a Tokyo brothel, a woman wanted some confidence that she would earn wages high enough to offset the risks and harshness of the job, and the hit to her reputation. To leave for a brothel on the military front, she incurred different vastly greater risks. Most obviously, she faced all the dangers of warwhether fighting, bombing, or the rampant disease on the front. She also faced far more serious risks of brothel non-performance. Should a brothel owner in Tokyo try to cheat on the contract, a prostitute might complain to the police. On the front, she would find no police except those that worked for the army. In Tokyo she might sue the brothel owner in court for nonperformance. On the front, she had no such option. In Tokyo, she could walk out of the brothel, and disappear into the anonymity of the Tokyo metropolis. On the front, she might be able to do this -but it all depended on where more specifically the brothel might be.
Translating the Tokyo brothel contracts to the front, in other words, required changes. The most basic contractual difference: a much shorter contractual term. Reflecting all these risks that followed from the brothel's location on the front, the contracts usually specified only two-year terms. Recall that the Japanese contracts typically provided six-year terms, and the Korean contracts threeyear terms. Some Korean comfort women in Burma worked on contracts as short as six months to a year (e.g., Josei, 1997: 1-19).

Contract prices
For these short but potentially dangerous assignments, the brothels paid (annual) wages much higher than those at the Tokyo brothels. Typically, for the two-year job they paid several hundred yen up-front. Sample contracts for Japanese women recruited to Shanghai comfort stations in 1937 provided advances of 500 to 1,000 yen (Naimusho, 1938). Similarly, Home Ministry documents from 1938 report Japanese women travelling to the Shanghai comfort stations on 600-700 yen advances, with one woman receiving an advance in the 700−800-yen range, and two in the 300−500-yen range (Naimusho, 1938).
Note what this means: in compensation for the much higher risks involved, prostitutes at the comfort stations earned much higher pay. Domestic prostitutes in Korea and Japan already earned considerably more than they would earn in other employment. Recall that those in Japan earned 1000 yen to 1200 yen on six year terms. At the comfort stations prostitutes from Japan earned 600-700 yen on two-year terms.

Contract terms
Some of the other contract terms reflected the greater insecurity on the front as well. Take the 1943 military regulations regarding the stations in Malaya. Women in Japan considering a job in Malaya could reasonably wonder: will I be robbed; when the army retreats will I be able to carry my savings; if I die will my family have any access to my money? In response, the regulations required the brothel to open a (Japanese) postal savings account for each prostitute in her name. It then required the brothel to deposit in the account 3 percent of the gross revenue the prostitute generated. In addition, the brothel was to pay the woman a fraction of the total gross revenue that turned on the amount of her outstanding debt. With 1,500 yen or more outstanding, she was to receive 40 percent of the revenue; if she had less than 1,500 yen, she was to receive 50 percent; and if she had no outstanding debt she was to take 60 percent. Of this share, the brothel was to apply 2/3 toward the remaining debt, and directly pay the prostitute the rest (Maree, 1943;see also U.S. Office, 1944).
Upon completing the contractual term or (if earlier) repaying the loan, the women could go home. A Korean receptionist for comfort stations in Burma and Singapore kept a diary for several years (Choe, 2017a,b). Regularly, comfort women from his brothel completed their terms and returned to their homes. In the course of his research on comfort women, Kako Senda met a veteran who had helped to recruit women from Japan. Obviously, he had selfinterested reasons to say what he said. But when Senda (1973: 26-27) asked him, "were there any women who actually paid back the 1,000 yen [advance] and went free?," "oh, there were," he replied. "There were lots. Among the ones who went with the first regiment, even those who were slowest paid it off in a few months and went free."

Prostitute savings
The amount that a prostitute earned beyond her up-front advance varied. Even by the contractual terms themselves, that amount depended on the revenue a prostitute generated. Scholars routinely suggest that that brothel owners must have cheated their prostitutes -and no doubt some did. People cheat each other in any industry.
Crucially, however, many brothel owners did indeed pay their prostitutes beyond that large up-front advance. The receptionist with the diary noted that the comfort women kept savings accounts. He noted that he regularly deposited money on their behalf in them. And he noted that he regularly sent money back to their homes on their behalf, and received telegrams confirming receipt (KIH, 2016a;Choe, 2017a,b). Indeed, some comfort women earned and saved enough to establish comfort stations of their own (Park, 2014: 111).
Of all the Korean comfort women who left accounts, Mun Okju seems to have done well most flamboyantly. She writes in her memoir (KIH, 2016b): I saved a considerable amount of money from tips. . . . I knew that all the soldiers put their earnings in the saving accounts in the field post office, so I decided to put my money in the saving account. I asked a soldier to make a personal seal and put 500 yen in the account. . . . I became the owner of the savings passbook for the first time in my life. I worked in Daegu as a nanny and a street seller from the childhood but I remained poor no matter how hard I worked. I could not believe that I could have so much money in my saving account. A house in Daegu cost 1,000 yen at the time. I could let my mother have an easy life. I felt very happy and proud. The savings passbook became my treasure. . . . It was fun to go shopping by rickshaw. I can't forget the experience of shopping in a market in Rangoon. There were lots of jewelry shops because many jewels were produced in Burma, and ruby and jade were not expensive. One of my friends collected many jewels. I thought I should have a jewel myself, so I went and bought a diamond. I became a popular woman in Rangoon. There were a lot more officers in Rangoon than near the frontlines, so I was invited to many parties. I sang songs at parties and received lots of tips.
3.6. The closing years of the war The Japanese government mobilized Korean workers most aggressively during the last two years of the war, and scholars have sometimes suggested that those were the years it most aggressively recruited comfort women. In fact, the opposite is true. The closing years were not ones where the government was trying to staff brothels. Those were years it was moving prostitutes out of brothels and into munitions factories.
As the war turned bad for Japan, the military began running out of men. In 1936, 240,000 men served in the army. Once the army invaded China, that number climbed to 950,000 (1937). It hit 3.58 million in 1943, 5.4 million in 1944, and 7.34 million in 1945. Increasingly, the army called up reservists approaching age 40. By the end of the war, 60.9 percent of the men aged 20-40 had served in the military, and 2 million had been killed (Watanabe 2014: 1, 8).
The military was also running out of supplies (see generally Miwa, 2014). As the army called up its 30-something reservists and sent them to the front, it needed others to take their place in the mines and factories. It had not drafted its young Koreans (despite their being Japanese citizens) into the army. By 1944, however, it did begin sending large numbers of Korean men to those mines and factories. Simultaneously, it began sending young unmarried Japanese and Korean women into the factories as well. 8 Brothels were the least of the government's worries. Steadily, brothels and high-end restaurants began to close. The army was shifting all plausible Japanese men from civilian production to the front. To replace them it was moving Korean men to Japan. It was moving both Japanese and Korean women out of homes and inessential jobs and into munitions production. 9 Think Rosy the Riveter in Korea: the Mainichi shimbun newspaper (1944) published a letter from a woman hauling freight in Pusan harbor. Our country needs us, she exclaimed.Just because we're women doesn't mean that we can closet ourselves in our homes.Between the general austerity in the air and the loss of prostitutes to the factories, brothels steadily went out of business. 10

Conclusion
The Japanese army had a problem. It did not lack for brothels. Prostitutes follow armies everywhere, and they had followed the Japanese army in the 1930s and 1940s. The problem was medical: these local prostitutes suffered from very high levels of debilitating venereal disease. If their soldiers were going to frequent brothels, the command at least wanted them in healthy brothels.
Toward that end -not toward better public health but toward maintaining a deadlier military force -the military imported the standard Japanese and Korean licensing system. Brothels and prostitutes registered with it. Designated physicians conducted weekly medical examinations. Brothels required condoms, and prostitutes were told to refuse clients who balked. Both clients and prostitutes were to wash with disinfectant after every encounter.
The contracts themselves followed basic game theoretic principles ofcredible commitments.Brothel owners (not the military) hired the bulk of the new prostitutes, and hired most of them from Japan and Korea. Realizing the incentive brothel owners had to exaggerate their future earnings, women wanted a large portion of their pay upfront. Brothels agreed. Knowing that they were headed for the front, women wanted a maximum service length. Brothels agreed. In turn, realizing the incentive the women had 8 Hatarakeru (1943), Romu (1943), Chosen (1944Chosen ( , 1945, Higuchi (2005: 53). 9 Senso (1943), Hanto (1944). 10 Senso (1943), Hata (1992: 330, 333), Hakken (1943). to shirk within their unmonitored quarters, the brothels wanted terms that gave women an incentive to work hard. The women agreed. Together, the women and brothels concluded indenture contracts that coupled a large advance with one or two year terms. Until the last months of the war, the women served their terms or paid off their debts early, and returned home.