Orply.

American Relief Lowered Food Prices and Improved Outcomes During Russia’s 1921–22 Famine

Natalya NaumenkoVolha CharnyshHoover InstitutionWednesday, November 18, 202612 min read

A study of the American Relief Administration’s work during the 1921–22 famine in Soviet Russia finds that aid followed food scarcity rather than local political or ethnic characteristics. Natalya Naumenko, Volha Charnysh and Andrei Markevich use newly assembled provincial data to argue that relief lowered staple-food prices, increased reported calorie intake and improved some health and demographic outcomes. Their estimate suggests the 1926 population was 4.4 million higher than it would have been without ARA relief, though they caution that this figure is indirect and depends on assumptions.

Aid followed famine severity, not local politics

The American Relief Administration’s operation in Soviet Russia was an unusually hard test of whether large-scale famine relief could work. More than 20 million people were at risk of starvation after civil war, disrupted trade and a severe drought. The Bolshevik government was hostile to the Americans, transport had collapsed, and the ARA had only about 200 Americans on the ground. At its peak, the operation fed 10 million people a day.

The study’s central finding is that aid reached areas with greater need without being allocated according to local political or ethnic characteristics. Natalya Naumenko and her coauthors assembled monthly ARA feeding reports from the Hoover Institution Archives and combined them with newly digitized Soviet data on agricultural conditions, prices, calorie consumption, disease, population and political outcomes. Their panel covers 82 provinces, representing 88% of the 1913 population.

The allocation question matters because aid went to the places hardest hit by famine. A simple comparison between more aid and subsequent outcomes could therefore make aid look ineffective: the worst-affected areas might remain worse off even if the relief helped. The researchers first ask whether the distribution itself tracked need, and whether politics or practical access also shaped it.

In unadjusted comparisons, the patterns are not simple. Provinces with more non-Russians, for example, appear to receive more aid. But once the researchers account for famine severity—using indicators including drought losses, harvests and grain procurement—the political and ethnic measures no longer explain the distribution. They find no evidence that aid was directed toward either Bolshevik-supporting or anti-Bolshevik areas, or systematically toward ethnic minorities. Nor do they find that transportation measures predict allocation.

That last result surprised an audience member who expected broken railways to affect where supplies could go. Naumenko suggested the ARA’s flexibility may help explain it: its workers used cars, boats, rail and, in some places, camels. The operation was difficult, but the researchers interpret the lack of a transport relationship as evidence that staff worked around infrastructure failures.

The presentation’s maps trace the expansion of feeding alongside investigators’ travels. ARA staff initially expected an urban food crisis, then found that peasants in grain-producing areas were in acute distress. Their monthly reports and local assessments guided the expansion from the Middle Volga toward southern Russia and Ukraine.

The arrangement gave Americans unusual influence over delivery. They traveled to assess conditions, decided where and how much aid to send, and used local committees to identify recipients and run kitchens. At the operation’s peak, they employed roughly 120,000 local staff, whom they selected themselves. They sought people with useful skills, including teachers and priests, rather than simply relying on Communist Party recommendations. Naumenko noted that secret-police informers may nevertheless have been present among ARA workers.

Bolshevik officials tried to restrict American movement, and the secret police intimidated Russian staff. The ARA also faced shortages of railcars, fuel and storage, along with armed bands and small-scale theft. The researchers argue that the structure of the operation helped limit diversion: aid arrived in monthly installments, allowing the ARA to threaten to suspend deliveries if the government interfered. Food for children was distributed through ARA-run soup kitchens; adults later received dry rations.

A question from the audience raised the gap between these findings and accounts of famine aid that failed to reach people in need. Naumenko agreed that the famine itself had political causes, including forced grain procurement that reduced incentives to plant. Her distinction was between those causes and the later allocation of outside assistance. In this case, Americans insisted on freedom of movement and on deciding distribution themselves, rather than handing allocation to the government or government-controlled organizations. She presented that control, and the ability to stop future deliveries, as a possible reason aid followed need. The researchers believe this mattered, but they cannot test the mechanism directly: the same system operated across the regions they study.

The famine joined policy failures to a severe harvest shock

The famine followed years of disruption. During the Civil War and the Bolsheviks’ policy of war communism, the government forcibly extracted grain from producing regions to feed workers and cities. Grain trade was banned, interrupting the normal movement of food between the South and the North. Peasants responded to the extraction by planting less: the presentation describes a decline in sown area of about one-third. In spring 1921 the government introduced the New Economic Policy, replacing forced requisitioning with a tax and allowing market activity in the countryside. But the change came as drought, following another drought the year before, sharply reduced the harvest.

The combination produced a counterintuitive geography. The southern Black Soil region was a major grain-producing area, yet it was among the hardest hit. In ordinary shortage situations, a producing area might be expected to have more food. Here, years of extraction had helped reduce production, and drought struck the region that had been expected to supply others. The presenters emphasized that the study does not assign a precise share of the famine to government policy versus weather: both contributed.

The Bolsheviks initially resisted acknowledging the famine. As conditions worsened and unrest became a concern, the writer Maxim Gorky appealed publicly for help. Herbert Hoover, then leading the ARA’s relief work in Europe, responded. After negotiations, the Americans secured freedom of movement and the authority to determine where aid went. They first planned to feed one million children. As staff traveled and found the crisis more extensive than anticipated, the effort expanded to adults, with additional American government funding.

The operation ran from 1921 to 1923. Its peak came in July 1922, just before the new harvest improved food availability; after the harvest, the program scaled down and shifted toward hospitals and orphanages. The ARA distributed 718,770 tons of food, equivalent to 61.4 million person-months. At one point it supplied about 390 grams per person per day, which the presentation described as roughly 30% of the average intake at the time. The operation ended in May 1923.

The path of expansion was shaped by what the Americans learned on the ground. They initially expected an urban shortage, then found that peasants in the grain-producing regions were in acute distress. They began near Kazan and the Middle Volga, expanded south, and later reached Ukraine, where the government had claimed there was no famine. ARA workers encountered the crisis there while traveling for a separate relief effort. Local committees estimated how many people needed feeding; reports went to Moscow and then New York, and supplies followed those assessments.

Monthly data help separate relief from worsening shortages

The researchers’ main economic challenge is that aid was not randomly assigned. As Volha Charnysh explained, more aid went to places where shortages were worse. Regressing outcomes directly on aid would therefore confound the effect of relief with the severity of the crisis.

Their approach uses monthly data and changes over time. The underlying shortages, they argue, changed more slowly than the rapid increases in ARA feeding. They estimate whether month-to-month increases in feeding are associated with subsequent changes in outcomes, while controlling for agricultural conditions and other factors. They also separate the period before the August 1922 harvest from the period after it, when food scarcity eased. If unmeasured shortages still worsened in places where feeding increased, Charnysh said, that would tend to bias the estimates against finding a benefit.

Rye flour is the primary price measure. It was a staple of peasant diets: the presenters said about 60% of what peasants ate was bread, and 80% of that bread was rye. The price series uses local bazaar prices, adjusted for inflation using a basket of consumer goods. These were market prices, not government-controlled prices. The audience asked whether the figures could be trusted amid a collapsed economy and semi-legal markets. Charnysh said the prices were collected by Soviet statisticians and published; Naumenko pointed to the series’ broader pattern as evidence that the data captured the shift from high prices in the North during the Civil War to a spike in the drought-affected South.

9%
lower rye-flour prices for each 10-percentage-point increase in the population fed, before the August 1922 harvest

Before the 1922 harvest, a 10-percentage-point increase in the population fed by the ARA was associated with a 9% decline in rye-flour prices in the following month. The relationship was statistically significant and remained similar when prices were not adjusted for inflation or when the researchers added controls for other food sources, livestock and refugees. After the harvest, they did not find a comparable price effect.

The researchers also tested whether the ARA simply increased feeding in places where prices were already falling. They found no evidence that it targeted provinces based on prior price improvements. In event-study estimates using several methods designed for staggered adoption, prices fell by about the third month after feeding began. The presenters also reported that wheat, potatoes and oats became cheaper, while beef and cotton cloth did not. Because the ARA supplied grain but not meat or cloth, they treated those unchanged prices as a check against the explanation that all prices were falling because markets were generally improving.

The price findings were accompanied by changes in reported food consumption. Soviet statisticians surveyed households about what they had eaten during the previous week and converted consumption into calories per adult male. The records are available only as province-level averages, not individual responses, but they are also divided by farm size. Where ARA feeding increased, measured calories rose; the gains were largest among landless and small-farm peasants. The researchers interpreted that pattern as evidence that the poorest benefited most.

An audience member questioned whether the calorie levels in the historical data were plausible. Naumenko said scholars had debated why reported intake was so high compared with estimates for British workers. One possible explanation was the physical demands of peasant labor and the cold climate; another was that Soviet estimates of physiological needs or the conversion of food into calories differed. The presenters did not resolve the issue. Their claim rests on changes within the series, rather than on treating the reported levels as an uncontested measure.

The disease findings were narrower. ARA feeding was associated with fewer cases of relapsing fever, which is louse-borne, but not with fewer cases of waterborne typhoid. The presenters noted that the ARA also supplied medical goods and carried out delousing campaigns in orphanages and refugee camps; in some cases, food was tied to participation in delousing. The measured feeding variable may therefore track a wider package of relief activity, not food alone.

The population estimate is consequential but indirect

The researchers did not have mortality data. Instead, they used birth cohorts recorded in the 1926 census as an indicator of demographic consequences, and compared the relationship between feeding and cohort sizes across provinces. More feeding was associated with larger subsequent birth cohorts, particularly in rural areas. Naumenko cautioned that the estimate was small in magnitude and that cohort size is not a direct measure of deaths prevented.

They also examined provincial population in 1926. More ARA feeding during the first phase of the operation was associated with a larger rural population; feeding after the 1922 harvest was not. This measure combines several possible effects: mortality, births and migration. The researchers cannot isolate each one.

4.4 million
estimated additional people in the 1926 population when ARA feeding is compared with a model setting feeding to zero

Their back-of-the-envelope estimate uses 43 provinces. In that sample, their model predicts a 1926 population of about 74.8 million, close to the reported 74.9 million. Setting ARA feeding to zero in the calculation lowers the predicted population to 70.4 million—a difference of roughly 4.4 million people. Using an estimate that about $49 million, or 80% of ARA spending, went to the relevant relief, the presenters calculated a cost of about $11 per additional person in 1926, or $213 in today’s money.

Naumenko repeatedly urged caution: the calculation depends on assumptions, covers only 43 of the 82 provinces and includes differences in migration as well as survival and births. She presented it as an indication that the effect may have been large, not as a direct count of lives saved.

An audience member argued that the estimate might understate the ARA’s role because it focused on feeding and not seed. In December 1921, Hoover had reportedly made an additional congressional appropriation for adult feeding conditional on the Soviet government contributing $10 million in gold toward seed purchases. The ARA supplied seed, much of it wheat seed from northern U.S. states. Timing mattered: food had to reach famine areas in time for spring planting, but not so early that people would eat the seed. The audience member argued that the resulting 1922 harvest helped break the famine and should be part of any assessment of the operation.

Naumenko agreed that the ARA provided seed and said the researchers had not assembled comparable province-level seed data. They treat feeding as a proxy for the broader ARA presence; provinces receiving more food aid may also have received more seed, medical supplies and help with spring sowing. The study separately finds faster agricultural recovery where feeding was greater, with the difference in sown area fading by 1925. But the presenters cannot attribute that recovery to food, seed or other correlated assistance individually.

The timing of Soviet policy also entered the discussion. The government stopped grain requisitions in March 1921, before ARA feeding began in September. Charnysh said the analysis controls for previous procurement and weather; Naumenko emphasized that requisitions had stopped everywhere before the Americans arrived. The distinction matters because expectations about whether peasants could keep what they planted may have influenced recovery independently of relief.

Relief improved material conditions without changing measured political behavior

The ARA and the Bolsheviks both had political hopes for the operation, though in opposite directions. American officials hoped that relief might improve attitudes toward the United States and weaken support for the Bolsheviks. The Soviet government hoped to claim credit for the assistance and bolster its legitimacy. The researchers find no evidence that feeding changed the political outcomes they measure.

They examined peasant unrest using secret-police reports that recorded terms associated with revolt, uprisings, banditry, desertion and discontent. Feeding was not associated with either an increase or a decrease in those mentions. They also found no effect on turnout in elections to the Soviets or on the share of people disenfranchised. A dataset of court convictions for counter-revolutionary activity was still being cleaned; preliminary results showed no effect. Finally, they used first names in a database of 38 million Second World War soldiers as a rough proxy for parents’ political preferences. Revolutionary names did not decline where ARA feeding was greater. The presenters described these results as preliminary.

An audience member raised a possible limitation: secret-police monitoring might have been stronger in places receiving more aid, and people might have concealed their views. Naumenko said the researchers also examined unrest in an event-study framework and found no effect.

The presenters also resisted a broader extrapolation to sustained aid. One audience member distinguished the ARA’s time-limited intervention—during a crisis that eased with the next harvest—from repeated relief that might weaken local production. The presenters agreed that their findings did not establish what would happen under year-after-year food assistance. Naumenko stressed that the ARA sent supplies month by month and could stop them if the government interfered; she saw donor control over distribution as central to the case.

The study’s conclusion is correspondingly specific. In this operation, large-scale aid reached areas with greater famine need, lowered staple-food prices, raised reported calorie consumption, reduced relapsing fever and was associated with better demographic and agricultural outcomes. It did not produce measurable changes in unrest, electoral participation, disenfranchisement, counter-revolutionary convictions or revolutionary naming patterns. Relief appears to have helped people materially without delivering the political shift its American sponsors hoped for.

The frontier, in your inbox tomorrow at 08:00.

Sign up free. Pick the industry Briefs you want. Tomorrow morning, they land. No credit card.

Sign up free