How to set defense spending when inflation is volatile

This month, as Congress starts crafting the fiscal 2023 Defense Department budget, policymakers must decide how to deal with inflation’s erosion of the Pentagon’s purchasing power. A growing number of legislators have called for increasing the budget above the Biden administration’s $773 billion request to compensate for high inflation, among other rationales. Many policymakers have shied away from proposing dollar amounts, instead endorsing specific levels of real growth — the year-to-year percentage change in spending after accounting for inflation, such that 0% means spending grows at the forecast inflation rate.

The trouble is that when inflation is changing quickly, like today, real growth resembles Potter Stewart’s definition of obscenity : “I know it when I see it.”

During periods with volatile inflation, policymakers have different outlooks on the accuracy of the inflation forecast. As a result, they also have different standards for what qualifies as “real” growth. To make sound judgments about the FY23 budget, policymakers need to consider not only their preferences about real growth but also their beliefs about future inflation. Combining a little history and a little math illustrates the wide-ranging options available to Congress as it sets this year’s defense budget.

The FY23 Pentagon budget request assumed inflation would average 2.2% in 2023. This rate represents the gross domestic product price index that anchors Defense Department budget planning, as mandated by the Office of Management and Budget. Critics have cited alternative indices reporting higher rates, such as the consumer price index, to suggest that the assumed inflation rate is potentially too low. As a practical matter, the GDP price index should continue to anchor budget planning because it performs well relative to alternatives, according to technical studies .

All the attention paid to alternative indices has distracted from a vitally important question: How accurate have GDP price index forecasts been in the past?

By surveying the range of prediction error in previous forecasts — specifically forecasts looking one year ahead — we can benchmark the range of potential error in the current forecast. This benchmark could prove inaccurate if something unprecedented happened with inflation. However, including periods with extreme inflation outcomes, such as the late 1970s, lowers that risk.

From 1977 to 2020, the period for which data exists in Defense Department reference volumes, forecasts looking one year ahead predicted the GDP price index inflation rate reasonably well. The U.S. government overestimated inflation (meaning the forecast exceeded the rate) more often than it underestimated inflation, with 24 overestimates, 16 underestimates and four bull’s-eyes. The maximum underestimate forecast the inflation rate 2.5 percentage points too low (1979), whereas the maximum overestimate forecast the inflation rate 2 percentage points too high (1986).

These historical outcomes illustrate how policymakers might, depending on their outlook on future inflation, adjust the FY23 budget. For example, policymakers deeply concerned that the assumed 2023 inflation rate of 2.2% is too low might consider increasing the budget by an additional 2.5 percentage points, matching the maximum underestimate from 1979. That adjustment would produce a $792 billion budget, representing 0% real growth with our higher assumed inflation rate. Policymakers could then increase spending further, if desired, based on their real growth preferences. In this scenario, 5% real growth would yield a $830 billion budget, nearly $60 billion above the request.

Policymakers skeptical about this worst-case thinking might instead consider increasing the budget by an additional 0.3 percentage points, a less extreme key value from the historical data that falls closer to the official forecast. That adjustment would result in a $776 billion budget, again representing 0% real growth with our assumed inflation rate. In this alternative scenario, 5% real growth would produce a $814 billion budget, an increase of $41 billion above the request that falls exactly halfway in between the increases recommended by the Senate and House Armed Services committees in June.

In weighing these options, policymakers should remember that inflation factors are not the only way to judge the budget’s sufficiency. Equally important is the question of how the U.S. forces supported by the budget would fare against enemy forces in likely conflict scenarios. If we believe that the forces and budget provide roughly the right level of defense, then estimating inflation’s deleterious effects matters greatly. By contrast, if we believe that the forces and budget do not provide the right defense, then devoting disproportionate attention to inflation distracts from larger problems.

In sum, inflation can, alongside other factors, rightfully inform judgments about the appropriate level of defense spending.

Real growth alone does not provide an adequate standard for setting defense spending given today’s uncertainty about tomorrow’s inflation. Specific real-growth percentages can yield different budgets depending on one’s outlook on future inflation. For this reason, legislators need to define, even if privately, the inflationary losses in defense buying power that they seek to avoid or accept in the FY23 budget. The illustrative options above provide one way to think about this central issue.

Travis Sharp is a fellow and the director of defense budget studies at the Center for Strategic and Budgetary Assessments. This commentary is adapted from his forthcoming CSBA report, “How I Learned to Start Worrying and Hate Real Growth: Analysis of the 2023 Defense Budget Request.”

本月，随着国会开始制定2023财年国防部预算，决策者必须决定如何应对通货膨胀对五角大楼购买力的侵蚀。越来越多的议员呼吁增加预算，使其超过拜登政府提出的7730亿美元，以弥补高通胀等因素的影响。许多决策者不愿提出具体的金额，而是支持具体的实际增长水平——即扣除通货膨胀因素后，支出同比的百分比变化，0%意味着支出以预测的通货膨胀率增长。

问题在于，当通货膨胀像今天这样快速变化时，实际增长就像波特·斯图尔特对淫秽的定义一样：“我一眼就能认出来。”

在通胀波动时期，政策制定者对通胀预测的准确性持有不同的看法。因此，他们对“实际”增长的定义也存在差异。为了对2023财年预算做出合理的判断，政策制定者不仅需要考虑他们对实际增长的偏好，还需要考虑他们对未来通胀的预期。结合一些历史背景和一些数学计算，我们可以了解国会在制定今年国防预算时所面临的多种选择。

2023财年五角大楼预算申请假设2023年平均通胀率为2.2%。这一通胀率代表了国内生产总值（GDP）价格指数，该指数是国防部预算规划的基准，由管理和预算办公室（OMB）规定。批评人士援引其他指数（例如消费者价格指数）的数据，指出GDP价格指数的通胀率更高，并以此暗示五角大楼假设的通胀率可能过低。但实际上，根据技术研究，GDP价格指数相对于其他基准指数表现良好，因此仍应继续作为预算规划的基准。

人们对其他指数的关注分散了人们对一个至关重要的问题的注意力：过去 GDP 价格指数预测的准确性如何？

通过考察以往预测（特别是展望未来一年的预测）的误差范围，我们可以对当前预测的潜在误差范围进行基准测试。如果通胀出现前所未有的波动，这个基准可能并不准确。然而，将通胀极端时期（例如20世纪70年代末）纳入考量，可以降低这种风险。

从1977年到2020年（国防部参考资料中有相关数据），对未来一年GDP价格指数通胀率的预测相当准确。美国政府高估通胀率（即预测值高于实际值）的次数多于低估通胀率的次数，其中高估24次，低估16次，准确预测4次。最大低估值是预测值比实际值低2.5个百分点（1979年），而最大高估值是预测值比实际值高2个百分点（1986年）。

这些历史结果表明，政策制定者可以根据对未来通胀的预期来调整2023财年的预算。例如，如果政策制定者非常担心假设的2023年通胀率2.2%过低，他们可能会考虑将预算增加2.5个百分点，与1979年的最大低估幅度持平。这一调整将使预算达到7920亿美元，相当于在我们假设的较高通胀率下实际增长率为0%。政策制定者可以根据其对实际增长的偏好，酌情进一步增加支出。在这种情况下，5%的实际增长率将使预算达到8300亿美元，比预算申请高出近600亿美元。

对这种最坏情况设想持​​怀疑态度的政策制定者可能会考虑将预算增加0.3个百分点，这是一个相对温和的关键数值，更接近历史数据和官方预测。这一调整将使预算达到7760亿美元，同样假设实际增长率为0%（基于我们假设的通胀率）。在另一种情景下，5%的实际增长率将使预算达到8140亿美元，比预算请求增加410亿美元，恰好位于参众两院军事委员会6月份建议增幅的中间值。

在权衡这些选项时，政策制定者应记住，通货膨胀因素并非衡量预算充足性的唯一标准。同样重要的是，在可能的冲突情境中，由该预算支持的美军能否有效对抗敌军。如果我们认为目前的兵力和预算能够提供大致合适的防御水平，那么评估通货膨胀的不利影响就至关重要。相反，如果我们认为目前的兵力和预算无法提供足够的防御，那么过度关注通货膨胀就会分散人们对更重大问题的注意力。

总之，通货膨胀可以与其他因素一起，为判断适当的国防开支水平提供合理的依据。

鉴于当前通胀形势的不确定性，仅凭实际增长率不足以作为制定国防开支的充分标准。具体的实际增长率会因对未来通胀预期不同而导致不同的预算。因此，立法者需要明确（即使是私下讨论）他们在2023财年预算中希望避免或接受的国防购买力通胀损失。以上示例提供了一种思考这一核心问题的思路。

特拉维斯·夏普是战略与预算评估中心的研究员兼国防预算研究主任。本文改编自他即将发表的CSBA报告《我如何开始担忧并憎恨真正的增长：2023年国防预算申请分析》。