What If We'd Been Watching Savings Instead of Delinquencies?

The conventional story of the Great Financial Crisis runs through housing: subprime mortgages, rising delinquencies, collapsing MBS portfolios, Lehman Brothers. All told through bank-reported data. When mortgage delinquencies spiked in mid-2007, the crisis was already underway. By the time Lehman filed for bankruptcy in September 2008, millions of households had already exhausted their savings, maxed their credit cards, and burned through retirement accounts. We published a comprehensive update of this analysis with current data in What the Savings Rate Told Us Nine Quarters Before the Last Crisis, and track the buffer signals quarterly in State of the Buffer.

Bank-reported delinquency is a lagging indicator. By the time someone misses a mortgage payment, they’ve already fallen off the financial cliff. The distress was there months — or years — earlier. The question is whether we could have seen it.

When we built the American Distress Index with five independent dimensions of household distress, we discovered something unexpected: one component — Buffer Depletion — predicted another — Debt Stress — by more than two years. This isn’t a model. It’s not a forecast. It’s a measurement of what happened, using data that was publicly available at the time.

The Pattern That Preceded the Crisis

We computed cross-correlations between Buffer Depletion (household savings and debt service ratios) and Debt Stress (bank-reported delinquency rates) across the full 2005-2025 time series. The correlation increases with lag. At lag 0 (simultaneous), the correlation is 0.57. At lag 3 quarters (Buffer Depletion leading by 9 months), it rises to 0.69. At lag 4 quarters, it peaks at 0.69.

This is the signature of a leading indicator. If two series just moved together randomly, correlation wouldn’t increase with lag. The fact that Buffer Depletion at time t correlates more strongly with Debt Stress at time t+3 than at time t means the signal is propagating forward in time.

The statistics tell us there’s a relationship. The actual history shows us what it looked like in real time.

Buffer Depletion crossed above +0.5 standard deviations in early 2005. It stayed elevated through the entire 2005-2007 period, peaking above +1.0 by mid-2005. Debt Stress didn’t follow until mid-2007 — nine quarters later. By the time bank call reports showed rising delinquencies, household savings had already been collapsing for more than two years.

The mechanism is straightforward. When savings decline and debt service rises, households lose their cushion. A missed paycheck, a medical bill, or a car repair that would have been absorbed during stable times instead triggers a cascade. Credit cards get maxed. Payments get skipped. The spiral begins. But it doesn’t show up in bank-reported delinquency data until those households have exhausted all other options.

This is why Buffer Depletion leads Debt Stress. Buffer depletion is the condition that makes delinquency inevitable once any shock arrives. It’s not the trigger — it’s the setup.

Because Buffer Depletion is a leading indicator, the American Distress Index weights it at 30% — the largest share of any component. This isn’t arbitrary — it’s statistically justified by the cross-correlation finding. During the 2005-2007 buildup to the crisis, the Buffer Depletion component was elevated well before Debt Stress followed, providing an early warning signal.

When Buffer Depletion is elevated while Debt Stress remains moderate, it means households are losing their cushion faster than they’re falling behind on payments. That pattern preceded the GFC. It’s a pattern worth watching for. If you’re a homeowner watching your own buffer thin, our guide on what to do when you fall behind covers the concrete steps available before delinquency hits.

Unemployment Followed, It Didn’t Lead

The Labor Market component (initial unemployment claims) does not lead Debt Stress. Job losses spiked after delinquencies, not before them. Unemployment claims didn’t cross above baseline until Q4 2008, well after the mortgage crisis was already visible in bank data. The sequence during the GFC was: buffer depletion (2005) → delinquency (2007) → foreclosures (2008) → bank losses (2008) → recession (2008-2009) → unemployment (2008-2009).

This contradicts the common narrative that job losses cause defaults. During the GFC, households were already in distress before the labor market collapsed. The recession amplified the crisis, but it didn’t initiate it. The ADI’s architecture captures this: labor market disruption is one dimension of financial distress, but it’s not the only one, and it’s not always the first to move.

What It Means Now

As of 2025-Q3, Buffer Depletion reads +0.57 — elevated, but not extreme. Household savings remain below the 2015-2024 baseline average, and debt service ratios are moderately elevated. Debt Stress reads +0.19 — rising, but not at crisis levels. The Labor Market reads -0.45 — unemployment claims remain below baseline.

The leading indicator is elevated but not alarming. Household buffers are thinner than the post-GFC baseline, which makes them more vulnerable to shocks. But the shock (job losses) hasn’t arrived. If the Labor Market component turns positive while Buffer Depletion remains elevated — labor market stress arriving while households have no cushion — that’s the pattern that preceded the major default wave of 2008-2009.

The ADI currently reads 57.1 (Elevated). Buffer Depletion is the most elevated component, but the pattern hasn’t reached alarming levels. The index is watching for the pattern to repeat, not predicting that it will.

Why This Matters

The ADI is designed to see distress before it reaches bank call reports. Buffer Depletion is the early warning. When household savings collapse and debt service rises, the pipeline into default is loading — even if delinquencies haven’t spiked yet. The 2005-2007 period proved this. Households were losing their cushion for two years before the crisis became visible in the Debt Stress data.

Policy responses are designed around lagging indicators. When delinquencies spike, the Fed cuts rates. When unemployment rises, Congress extends benefits. But by then, millions of households have already fallen through. A leading indicator framework — tracking buffer depletion before defaults materialize — creates the possibility of earlier intervention.

This isn’t a forecast. It’s a measurement. The ADI tracks what is, not what will be. But it tracks the right things — the things that move first. If the pattern that preceded 2008 begins to repeat, we’ll see it in Buffer Depletion long before it shows up in Debt Stress.

We’re watching.

Explore the data yourself

All data underlying this analysis is available on the American Default homepage. Download the JSON files, verify the cross-correlations, check the methodology. See Methodology for full details on how the ADI warning lights are computed. For the full savings rate and mortgage delinquency data, see our Savings Rate Statistics and Mortgage Delinquency Statistics 2026 roundups. Active structural projections based on validated leading relationships are tracked on the Structural Outlook page.