Econometric Advisors’ U.S. macro model is a compact model that captures the main high-level features and interactions of the U.S. economy, including the macroeconomic variables needed to drive EA’s metro-level real estate forecasts. The core model covers GDP, employment, inflation, interest rates and corporate bond spreads (as a proxy for financial stress). Additional equations drive forecasts of imports, stock-building and household disposable incomes.
No attempt is made to explicitly model variables like fiscal policy, or the world economic environment. This is not because we do not think they are important; quite the contrary. Rather, we prefer to treat them as shocks to GDP relative to trend, on which we make an informed judgement.
Using a Hodrick-Prescott Filter, we split GDP into “trend” and “cyclical” (or GDP relative to trend) components. The “trend” is exogenous and is set to be broadly in line with consensus forecasts of U.S. long-term economic growth. GDP relative to trend acts as a measure of the output gap in the inflation equation and others.
The change in GDP relative to trend is a function of:
The expressions in brackets show the impact (positive or negative) of the variable on the dependent variable. The negative relationship between the change in GDP relative to trend and the lagged value of GDP relative to trend means that GDP will, with all other things held constant, tend to move back towards trend. The larger the deviation of GDP from trend, the faster is the suggested movement back toward trend.
The corporate bond spread is used as a measure of financial stress. We continue to experiment with alternative measures, but corporate bond spreads perform quite well for our purpose. The spread is expressed relative to a 16-quarter moving average to make the series stationary (there is a marked positive trend in corporate bond spreads before 2003).
The presence of a negative coefficient on lagged GDP relative to trend means that the relationship tends to be “trend-reverting” but it does not mean that GDP will always revert to trend. If GDP is above trend and there is decline in corporate bond spreads (financial stress), then GDP can still move further above trend. This effect will cease when the decline in financial stress slows enough, and when financial stress begins to rise, GDP can revert rapidly towards trend. If GDP is below trend, the opposite applies.
Core PCE inflation relative to trend is determined by GDP relative to trend and the change in GDP relative to trend. Headline CPI inflation depends on Core PCE Inflation and changes in oil prices.
Both trend Core PCE and headline CPI inflation are exogenous and fall gradually over time before leveling off at the beginning of 1999.
10-year Treasury Bond yields move one-for-one with Core PCE inflation over the long term. These are also affected by:
This specification is broadly in line with Barkham and Blake (2018).
The short-term policy rate (the effective Fed Funds Rate) is driven by 10-year Treasury Bond yields (a one-for-one relationship over the long-term) and the level of GDP relative to trend.
AAA corporate bonds spreads over 10-year Treasury Bond yields depend on:
Employment is determined by:
The Standard Scenarios
The standard EA Upside and Downside scenarios are generated by making exogenous overrides to GDP and then allowing this to feed through to the other variables in the model. These overrides are made to levels of GDP, rather than to growth rates. The current overrides are:
The Upside Scenario: GDP relative to base
The Downside scenario overrides are simply the opposite of the Upside overrides.
The Severe Downside Scenario
Our Severe Downside scenario is meant as a proxy for the Federal Reserve’s Severely Adverse supervisory scenario under its Comprehensive Capital Analysis and Review (CCAR) program. CCAR is meant to evaluate the capital planning processes and capital adequacy of U.S. banks under stressful macroeconomic scenarios. This scenario is generated by overriding EA’s baseline GDP with the Fed’s forecasts for GDP under their Severely Adverse scenario, which is then fed through to the other variables in the model. As with the Upside and Downside scenarios, the overrides are to GDP levels, rather than growth rates. The estimates for the Fed’s Severely Adverse scenario are released annually in February and can be found here.
The Main Relationships in the EA U.S. Macro Model