Third-quarter US GDP is expected to increase 1.9% (real seasonally adjusted annual rate), according to The Capital Spectator’s average econometric nowcast. This is the initial estimate that uses limited Q3 data and so the projection is a preliminary review that will be revised several times as new economic data is published. The final nowcast will be published shortly before the official Q3 GDP report, which will arrive on October 30, when the Bureau of Economic Analysis (BEA) releases the first of three estimates.
In 2013’s second quarter, GDP increased 1.7%, BEA reported in its July 31 “advance” estimate. The modest rate of expansion for the second quarter fell short of the 2.0% gain projected in the final Q2 nowcast that preceded the government’s report.
Here’s a review of the current Q3 nowcast in context with actual data and several recent forecasts from other sources:
Here’s how the individual nowcasts compare:
Finally, here’s a brief profile for each of The Capital Spectator’s nowcast methodologies:
R-4: This estimate is based on a multiple regression in R of historical GDP data vs. quarterly changes for four key economic indicators: real personal consumption expenditures (or real retail sales for the current month until the PCE report is published), real personal income less government transfers, industrial production, and private non-farm payrolls. The model estimates the statistical relationships from the early 1970s to the present. The estimates are revised as new data is published.
R-10: This model also uses a multiple regression framework based on numbers dating to the early 1970s and updates the estimates as new data arrives. The methodology is identical to the 4-factor model above, except that R-10 uses additional factors—10 in all—to nowcast GDP. In addition to the data quartet in the 4-factor model, the 10-factor nowcast also incorporates the following six series:
• ISM Manufacturing PMI Composite Index
• housing starts
• initial jobless claims
• the stock market (S&P 500)
• crude oil prices (spot price for West Texas Intermediate)
• the Treasury yield curve spread (10-year Note less 3-month T-bill)
ARIMA GDP: The econometric engine for this nowcast is known as an autoregressive integrated moving average. This ARIMA model uses GDP’s history, dating from the early 1970s to the present, for anticipating the target quarter’s change. As the historical GDP data is revised, so too is the nowcast, which is calculated in R via the “forecast” package, which optimizes the parameters based on the data set’s historical record.
ARIMA R-4: This model combines ARIMA estimates with regression anlaysis to project GDP data. The ARIMA 4 model analyzes four historical data sets: real personal consumption expenditures, real personal income less government transfers, industrial production, and private non-farm payrolls. This model uses the historical relationships between those indicators and GDP for projections by filling in the missing data points in the current quarter with ARIMA estimates. As the indicators are updated, actual data replaces the ARIMA estimates and the nowcast is recalculated.
VAR 4: This vector autoregression model uses four data series in search of interdependent relationships for estimating GDP. The historical data sets in the R-4 and ARIMA 4 models above are also used in VAR-4, albeit with a different econometric engine. As new data is published, so too is the VAR-4 nowcast. The data sets range from the early 1970s to the present, using the “vars” package in R to crunch the numbers.
ARIMA R-NIPA: The model uses an autoregressive integrated moving average to estimate future values of GDP based on the datasets of four primary categories of the national income and product accounts (NIPA): personal consumption expenditures, gross private domestic investment, net exports of goods and services, and government consumption expenditures and gross investment. The model uses historical data from the early 1970s to the present for anticipating the target quarter’s change. As the historical numbers are revised, so too is the estimate, which is calculated in R via the “forecast” package, which optimizes the parameters based on the data set’s historical record.