Abstract: This paper compares the eight most widely used methods for estimating the U.S. output gap, applying each to both revised and real-time quarterly data from 1980 onward. The resulting gap estimates disagree markedly in level, volatility, and cyclical timing, and these differences are compounded by data revisions. Some methods, particularly the refined Beveridge-Nelson decomposition and Hamilton filter, maintain strong agreement between their real-time and final estimates, while deterministic trend approaches show large and persistent revisions. The disagreement carries through to two empirical applications. In recession prediction, the HP filter provides the most accurate forecasts with revised data at longer horizons, but the refined Beveridge-Nelson is a more reliable real-time indicator at shorter horizons. In output growth forecasting, the HP filter again leads with revised data, but the refined Beveridge-Nelson is the most accurate individual method in real time. Forecast combinations built on Bates-Granger weighting and partially-egalitarian LASSO selection offer stable performance across both vintages.
Abstract:Whether financial development promotes industrial innovation depends not just on how developed a country’s financial system is, but on which dimensions of that system are well developed. This paper examines how depth, access, and ef- ficiency of both financial institutions and financial markets shape R&D investment across industries that differ in their reliance on external finance. Using industry- level data from the ISIC Rev. 4 classification across 18 OECD countries from 1995 to 2019, and drawing on the IMF’s multidimensional Financial Development In- dex, we analyze how country-level financial development measures interact with an industry-level external finance dependence measure to influence R&D intensity measured relative to output and value added. Our findings show that the overall level of financial development matters primarily through depth. In particular, the depth of financial institutions and, to a lesser extent, the depth of financial mar- kets significantly raise R&D intensity in industries that depend more heavily on external funding. Measures of access and efficiency display little systematic effect. The results are strongest within manufacturing industries, where innovation activ- ity is concentrated. These findings highlight the importance of financial structure and, in particular, the scale and capacity of financial intermediation, in shaping the allocation of innovative investment across industries.