We examine abrupt deviations in measured technology-growth residuals using Penn World Table 11.0 data for 180 countries over 1950-2015. We adjust output growth for capital-stock and population growth and fit a Bayesian three-state Jump-AR model separately to each country. We classify each observation by the state with the highest posterior probability. Crisis dates and country groups do not enter estimation. Negative jump candidates are about twice as common as positive candidates and remain more common when periods, capital shares, priors, or classification thresholds change. Negative candidates cluster in transition economies in the early 1990s. Over the full evaluation periods, negative shares are 6.3% in 29 transition economies and 3.0% elsewhere. These comparisons are unadjusted and exploratory, and do not identify a causal effect of institutional transition. The residual can also reflect labor use and capacity utilization. Simulations show both missed jumps and false detections. The results document asymmetric, concentrated deviations in input-adjusted growth.