Search Results

You are looking at 1 - 10 of 19 items for :

  • "expectation-maximization algorithm" x
Clear All

automation are also highest for less educated workers and for clerical and sales workers ( Figure 1 ). Figure 1. Gender Gaps in High Risk of Automation by Generation and Educational Level (Difference in automatability between females and males) Sources: Organisation for Economic Co-operation and Development, Programme for the International Assessment of Adult Competencies; and IMF staff estimates. Note: The probability of automation is estimated using an expectation-maximization algorithm that relates individual characteristics (age, education, and training

Mr. Eduard J Bomhoff

compute adjustments to the three unknown variance terms. I use the expectation maximization algorithm, described by Dempster, Laird, and Rubin ( 1977 ) and adapted to the case here by Shumway and Stoffer ( 1982 ). 10 Then, the separate forward and backward Kalman filters (blocks 1 and 2) are run again in order to prepare inputs for the Kalman smoother in the next iteration. This process stops when the estimated values of the unknown parameters have converged to their optimal values. 11 Finally, the fifth block of the algorithm is applied just once. It starts with the

Mr. Eduard J Bomhoff

states, one has to start both filters with an uninformative prior distribution for the covariance matrix of the states. With this initialization, the smoothing algorithm will reproduce the OLS variance matrix of the parameters (and the OLS residuals) in the special case that all the states are constant and correspond to OLS parameters. The fourth block of the algorithm uses the results of the Kalman smoother to compute adjustments to the unknown parameters (the four unknown variance terms), based on the Expectation Maximization algorithm, described by Dempster et al

Mariya Brussevich, Ms. Era Dabla-Norris, and Salma Khalid

-workability: occupation-level classification of feasibility of working from home derived by Dingel and Neiman (2020) for the US and individual-level data from the OECD’s Programme for the International Assessment of Adult Competencies (PIAAC). The latter has the advantage of measuring task or skill content at the worker level for a large sample of countries. Our estimation approach relies on an Expectation Maximization algorithm to map occupation-level measures of the feasibility of working at home to individual-level observations in the PIAAC dataset and derive predicted tele

Mariya Brussevich, Ms. Era Dabla-Norris, and Salma Khalid
Using individual level data on task composition at work for 30 advanced and emerging economies, we find that women, on average, perform more routine tasks than men?tasks that are more prone to automation. To quantify the impact on jobs, we relate data on task composition at work to occupation level estimates of probability of automation, controlling for a rich set of individual characteristics (e.g., education, age, literacy and numeracy skills). Our results indicate that female workers are at a significantly higher risk for displacement by automation than male workers, with 11 percent of the female workforce at high risk of being automated given the current state of technology, albeit with significant cross-country heterogeneity. The probability of automation is lower for younger cohorts of women, and for those in managerial positions.
Mariya Brussevich, Ms. Era Dabla-Norris, and Salma Khalid

mapped to multiple occupations in the Frey and Osbourne (2017) estimates. Therefore, in the spirit of Arntz, Gregory, and Zierahn (2017) we use the Expectation Maximization algorithm and estimate an individual-level regression: p i j = Σ n = 1 N β n X i n + ϵ i j , ( 3 ) where i denotes individuals, j denotes duplicates of these individuals when multiple probabilities are associated with one individual due to

Mr. Michael Weber and Ms. Michaela Denk

( Rubin, 1996 ). Iterative imputation procedures based on the EM (expectation maximization) algorithm ( Dempster, Laird, Rubin, 1977 ) are also closely related to multiple imputation. The EM algorithm consists of (i) an expectation (E) step that replaces missing value by expected values of the distribution based on estimated distribution parameters and (ii) a maximization (M) step that estimates the parameters of the distribution by maximizing the data log-likelihood function. This means that first missing values are replaced by some initial estimates (may be taken

Mariya Brussevich, Ms. Era Dabla-Norris, and Salma Khalid
Lockdowns imposed around the world to contain the spread of the COVID-19 pandemic are having a differential impact on economic activity and jobs. This paper presents a new index of the feasibility to work from home to investigate what types of jobs are most at risk. We estimate that over 97.3 million workers, equivalent to about 15 percent of the workforce, are at high risk of layoffs and furlough across the 35 advanced and emerging countries in our sample. Workers least likely to work remotely tend to be young, without a college education, working for non-standard contracts, employed in smaller firms, and those at the bottom of the earnings distribution, suggesting that the pandemic could exacerbate inequality. Crosscountry heterogeneity in the ability to work remotely reflects differential access to and use of technology, sectoral mix, and labor market selection. Policies should account for demographic and distributional considerations both during the crisis and in its aftermath.
Mr. Jorge A Chan-Lau

-peaked probability distribution as close as possible to a delta distribution. This is equivalent to perform K-means clustering by maximizing the likelihood function: L ( μ , g ; x ) = Π c = 1 K Π c = 1 K P ( μ c | x i ) P ( x i ∈ c l u s t e r c ) ( 3 ) or its log-likelihood. The expectation-maximization algorithm of Dempster et al. (1977) yields the following iterative