Bureau of the Fiscal Service

Topic models, an unsupervised technique for inferring translation domains improve machine translation quality. However, previous work uses only the source language and completely ignores the target language, which can disambiguate domains. We propose new polylingual tree-based topic models to extract domain knowledge that considers both source and target languages and derive three different inference schemes. We evaluate our model on a Chinese to English translation task and obtain up to 1.2 BLEU improvement over strong baselines.

In this paper, we argue that there is much room for China to strengthen its regulatory framework for public-private partnerships (PPPs). We show that infrastructure projects carried out through local government financing vehicles (LGFVs) are largely unregulated PPPs, and significant fiscal risks have already manifested themselves. While PPPs can potentially provide efficiency gains, they can also be used by governments to circumvent budgetary borrowing constraints. Therefore, effective PPP regulation is key to delivering PPPs’ benefits while containing their potential fiscal risks. The authorities have taken concrete steps in order to establish a sound regulatory framework and foster a new generation of PPPs. However, to make the framework effective, we highlight a few issues to be resolved. Based on international best practice, we propose a four-pillar regulatory framework for China, which could be implemented gradually in three stages.

In 1982, the Pennsylvania Department of Transportation (PennDOT) became one of the first transportation agencies to introduce strategic management as an operating principle. PennDOT's process includes an agency strategic agenda that is developed every 4 years under the guidance of a strategic management committee and is implemented through annual business plans. Strategic plans are used to drive other management and decision-making processes to achieve agency goals and objectives. Four separate administrations representing different political parties have used this process to improve PennDOT's business results and have enhanced the process in each successive iteration. Despite this record of success, a 1998 Baldrige assessment revealed that PennDOT's strategic management process had significant shortcomings—particularly in the area of implementation. A 2-year gap closure effort was initiated in response to these findings; the effort strengthened the overall process and established scorecard targets, linking planning to implementation and facilitating monitoring and review throughout the department. Strengths and opportunities were identified along with recent enhancements made by the current strategic management committee after the change of administration in 2003.

Federal administrative agencies are one of the primary policymaking venues in the United States. One of the core features of U.S. administrative practice is the notice-and-comment process in which agencies solicit, collect, and respond to comments from the public before issuing new regulations. In this paper, we develop a model of commenting based on three motivations—litigation preservation; agency persuasion; and expression—and analyze public comments to determine how features of the political environment, and specifically the president in power, affect the pool of commenters. We focus on the 2017 presidential transition, when there was both a change in Presidents and the party in control of the White House. We find that there were greater differences in the pool of commenters between administrations than within administrations and that interest groups tended to participate more when they were more closely associated with the party in power. Our findings support the view that many commenters use the public comment process for persuasive purposes, and not only to preserve litigation opportunities or for purely expressive reasons.