scispace - formally typeset
Search or ask a question
Topic

Proxy (statistics)

About: Proxy (statistics) is a research topic. Over the lifetime, 5257 publications have been published within this topic receiving 94504 citations. The topic is also known as: proxy variable & proxy measurement.


Papers
More filters
Journal ArticleDOI
TL;DR: In this article, the authors developed three new spread proxies based on price clustering, extending the roll serial covariance model, and using daily quoted spread data available for no-trade days only.
Abstract: I develop new proxies for liquidity as computed from low-frequency (daily) data. Specifically, I develop three new spread proxies based on: (1) price clustering, (2) extending the Roll serial covariance model to include no-trade midpoints, and (3) using daily quoted spread data available for no-trade days only. I introduce multi-factor liquidity proxies (linear combinations of individual liquidity proxies) and show theoretically that they can diversify away some imperfectly correlated measurement error. I test many variations of each new proxy on three performance dimensions using both U.S. and Mexican stock data. For the U.S., I find that on all three performance dimensions the best proxy is a multi-factor liquidity measure composed of the best (or near best) new measures and exploiting all (or most) of the variables available in the U.S. Also, I find that the best multi-factor proxy is statistically significantly better than any of the individual measures. For Mexico in both price-based comparisons and general comparisons, I find on the two correlation-based performance dimensions that the new measures are not the best on an individual basis. However in each of these cases, the best proxy overall is a multi-factor proxy that mixes in the some of the new measures as components. On the prediction error tests, no measure does especially well, but the best performer overall is a multi-factor proxy that mixes in the some of the new measures as components.

32 citations

Journal ArticleDOI
TL;DR: In this paper, the ECRI leading and long leading indexes are used as putative determinants of these transition probabilities, and a broader, more comprehensive proxy for the business cycle is employed to test the usefulness of leading indicators in forecasting the likelihood of future business cycle phase shifts.

32 citations

Journal ArticleDOI
TL;DR: Physical connectivity as measured in this paper has a significant effect on social cohesion, and that this measure is unlikely to proxy either deprivation or the urban/rural status of communities.
Abstract: Background: There is now a substantial body of research suggesting that social cohesion, a collective characteristic measured by the levels of trust, reciprocity and formation of strong social bonds within communities, is an important factor in determining health. Of particular interest is the extent to which factors in the built environment facilitate, or impede, the development of social bonds. Severance is a characteristic of physical environments which is hypothesized to inhibit cohesion. In the current study we test a number of characteristics of spatial networks which could be hypothesized to relate either to severance, or directly to community cohesion. Particular focus is given to our most promising variable for further analysis (Convex Hull Maximum Radius 600 m). Methods: In the current study we analysed social cohesion as measured at Enumeration District level, aggregated from a survey of 10,892 individuals aged 18 to 74 years in the Caerphilly Health and Social Needs Cohort Study, 2001. In a data mining process we test 16 network variables on multiple scales. The variable showing the most promise is validated in a test on an independent data set. We then conduct a multivariate regression also including Townsend deprivation scores and urban/rural status as predictor variables for social cohesion. Results: We find convex hull maximum radius at a 600 m scale to have a small but highly significant correlation with social cohesion on both data sets. Deprivation has a stronger effect. Splitting the analysis by tertile of deprivation, we find that the effect of severance as measured by this variable is strongest in the most deprived areas. A range of spatial scales are tested, with the strongest effects being observed at scales that match typical walking distances. Conclusion: We conclude that physical connectivity as measured in this paper has a significant effect on social cohesion, and that our measure is unlikely to proxy either deprivation or the urban/rural status of communities. Possible mechanisms for the effect include intrinsic navigability of areas, and the existence of a focal route on which people can meet on foot. Further investigation may lead to much stronger predictive models of social cohesion.

32 citations

Posted Content
TL;DR: In this article, a simple estimator is developed to identify structural shocks in vector autoregressions (VARs) by using a proxy variable that is correlated with the structural shock of interest but uncorrelated with other structural shocks.
Abstract: This paper develops a simple estimator to identify structural shocks in vector autoregressions (VARs) by using a proxy variable that is correlated with the structural shock of interest but uncorrelated with other structural shocks. When the proxy variable is weak, modeled as local to zero, the estimator is inconsistent and converges to a distribution. This limiting distribution is characterized, and the estimator is shown to be asymptotically biased when the proxy variable is weak. The F statistic from the projection of the proxy variable onto the VAR errors can be used to test for a weak proxy variable, and the critical values for different VAR dimensions, levels of asymptotic bias, and levels of statistical significance are provided. An important feature of this F statistic is that its asymptotic distribution does not depend on parameters that need to be estimated.

32 citations


Performance
Metrics
No. of papers in the topic in previous years
YearPapers
20231,242
20222,473
2021334
2020262
2019250
2018282