Ordered Response Model Research Articles

What are the determinants of road users' experiences with congestion: Econometric assessment using ordered response models

We investigate factors that determine road users' experiences with congestion based on a questionnaire survey conducted in the Oslo area of Norway. The rationale is to add more knowledge on factors that determine users' experience with congestion. Furthermore, we use a succinct econometric framework to assess the data. We find that the following factors influence road users' experiences with congestion: (1) whether they experienced congestion on their reference trip; (2) how often road users undertake trips during congestion; (3) the extent to which road users had potential alternative modes of transport other than car use; (4) education; (5) whether respondents had time commitments at their destinations; (6) travel time used during their journey; (7) how often they experience congestion as a problem during their journey; (8) when participants begin to experience discomfort with congestion; and (9) age. The direction of these factors' impact on experience is explained in the results section. For example, those who did not experience congestion on their previous trip are 26 percentage points more likely to report a negative experience with congestion than those who did. The results provide new insight into the factors that determine road users' experience with congestion. Finally, we warn that one should not be indifferent regarding the logit models used, as the results can be quite different.

A framework for assessment of pedestrianization impacts on quality of life: Combining subjective and objective measures

Pedestrianization is a useful strategy to tackle the challenges resulting from increased vehicular movement on the road that adversely affects the Quality of Life (QOL). Evidence from developed economies shows that pedestrianization has an optimistic influence on cities by advocating liveability and QOL. However, more information is needed about the effect of pedestrianization on QOL in developing economies like India. This study examines the QOL consequences of the pedestrianized urban street by integrating subjective information obtained through revealed preference surveys and objective street environment conditions. The research involves data collected when the pedestrianization measure was implemented in an urban street in Bengaluru, India, as a testbed. The strategy was enforced on all weekends from November 2020 to February 2021. The association between perceived QOL and latent factors is analyzed using the ordered response model to estimate the subjective QOL indicator. The principal component analysis is applied to objective variables including pollution, mobility, and cleanliness, to determine the Objective QOL indicator. Then, the subjective and objective QOL indicators are linearly aggregated to formulate a composite Local QOL index. Results from subjective analysis indicate that satisfaction with health, safety, comfort, and security due to pedestrianization positively affects QOL. Males and married people exhibit positive QOL consequences compared to females and unmarried people. Pedestrian footfall and pollution levels hold high weightage in determining the Objective QOL indicator. The combined evaluation demonstrates that the composite index falls under the category “A” explaining a strong positive influence on QOL. The proposed composite index classification can serve as a unique tool to evaluate the outcomes of pedestrianization measure and support decision-makers in effective strategy enforcement.

Happy Times: Measuring Happiness Using Response Times

Surveys measuring happiness or preferences generate discrete ordinal data. Ordered response models, which are used to analyze such data, suffer from an identification problem. Their conclusions depend on distributional assumptions about a latent variable. We propose using response times to solve that problem. Response times contain information about the distribution of the latent variable through a chronometric effect. Using an online survey experiment, we verify the chronometric effect. We then provide theoretical conditions for testing conventional distributional assumptions. These assumptions are rejected in some cases, but overall our evidence is consistent with the qualitative validity of the conventional models. (JEL C14, D60, D91, I31)

Specification test in panel ordered response models

This paper develops a new simple method to simultaneously test the threshold effect and the correlation between unobserved individual effects and covariates in a panel ordered response model. We first use logistic function to represent the threshold effect, and use Mundlak’s projection formula to characterize the potential correlation between individual effect and covariates. We then construct Wald statistics to test the existence of such correlation and threshold effect separately, and then to test whether the two phenomena exist simultaneously by controlling the false discovery rate (FDR). The advantage of using FDR approach is that we can accurately know the specific form that the model should adopt when rejecting the multiple hypotheses. Simulation results show that our method can effectively capture the actual model even in the finite sample. We apply our method to test the relationship between income and happiness using data from the survey of China Family Panel Studies. While the findings suggest that income is significantly correlated with unobserved individual effect, no evidence is found in support of the threshold effect between income and happiness.

A statistical approach for assessing cyber risk via ordered response models.

Proper evaluation of the risk associated to a cyber attack is a crucial aspect for many companies. There is an increasing need to plan for and implement effective ways to address cyber security, data security, and privacy protection. Estimating the risk of a successful cyber attack is an important issue, since this type of threat is proliferating and thus poses increasing danger to companies and the customers who use their services. While quantitative loss data are rarely available, it is possible to obtain a qualitative evaluation on an ordinal scale of severity of cyber attacks from experts of the sector. Hence, it is natural to apply order response models for the analysis of cyber risk. In particular, we rely on cumulative link models. We explain the experts' assessment of the severity of a cyber attack as a function of a set of explanatory variables describing the characteristics of the attack under consideration. A measure of diffusion of the effects of the attacks obtained via the use of a network structure is also incorporated into the set of explanatory variables of the model. Along with the description of the methodology, we present a detailed analysis of a real data set that includes information on serious cyber attacks occurred worldwide in the period2017-2018.

Influential Factors of Household Car and Vehicle Ownership in Urban Areas of Turkey

Car and vehicle ownership levels in developing countries are gradually increasing, leading to traffic congestion, energy consumption, and air pollution. As a developing country, Turkey encounters many of the issues caused by this rapid increase in car ownership. This paper aims to examine potential determinants of household car and vehicle ownership behavior in urban areas of Turkey by concentrating on three cities (Ankara, Erzurum, and Bayburt) that are distinct in population, geographical region, and public transit services. For this purpose, a written survey was administered to 1389 households living in these cities. The data is analyzed using alternative ordered response models—the generalized ordered logit, partial proportional odds (PPO), and heteroskedastic ordered logit models. To the best of our knowledge, this study is the first attempt to compare alternative ordered response models to determine car and vehicle ownership behavior in Turkey. Consequently, the PPO model was found to be the most parsimonious model among the others. The results demonstrated that household car and vehicle ownership behavior in Turkey differs significantly with urban area size. The estimation results also revealed that several key drivers, including type of dwelling, parking availability, household monthly income, and public transit use, significantly influence Turkish households’ car and vehicle ownership levels. This study’s empirical findings may provide insight to policymakers and stakeholders for future transportation strategies and policies.

Effect of childbearing-age women's family status on the health status of three generations: evidence from China.

It is widely recognized that inequalities in social status cause inequalities in health. Women in a family often directly influence three generations-women themselves, their children and their parents -yet the effect of women's family status on their own health status and that of the two generations before and after is not clear. Taking data from the China Family Panel Studies, this study used an ordered response model to investigate the effect of childbearing-age women's family status on the health status of three generations. The results showed that increases in childbearing-age women's family status improved the health status of the women themselves and their children. Unlike previous studies, however, we found that higher family status did not improve parents' health status but decreased it. The mechanism analysis indicated that women's family status influenced the health status of three generations through economic conditions, resource allocation, and child discipline. The results held after robustness testing. Our findings contribute to knowledge in related fields and provide theoretical support for policies that empower women.

Selection of random coefficients in ordered response models: a framework to detect heterogeneity in household surveys

This paper develops a Bayesian method to detect heterogeneity in the relationship between covariates and the outcome in models with ordered responses. To this end, we construct an efficient Markov chain Monte Carlo algorithm for a hierarchical Bayesian model that selects random coefficients in ordered models. This method extends an approach for selecting random coefficients in linear mixed models into the ordered setting by adding two enhancements that are relevant to the latter category of models. First, we construct steps to efficiently estimate cut-points by addressing identification and ordering constraints. Second, we develop a framework to evaluate marginal effects that combine the fixed and random effects of each covariate. The marginal effects additionally allow for model uncertainty by averaging across models visited by the selection algorithm. Simulation studies demonstrate that this method detects random effects when they are present, estimates parameters accurately and efficiently samples from the posterior with low autocorrelations across successive draws. On applying this method on data from the survey of consumer expectations, we find clear support for the presence of household-level heterogeneity in relationships between demographic variables, and current as well as expected financial conditions.

Association between Air Pollution and Squamous Cell Lung Cancer in South-Eastern Poland.

Air pollution is closely associated with the development of respiratory illness. The aim of the present study was to assess the relationship between long-term exposure to PM2.5, PM10, NO2, and SO2 pollution and the incidence of lung cancer in the squamous subtype in south-eastern Poland from the years 2004 to 2014. We collected data of 4237 patients with squamous cell lung cancer and the level of selected pollutants. To investigate the relationship between the level of concentrations of pollutants and the place of residence of patients with lung cancer in the squamous subtype, proprietary pollution maps were applied to the places of residence of patients. To analyze the data, the risk ratio was used as well as a number of statistical methods, i.e., the pollution model, inverse distance weighted interpolation, PCA, and ordered response model. Cancer in women and in men seems to depend in particular on the simultaneous inhalation of NO2 and PM10 (variable NO2PM10) and of NO2 and SO2 (variable NO2 SO2), respectively. Nitrogen dioxide exercises a synergistic leading effect, which once composed with the other elements it becomes more persistent in explaining higher odds in the appearance of cancers and could constitute the main cause of squamous cancer.

SIMPLE SEMIPARAMETRIC ESTIMATION OF ORDERED RESPONSE MODELS

We propose two simple semiparametric estimation methods for ordered response models with an unknown error distribution. The proposed methods do not require users to choose any tuning parameters, and they automatically incorporate the monotonicity restriction of the unknown distribution function. Fixing finite-dimensional parameters in the model, we construct nonparametric maximum likelihood estimates for the error distribution based on the related binary choice data or the entire ordered response data. We then obtain estimates for finite-dimensional parameters based on moment conditions given the estimated distribution function. Our semiparametric approaches deliver root-n consistent and asymptotically normal estimators of the regression coefficient and threshold parameter. We also develop valid bootstrap procedures for inference. The advantages of our methods are borne out in simulation studies and a real data application.

Estimating Unobserved Soft Adjustment in Credit Rating Models: Before and after the Dodd–Frank Act

Abstract Credit Rating Agencies (CRAs) adjust preliminary bond ratings with knowledge beyond publicly available information. These unobserved “soft adjustments” may reflect material nonpublic information and rating biases due to conflicts of interest, making certain bond characteristics endogenous. We model and quantify soft adjustments as bond-specific thresholds in a semiparametric ordered-response model and exploit ownership structures of bond-issuers to control for endogeneity. Relying on the shift restrictions, we develop a location estimator for models of ordered choices with correlated heterogenous thresholds. Using Moody’s initial ratings from 2000 to 2016, we find a significant reduction of soft adjustment after the Dodd–Frank reform.

Factors Influencing Use and Frequency of Rotational Grazing for Beef Cattle in Tennessee

AbstractThis study determines which factors are associated with the use of rotational grazing and the frequency with which Tennessee producers rotate cattle during the summer months. Survey data were used to estimate an ordered response model with sample selection. Most respondents used rotational grazing, and the most frequent rotational schedule was rotating cattle one to two times per month. Factors including labor, capital, knowledge, and water availability influenced the use of rotational grazing and the frequency of rotating cattle. The insights from this study can inform the development of incentives to promote more intensive use of rotational grazing.

Factors determining teleworking before and during COVID-19: some evidence from Spain and Andalusia

PurposeThis study aims to analyse the inequalities in access to teleworking in Spain. More specifically, the study examines what factors determine remote working in a pre-pandemic period and during the state of alarm (March 2020). Moreover, the study examines whether telecommuting is related to the likelihood of suffering emotional disorders during lockdowns.Design/methodology/approachThe methodological approach followed to analyse access to teleworking mainly consists of the estimation of ordered response models. Two datasets are used: the first is the Survey on Equipment and Use of Information and Communication Technologies (ICTs) in Households, conducted by the Spanish National Statistics Institute (INE) in the pre-pandemic period (2018). The second is the Social Survey 2020. Habits and Living Conditions of the Andalusian Population during the State of Alarm, conducted by the Institute of Statistics and Cartography of Andalusia (IECA) once the state of alarm was declared in Spain.FindingsThe results obtained indicate that ICT training is a key element in helping to explain the likelihood of working from home. However, some groups of workers might experience difficulties in their transition to teleworking. This could increase labour market segmentation and hinder the transition to the knowledge economy. Moreover, the findings detect that employees working both on the company premises and from home are more likely to suffer emotional disorders.Originality/valueThe paper sheds new empirical evidence on teleworking in Spain and Andalusia, including some novel methodological aspects to estimate the regressors’ effects on the probability of working from home.

Partial Identification and Estimation of Semiparametric Ordered Response Models with Interval Regressor Data*

AbstractIn many micro‐data studies, the dependent variable often involves ordered categories and at least one regressor is measured by the interval rather than the precise value. This paper considers partial identification of such an ordered response model when point identification fails. We show the identified set of non‐intercept coefficients is the intersection of those for composite binary response models. We also propose a generalized modified maximum score set (GMMS) estimator. A practical implication of our finding is researchers can shrink the identified set and obtain more precise inference by designing as many as categories of response in a questionnaire during data collection. Another advantage is our theoretical finding can be used to infer the identified region in the multinomial choice model. A Monte Carlo study is conducted to illustrate the main finding in a finite sample. Finally, we apply GMMS estimator to a job satisfaction study using US data with the interval income.

Do financial markets reward government spending efficiency?

We provide a novel set of government spending efficiency scores for the OECD countries and then assess to what extent capital markets perceive government efficiency increases (decreases) as part of the determinants of sovereign rating decisions. Public efficiency scores are computed via data envelopment analysis. Then, we rely notably on ordered response models to estimate the response of sovereign ratings to changes in efficiency scores. Covering 35 OECD countries over the period 2007–2020, we find that increased public spending efficiency is rewarded by financial markets via higher sovereign debt ratings. In addition, higher inflation and government indebtedness lead to sovereign rating downgrades, while higher foreign reserves contribute to rating upgrades.

Uncertainty and disagreement of inflation expectations: Evidence from household‐level qualitative survey responses

AbstractWe propose a procedure that jointly estimates expectation, uncertainty, and disagreement using a flexible hierarchical ordered response model and individual‐level qualitative data. Based on the Michigan survey of US consumers, our results reveal how their inflation expectations and the associated uncertainty are affected by various factors, including their perceptions of economic conditions, recollections of relevant news reports, and sociodemographic characteristics. An examination of the dynamics of inflation uncertainty and disagreement produces evidence in support of using the latter as a proxy of the former. However, our results also highlight important episodes (such as the start of the COVID pandemic) in which the two series diverge.

A Psychophysical Ordered Response Model of Time Perception and Service Quality: Application to Level of Service Analysis at Toll Plazas

This work attempts to bring to the fore the importance of explicitly modelling how time (or other physical quantities like distance) is perceived in the analyses of phenomena or behaviors where time (or other physical quantities) plays an important role. To do so the application area of level of service at toll plazas is chosen. Principles arising out of past work on the psychophysics of time perception are seamlessly incorporated into a novel unified model that accounts for both bias and random errors in perception. Multiplicative error terms arise naturally in this model. Results indicate that it is important to include both bias and random errors while modelling perception as ignoring the bias leads to high variance in the error term. Interestingly the pattern of bias in perception of time seems to remain steady across various waiting line situations from disparate contexts. Finally, the proposed model creates a framework that can be used to determine how time durations are perceived by humans when responses about perceptions are provided through categories like very small, small, etc. whose definitions are also concurrently identified. Such a framework can be used to study a wide range of situations both in transportation (such as route choice, mode choice, gap acceptance, and overtaking) and elsewhere. From a narrower perspective, this methodology can be used, as has been done here, to (i) determine level of service category definitions in terms of observable and “engineerable” variables (actual or measured delay faced by vehicles) even though it is the perceived value (a latent variable) that determines travelers’ responses on the quality of service, and (ii) identify parameters that define the systematic bias and randomness in perceptions.

The number of response categories in ordered response models.

The choice of the number m of response categories is a crucial issue in categorization of a continuous response. The paper exploits the Proportional Odds Models' property which allows to generate ordinal responses with a different number of categories from the same underlying variable. It investigates the asymptotic efficiency of the estimators of the regression coefficients and the accuracy of the derived inferential procedures when m varies. The analysis is based on models with closed-form information matrices so that the asymptotic efficiency can be analytically evaluated without need of simulations. The paper proves that a finer categorization augments the information content of the data and consequently shows that the asymptotic efficiency and the power of the tests on the regression coefficients increase with m. The impact of the loss of information produced by merging categories on the efficiency of the estimators is also considered, highlighting its risks especially when performed in its extreme form of dichotomization. Furthermore, the appropriate value of m for various sample sizes is explored, pointing out that a large number of categories can offset the limited amount of information of a small sample by a better quality of the data. Finally, two case studies on the quality of life of chemotherapy patients and on the perception of pain, based on discretized continuous scales, illustrate the main findings of the paper.

Investigating autonomous vehicle impacts on individual activity-travel behavior

This paper develops an analytic system to investigate the effects of AV availability on multiple dimensions of activity-travel behavior at once, based on a direct survey-based modeling approach. In particular, the model uses individual socio-demographics, built environment variables, as well as psycho-social variables (in the form of latent psychological constructs) as determinant variables to explain likely AV impacts on five dimensions of short-term activity-travel choices: (1) Additional local area trips (that is, those that would not characterized as long distance trips; a long distance trip was defined in the survey as a trip more than 75 miles one-way), (2) Trip distance to shop or eat-out activities in the local area, (3) Trip distance to leisure activities in the local area, (4) Additional long distance road trips beyond the local area, and (5) Commute travel time. The model system includes a confirmatory factor analysis step, a multivariate linear regression model for the latent constructs, and a multivariate ordered-response model for the five main outcomes just listed. Data from a 2019 Austin area survey of new mobility service adoption and use forms the basis for our empirical analysis. Our results, when aggregated across all respondents, does suggest that AVs may not after all have a substantial impact on overall trip-making levels, although local area trips are likely to become longer (for all purposes, including the commute). The highest impact of AVs will, it appears, be on the number of long distance trips (with such trips increasing). Our in-depth examination of the variations in AV activity-travel responses across population segments and geographies underscores the importance of modeling multiple activity-travel dimensions all at once. In addition, our results highlight the value of using psycho-social latent constructs in studies related to the adoption/use of current and emerging mobility services, both in terms of improved prediction fit as well as proactive strategies to design equitable, safe, and community-driven AV systems. There is likely to be considerable heterogeneity in how different population groups view and respond to AVs, and it is imperative that AV campaigns and AV design consider such heterogeneity so as to not “leave anyone behind”.

Semiparametric Estimation of a Corporate Bond Rating Model

This paper investigates the incentive of credit rating agencies (CRAs) to bias ratings using a semiparametric, ordered-response model. The proposed model explicitly takes conflicts of interest into account and allows the ratings to depend flexibly on risk attributes through a semiparametric index structure. Asymptotic normality for the estimator is derived after using several bias correction techniques. Using Moody’s rating data from 2001 to 2016, I found that firms related to Moody’s shareholders were more likely to receive better ratings. Such favorable treatments were more pronounced in investment grade bonds compared with high yield bonds, with the 2007–2009 financial crisis being an exception. Parametric models, such as the ordered-probit, failed to identify this heterogeneity of the rating bias across different bond categories.