nep-dcm New Economics Papers
on Discrete Choice Models
Issue of 2014‒09‒05
ten papers chosen by
Edoardo Marcucci
Università degli studi Roma Tre

  1. A discrete continuous model of vehicle ownership and use in Flanders By Franckx, Laurent; Michiels, Hans; Mayeres, Inge
  2. "Preference for Flexibility and Random Choice: an Experimental Analysis" By Mark Dean; John McNeil
  3. Classification using Random Forests in Stata and R By Linden McBride; Austin Nichols
  4. Household Formation and Tenure Choice By Byrne, David; Duffy, David; FitzGerald, John
  5. The Effects of the EU-Ukraine FTA: An Inequality Analysis using a CGE-Microsimulation Model for Ukraine By Miriam Frey
  6. Household Formation and Tenure Choice: Did the great Irish housing bust alter consumer behaviour? By Byrne, David; Duffy, David; FitzGerald, John
  7. A midas retouch regarding diagnostic meta-analysis By Ben Dwamena
  8. Balanced use of fertilizer nutrients and its determinants: a case of cotton crop By Khuda Bakhsh -
  9. Job Networks in Izmir: Why are Migrants Different? By Idil Goksel; Alper Duman
  10. Wage rigidity and employment adjustment at the firm level: Evidence from survey data By FERNANDO MARTINS; Daniel Dias; Carlos Marques

  1. By: Franckx, Laurent; Michiels, Hans; Mayeres, Inge
    Abstract: We estimate a discrete-continuous model of vehicle demand and use for the Belgian region of Flanders, combining the results of the official regional travel survey with a detailed database of vehicle characteristics. The overall predictive value of the submodel predicting the number of vehicles owned by each household is satisfying, and in line with expectations. However, existing data turn out to be relatively poor predictors of the vehicle class owned by households and of the annual mileage per vehicle. We argue that the current travel survey focuses on determinants of travel in the peak periods. In order to predict overall travel demand, future versions of the travel survey should identify indicators with a higher predictive value for travel behavior for other than commuting purposes.
    Keywords: vehicle choice, discrete-continuous choice modelling
    JEL: C25 H23 R41 R48
    Date: 2014–08–25
  2. By: Mark Dean; John McNeil
    Abstract: People may be uncertain about future preferences, leading to both a preference for flexibility in choice between menus and stochastic choice from menus. This paper describes an experimental test of preference uncertainty in a realeffort task. We observe subjects’ preferences over menus of work contracts, along with their choices of effort levels from those contracts. Our results suggest that preference uncertainty is important: 48% of our subjects exhibited strict preference for flexibility. A model of preference uncertainty (Ahn and Sarver (2013)) well describes the relationship between choice of and from menus: subjects willing to pay to include an option in a contact were more likely to use that option, and those that used an option were prepared to pay for it. We show that the introduction of an explicit stochastic element to the contract increased preference for flexibility, suggesting a causal role for uncertainty in menu preferences
    Keywords: #
    Date: 2014
  3. By: Linden McBride (Cornell University); Austin Nichols (The Urban Institute)
    Abstract: Many estimation problems focus on classification of cases (into bins) with tools that aim to identify cases using only a small subset of all possible questions. These tools can be used in diagnoses of disease, identification of advanced or failing students using tests, or classification into poor and nonpoor for the targeting of a means-tested social program. Most popular estimation procedures for generating these tools prioritize minimization of in-sample prediction errors, but the objective in generating such tools is the minimization of out-of-sample prediction errors. We provide a comparison of linear discriminant, discrete choice, and random forest methods, with applications to means-tested social programs. Out-of-sample prediction error is typically minimized by random forest algorithms.
    Date: 2014–08–02
  4. By: Byrne, David; Duffy, David; FitzGerald, John
    Date: 2014–08
  5. By: Miriam Frey
    Abstract: This paper analyzes the effects of the planned free trade agreement (FTA) between the European Union (EU) and Ukraine on inequality and poverty in the latter using a CGE-microsimulation model for Ukraine. Special attention is thereby given to the in-house production of agricultural and processed food goods generated by the Ukrainian households. Due to a lack of information on these activities in the national accounts data, this type of household production is neglected in an already existing CGE analysis of the welfare effects resulting from the EU-Ukraine FTA. However, in-house production of food products for own consumption plays a crucial role in Ukraine. According to a household expenditure survey for Ukraine 60.5% of the households are engaged in the production of agricultural goods like potatoes, eggs and cabbage and an even slightly higher percentage (61.6%) of them reported to produce processed food goods like dairy products and preserves. The CGE-microsimulation model used in this paper follows the top-down approach, meaning that variables, such as prices and factor returns which change as a result of the simulation of the trade integration in the CGE model are then transferred to the microsimulation model where they are treated as exogenous variables. The CGE model used here is a rather standard, small open economy, single-country model for Ukraine exhibiting perfect competition and constant returns to scale. It incorporates 38 sectors of production and a representative household which is disaggregated into four types according to the domestic poverty line and the place of residence (rural or urban). Labor is differentiated based on the level of education in skilled and unskilled labor and is assumed to be fully employed. The microsimulation model consists of a log-income and a discrete choice labor supply equation as well as of accounting identity and arithmetical computation equations. In-house production of agricultural and processed food products is not treated as being part of the – in most of the literature existing – labor market alternative “being self-employed”, but is modeled explicitly and analogous to the labor status as a discrete choice. This is mainly done for the following two reasons. Labor market status is assumed to be a choice made on the individual level, whereas the decision whether or not to participate in household production of agricultural and processed food goods for own consumption is made on the household level. The second reason is that the explanatory variables which influence the choice of the labor market alternative do not necessarily have to coincide with the ones that determine the decision on in-house production of food products. The data needed for the CGE model include the Ukrainian national accounts and input-output tables for 2007, additional statistics from national sources like information on indirect taxes, labor remuneration and tariffs, international trade statistics and a household expenditure survey for 2007 covering more than 10,000 Ukrainian households and more than 20,000 household members. The latter is also the basis for the microsimulation model. Concerning the household level, information on expenditures, place of residence, characteristics of the household head and land ownership are the most important variables. With respect to the household members, information on sex, age, education, labor market status and income are crucial.
    Keywords: Ukraine, Trade issues, General equilibrium modeling
    Date: 2013–06–21
  6. By: Byrne, David; Duffy, David; FitzGerald, John
    Date: 2014–07
  7. By: Ben Dwamena (University of Michigan)
    Abstract: The talk describes recent updates for midas, a comprehensive and medically popular program for diagnostic test accuracy meta-analysis. A major change is that midas is now an estimation command and a wrapper for meglm of Stata 13 . The update allows more flexibility for specifying covariance structures, link functions other than logit, more extensive post-estimation options and specification of starting values especially with sparse data; and the possibility of estimating univariate (independent) versus bivariate (correlated) modeling of sensitivity and specificity
    Date: 2014–08–02
  8. By: Khuda Bakhsh -
    Abstract: Although fertilizer use has substantially increased overtime in Pakistan, its application is highly imbalanced resulting in wastage of scarce resources and additional cost to farmers and economy as well because fertilizer industry is among major consumers of natural gas in Pakistan. Nitrogenous fertilizers are commonly used in cotton production compared to other fertilizer nutrients. Sine, production of nitrogenous fertilizers involves natural resources, namely natural gas, its indiscriminate use leads to demand for more of natural gas. Pakistan is facing severe supply problems of oil and gas resources, so society suffers from external cost due to allocation of natural resources to fertilizer production. The need is to explore reasons for imbalanced use of fertilizer nutrients so policies may be made accordingly.The present study was designed to estimate the determinants of imbalanced use of fertilizer nutrients in Pakistani PunjabRespondents growing cotton crop were randomly selected for the present study. Farmers were grouped into two categories, those using balanced amount and those using imbalanced amounts. So, it is necessary to estimate factors causing this type of beahviour among farmers. As the dependent variable is binary, Logit model was employed to estimate factors having impact on imbalanced use of fertilizer nutrients.Results indicated that social networks, access to information and institutions, availability of financial resources augmented by off-farm income were significantly related with balanced use of fertilizer nutrients. Age, farming experience and primary education also affected the selection of nutrients. Thus, strengthening institutions, increasing access to information and capacity building of farmers may lead towards the optimum and balanced use of fertilizer nutrients in the study area, giving sustainable cotton production.
    Keywords: Pakistani Punjab, Agriculture, Energy
    Date: 2013–09–05
  9. By: Idil Goksel; Alper Duman
    Abstract: The aim of this paper is to investigate the network effect on the probability of job finding. This paper uses a specific data set from the Izmir region, prepared by the Turkish Statistical Institute for a specific project carried out by Izmir University of Economics in cooperation with the Izmir Commerce Centre, the Izmir branch of the Turkish Statistical Institute and the Turkish Labour Institute. Its aim was to investigate the labour market situation of Izmir and to understand the main problems of labour market and reasons of unemployment and suggest possible policy implications. During this study we found that the migrants who have moved to Izmir five years or more ago have higher probability to find a job than Izmir born ones. The same argument is also valid for the ones who moved less than two years ago. Only the ones that have moved 2-5 years ago have less chance to find a job. These trends show also difference according to the gender. Furthermore, it is found that migrants tend to earn more relative to the natives. This variety made us think that it will be interesting to analyze this subject a bit deeper. Izmir is the third biggest city in Turkey and has been one of the preferred destinations of migrants. Izmir attracts both skilled and unskilled migrants. Moreover, we believe doing such an analysis at a local level is more promising than doing it in the state level, as it will be easier to isolate the network effects in local level. The paper aims to analyze how the network effect differs between skilled and unskilled people and the nonlinearity in the network effects. It is assumed that the more time spent in a city, the larger is the network and quality of the network connections depend on the job seeker’s qualifications. Using Armengol and Jackson (2007) theoretical model and the empirical framework of Munshi (2003), a probit model that estimates the probability of being employed is used. In the model personal characteristics, the sectors, and the number of years spent in Izmir are controlled for. Moreover, the education levels and occupations of the parents are used as proxies for the size and the quality of the network. It is expected to find nonlinearity in network effects. We also expect to find higher influence of quality of network with respect to its dimension. It will be also interesting to investigate whether network effects differ according to the gender or not.
    Keywords: Turkey, Labor market issues, Regional modeling
    Date: 2013–06–21
  10. By: FERNANDO MARTINS; Daniel Dias; Carlos Marques
    Abstract: This paper uses firm level survey data from Portugal to investigate how firms adjust their labour costs in the presence of wage rigidities. In particular, the paper contributes to the literature by analysing how firms, in the presence of wage rigidity, combine different channels of labour-cost adjustment in response to adverse shocks. Wage rigidity is expected to have implications for unemployment because, in the face of negative shocks, employment adjustment is likely to be larger when wages are rigid downwards. Wage rigidity is also thought to have important implications for monetary policy, as it may condition the inflation target that monetary authorities should pursue. If nominal wages were perfectly flexible it would be optimal to aim at zero inflation but, in the presence of downward nominal wage rigidity, a certain amount of inflation may be required to "grease the wheels" of the labour market by easing reductions in real wages. Model estimated by single equation methods (probit model). Recursive triangular model. We conclude that base-wage flexibility has a strong positive impact on employment, and that such positive impact has been significantly strengthened by the possibility of firms resorting to alternative margins of labour cost adjustment, like more flexible compensation components (bonus, benefits and promotions) and the recruitment of new employees at wages lower than those received by the employees that have left the firm.
    Keywords: Portugal, Labor market issues, Microsimulation models
    Date: 2013–06–21

This nep-dcm issue is ©2014 by Edoardo Marcucci. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.