Parallel Session 3

Parallel Session 3

Thursday 9 January 2020, 16.00-17.00.

A new presentation starts every half hour if nothing else is stated. Open tabs to read abstracts. Click on venue name to go to an electronic map.

Business Economics II

Aud.A Session Chair: Denis Utochkin

  • Aras Khazal: Revealing the Pure Energy performance Certificate Impact and the role of asymmetric information in housing utility cost decision-making

    Aras Khazal: Revealing the Pure Energy performance Certificate Impact and the role of asymmetric information in housing utility cost decision-making

    This paper uses different identification strategies to investigate the capitalization of Energy Performance Certificates (EPCs) in the Norwegian residential rental market, highlighting the endogeneity issue. Controlling for locational heterogeneity and using the Technical Standard Regulations of buildings information, we find a premium for labeled homes overall and green homes specifically. The IV-approach confirms the results and suggests that the only source of endogeneity is due to locational heterogeneity, rather than unobserved quality standards. We further construct a model to explain the underlying mechanisms of utility cost premiums in relation to EPCs. Labeled homes are more likely to include utility costs, and the likelihood is higher among green homes than non-green homes. We find that marketing, asymmetric information and uncertainty factors that explain the utility cost premiums. The findings suggest that the labeling policy is important for actor’s decision-making and reduces asymmetric information and uncertainty price effects.

Accounting III

Aud.B Session Chair: Mohammad Zarei

  • Even Fallan, Henrik Höglund and Dennis Sundvik: Cost behavior and external accounting help in private firms

    Even Fallan, Henrik Höglund and Dennis Sundvik: Cost behavior and external accounting help in private firms

    This is an investigation of cost behavior in private firms. It discriminates between firms that use external help with accounting processes and firms that perform accounting tasks internally. We argue for two conflicting reasons of a difference in cost behavior between these groups of firms. External accounting professionals could help firms without sufficient internal resources to achieve more symmetrical cost behavior through sophisticated business advice and industry expertise. Or, most firms utilize only statutory accounting services from accounting service providers and firms with internal accounting processes could achieve more efficient resource allocation and utilization since they are closer to the numbers and have increased awareness of the underlying business economics, future prospects, and cost accounting in general. Based on a large sample of small private firms, our findings are consistent with the second reasoning. Additional tests suggest that while external accounting help has a general negative effect on cost behavior, auditing has a positive effect. Taken together, the study highlights consequences of engaging external parties in the financial process of small firms.

    Discussant: Kenneth Fjell

  • Ranik Raaen Wahlstrøm, Florentina Paraschiv & Markus Schmid: Bankruptcy prediction of privately held SMEs: a study of input variables using feature selection methods

    Ranik Raaen Wahlstrøm, Florentina Paraschiv & Markus Schmid: Bankruptcy prediction of privately held SMEs: a study of input variables using feature selection methods

    Our goal is to illustrate the appropriateness and importance of feature selection methods when identifying input variables for predicting bankruptcy in general, and for SMEs in particular. We consider 154 different accounting variables from existing literature and use a filter, a wrapper and an embedded feature selection method, respectively, to identify optimal subsets. Empirical evidence is found by using a comprehensive data set of all annual financial statements of all Norwegian limited SMEs over nine accounting years, used in an oversampling technique for achieving a balanced data set of 29,884 financial statements. As tools we use logistic regression and an artificial neural network, respectively. Subsets identified using the wrapper and LASSO feature selection methods outperform established subsets from existing literature. In general, we found that accounting-based input variables provide a significant explanatory power when it comes to predicting bankruptcy of privately held SMEs in a one-year perspective.

Method

Aud.C Session Chair: Jonas Andersson

  • Espen Sirnes: Paneltime - an unbalanced panel data application

    Espen Sirnes: Paneltime - an unbalanced panel data application

    Paneltime is a python based application for analyzing large unbalanced panel data sets with severe ARIMA and GARCH problems. The integrated approach avoids the common problem that the estimation procedure determine the results. 
    There does not currently appear to exist software that allow for time series effects in panels. There are even very few options to handle GARCH and ARIMA simultaneously, possibly is only rugarch in R able to do that. 
    paneltime is a maximum likelihood estimator with analytical calculation of both the gradient and the hessian, which makes it unprecedented accurate. By analyzing collinearity at each iteration, the estimation rarely stalls, which is a common problem with GARCH and ARIMA estimators. 
    In addition, the analytical hessian allows identification of “valleys” in the likelihood terrain, where the objective function is convex. Most GARCH and ARIMA software rely on a numerical hessian, and will stall when that happens.
  • Ingvild Margrethe Helgøy: A Noise-Robust Sparse Bayesian Learning Model

    Ingvild Margrethe Helgøy: A Noise-Robust Sparse Bayesian Learning Model

    This paper presents a sparse Bayesian Machine Learning model. Bayesian learning models relies on Bayes' theorem, where data can be used to update the prior beliefs about parameter values. From the Bayesian framework we obtain the posterior distribution of our parameters, and the corresponding covariance matrix of this distribution gives the uncertainty of each estimated parameter value. Sparse models are models where only a small fraction of the estimated parameters are non-zero, and arise frequently in machine learning. The sparsity is an important property, since it reduces the computational time and sparse models may also be more easily interpreted.

Finance III

PC Lab 1 Session Chair: Nicolas Gibney

  • Negar Ghanbari: Do Creditor Rights Affect Financial Contracts? Evidence from the Anti-Recharacterization Statute

    Negar Ghanbari: Do Creditor Rights Affect Financial Contracts? Evidence from the Anti-Recharacterization Statute

    This paper examines the effect of creditor rights on bank loan contract design.
    Focusing on the confict of interest between creditors, I study how bank lenders respond to a legal change that strengthens the rights of securitization creditors. Improving the power of securitization creditors to seize their collateral in bankruptcy reduces their incentives to maximize recoveries in chapter 11, increasing the risk of other competing creditors, such as banks. I find that loans granted to firms using asset securitization have higher interest rates, higher fees, smaller size, and more covenant restrictions after the law change. These effects are stronger for firms with higher default risk, for which the legal change may have a bigger impact. My findings thus highlight how increasing the power of some corporate creditors affects the other financial contracts of the firm.
     

    DISCUSSANT: Thore Johnsen

  • Ane Haugdal: Earnings management as a tool to delist rom ROBEK

    Ane Haugdal: Earnings management as a tool to delist rom ROBEK

    The purpose of this paper is to supplement the earnings management literature on the public sector by investigating whether Norwegian municipalities undertake earnings management actions in order to improve their financial profile to delist from Robek.
     
    In this study, I want to use, among others, the Jones (1991) model and the modified Jones model (Dechow et al. 1995) to estimated the size of the discretionary accruals for a group of municipalities which have left Robek, and a control group consisting of municipalities who have never been on the Robek register (during the same time period). 
    In addition, I want to look into whether there are differences for the Robek group in the average size of the discretionary accruals the year before delisting, the year they were delisted and the year after.

Auditing

PC Lab 2 Session Chair: Rujuta Vaidya

  • Brynjar Gilberg: Konsernperspektiv i kommunal revisjon

    Brynjar Gilberg: Konsernperspektiv i kommunal revisjon

    Regnskapsskandaler i kommunalt eide selskaper har fått stor oppmerksomhet i media, og i enkelte tilfeller endt med kostbare granskninger. Det synes ikke å ha vært like stor oppmerksomhet blant kommunale eiere og revisorer. Ny kommunelov krever at det fra 2020 utarbeides konsoliderte regnskaper for kommunens virksomhet, men kravet gjelder kun foretak som inngår i samme juridiske enhet. Utviklingen viser en økning i antall og størrelse av kommunalt eide selskaper. Etter revisorloven skal revisor gjennomgå revisjonsarbeid utført av revisor i datterselskap av betydning for konsernregnskapet. Tilsvarende krav gjelder der deler av virksomheten blir utført av en serviceorganisasjon. Det foreligger ikke informasjon om hvorvidt god kommunal revisjonsskikk skiller seg fra god revisjonsskikk på dette området. En undersøkelse av praksis blant kommunale revisorer vil kunne gi svar på dette, og bidra til en diskusjon av hvorvidt det er gode grunner for at god kommunal revisjonsskikk kan avvike fra god revisjonsskikk på dette området.

  • Øivind André Strand Aase: Introducing Voluntary Audit for Small Private Firms - Effects on Accounting Quality and Bunching-behavior

    Øivind André Strand Aase: Introducing Voluntary Audit for Small Private Firms - Effects on Accounting Quality and Bunching-behavior

    Small private firms generally prepare financial statements under a more lenient disclosure and audit regime relative to other private and public firms. Private firms around thresholds for mandated audit have incentives to size down to avoid audit costs unless perceived benefits of audit outweigh these costs. I utilize the introduction of a revenue threshold in Norway in 2011, emanating from a legislative amendment that introduced voluntary audit for small private limited liability firms. The aim is to study whether this introduction affects accounting quality, and whether firms tend to bunch below the threshold for mandatory audits. In terms of accrual quality and conditional conservatism, the firm-fixed effects analyses indicate no consistent significant fall in accounting quality among eligible firms after the reform, and neither among firms that drop audit. The results show bunching tendencies under the revenue threshold in the post-reform period, implying that avoidance of audit is important for certain firms. However, I find no indication of lower accounting quality in bunching firms. Real earnings management is also investigated, and findings reveal no such behavior among bunchers. The audit reform, hence, seems to be a welcome reform with low costs for firms and firms’ stakeholders.