Var Özge, Durmuşoğlu Alptekin, Dereli Türkay
Department of Industrial Engineering, Gaziantep University, Gaziantep, Turkey.
Department of Industrial Engineering, Samsun University, Samsun, Turkey.
Heliyon. 2024 Dec 27;11(2):e41447. doi: 10.1016/j.heliyon.2024.e41447. eCollection 2025 Jan 30.
Consumer Confidence Index (CCI) is a measure obtained from consumer surveys (CS) that gauges assessments and expectations of the economic environment. Common practice uses 4 of the 12 questions in CCI calculation. However, efforts to find best set of questions continue, such as the European Commission swapping two questions in 2019. Literature studies employ different combinations of questions; however all-alternative combinations take too much time and computational power. The questions also exhibit cause-and-effect relationships as household consumption predictors and are not statistically independent of one another.
We suggest classifying the CS questions as "Causes" and "Effects." It makes sense that inquiries in the cause group should provide a better explanation of household consumption. If this theory turns out to be correct, a smaller solution space will be able to be used to find the ideal substitute CCI.
A fuzzy DEMATEL (Decision-Making Trial and Evaluation Laboratory), a reliable method to present causal relationships, is used to classification. The prediction power of cause group (in terms of explaining household expenditures) is measured with the Lasso regression (Least Absolute Shrinkage and Selection Operator), which provides more interpretable regression models. This approach was applied to European Union dataset from 2007Q3 to 2021Q2.
The cause group included four CS questions and explained the 75% variability of the consumption expenditures. It is performed comparably to earlier studies that took into account all possible question combinations. The Türkiye case, covering data from 2007 to 2021, supported the finding of EU case, explaining 84% variation in consumption expenditures.
These encouraging results suggest that comparable prediction power can be attained with a significant reduction in effort (in comparison to all brute force). Therefore, this approach would provide shortcut for constructing alternative CCIs to the authorities.
消费者信心指数（CCI）是通过消费者调查（CS）得出的一项指标，用于衡量对经济环境的评估和预期。通常做法是在计算CCI时使用12个问题中的4个。然而，寻找最佳问题集的工作仍在继续，例如欧盟委员会在2019年更换了两个问题。文献研究采用了不同的问题组合；然而，所有可能的组合花费时间过多且计算量过大。这些问题作为家庭消费预测指标还呈现出因果关系，彼此之间并非统计独立。
我们建议将消费者调查问题分为“原因”和“结果”两类。原因组中的问题应该能更好地解释家庭消费，这是合理的。如果这一理论被证明是正确的，那么就可以用一个更小的解空间来找到理想的替代CCI。
使用模糊DEMATEL（决策试验与评价实验室）这一可靠的呈现因果关系的方法进行分类。用套索回归（最小绝对收缩与选择算子）来衡量原因组（在解释家庭支出方面）的预测能力，该方法能提供更具解释性的回归模型。此方法应用于2007年第三季度至2021年第二季度的欧盟数据集。
原因组包括四个消费者调查问题，解释了消费支出75%的变异性。其表现与早期考虑所有可能问题组合的研究相当。涵盖2007年至2021年数据的土耳其案例支持了欧盟案例的研究结果，解释了消费支出84%的变化。
这些令人鼓舞的结果表明，与所有蛮力方法相比，付出显著更少的努力就能获得相当的预测能力。因此，这种方法将为当局构建替代CCI提供捷径。
