Consumer Sentiment, Managerial Expectations and Resource Adjustment Decisions

Abstract

This study presents evidence on the relationship between consumer sentiment and resource adjustment decisions. Consumer sentiment is an important piece of economic information, accepted as a reliable predictor of future economic activity, which is why it should influence managerial expectations underlying future-oriented resource adjustment decisions. In line with these considerations, I find that managers are more likely to retain slack resources following a decrease in sales when consumer sentiment about future business prospects is improving. Managers seem to adopt consumers’ optimism about future economic prospects by deciding to stall resource adjustments until sales recover to avoid current and future adjustment costs, thus increasing firms’ level of sticky cost behavior. Together with various additional analyses, this study provides new insights into managers’ resource adjustment decision-making process and enhances our understanding of the information upon which managers form their expectations.

Keywords:

• Cost behavior
• Cost management
• Cost stickiness
• Consumer confidence
• Managerial optimism
• Managerial decision-making

JEL Classifications:

• D8
• E20
• M41

Acknowledgements

I thank Victor Maas (Associate Editor) and two anonymous referees for their valuable comments and guidance. I also thank Brigitte Eierle, Maximilian Margolin (discussant), Francesco Mazzi and seminar participants at the EAA 2021 Virtual Congress and the University of Bamberg for their helpful comments and suggestions. All errors remain my own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For instance, the former Chair of the Federal Reserve Alan Greenspan frequently referred to consumer sentiment as a key economic indicator (Ludvigson, 2004).

2 Generally, how people and in particular managers make decisions and form expectations is an important and continuous question for economists and psychologists (Anderson & Goldsmith, 1994).

3 I identify at least four articles dealing with consumer sentiment that appeared in the first half of 2019 in the New York Times. Moreover, a recent Internet search (Google News) for ‘consumer sentiment’ produced almost 5,000,000 hits.

4 ‘Animal spirits’ shocks are considered as irrational whereas ‘sunspot’ shocks are self-fulfilling under rational expectations (Benhabib & Spiegel, 2019). In the context of stock prices, the specific effect of consumer sentiment survey outcomes on the behavior of economic players has also been referred to as a ‘publication effect’ (Jansen & Nahuis, 2003).

5 In the finance/financial accounting literature, consumer sentiment is often also referred to as investor sentiment, but it is basically measured with the same surveys. Other literature in this area shows that investor sentiment is significantly associated with disclosure policies (Bergman & Roychowdhury, 2008; Brown et al., 2012) and earnings management (Simpson, 2013).

6 Conversely, in times of sales upswing, managers might rely on consumer sentiment in order to decide whether or not to use the rents from current growth to invest in future revenue-generating resources. If consumer sentiment improves, managers might believe that demand growth is permanent so that such an investment would be likely to yield the desired future revenue. Alternatively, if managers believe that demand growth is temporary based on consumer-sentiment information, they could refrain from adding resources to avoid adjustment costs and instead temporarily raise selling prices (Cannon, 2014). However, in line with prior cost asymmetry literature I primarily focus on the decision to retain slack resources in periods of sales downturn in the development of my main hypothesis and I do not formulate a separate hypothesis for sales-increasing resource adjustments.

7 However, some studies also exclusively focus on the forward-looking subset of questions or one specific question of the consumer sentiment survey (e.g., Benhabib & Spiegel, 2019).

8 Cannon et al. (2020) also argue that increased takeover threats induce managers to make myopic resource adjustment decisions.

9 I thank an anonymous referee for pointing out this alternative perspective of how managers react to sales downturns in adjusting slack resources.

10 The role of uncertainty also plays a role in cost management research. Specifically, Banker, Byzalov, and Plehn-Dujowich (2014) investigate empirically and analytically how demand uncertainty might influence cost structure decisions.

11 In an untabulated test, I have also retrieved data from the OECD which publishes the Consumer Confidence Index (CCI) and the Business Confidence Index (BCI). Although slightly adjusted, the CCI for the US is based on data from the University of Michigan and the BCI for the US is based on the Manufacturing Index from the Institute for Supply Management (ISM), which measures sentiment among managers at manufacturing firms. While I find the expected significant results for the CCI, no significant effects on resource adjustments can be found for the BCI. A possible explanation for this result is that the BCI rather reflects sentiment towards current conditions while the consumer sentiment indices are more future-oriented.

12 Another widely recognized index is the Conference Board Consumer Confidence Index, for which data is not freely available. However, the results from both indices are considered to be largely similar (Doms & Morin, 2004).

13 Data is publicly available online: http://www.sca.isr.umich.edu/tables.html. More precisely, the index is published twice a month because the University of Michigan publishes preliminary results in the middle of the month and the final results at the end of the month.

14 I report only the short versions of the questions. For the exact wording, please refer to the Appendix.

15 This is why the index is normalized to have a benchmark value of 100 in the first quarter of calendar year 1966.

16 Evidence and theories from psychology and economic behavior literature suggest that people are more sensitive to general changes than absolute levels when making decisions (e.g., Helson, 1964; Rabin, 1998). Changes in consumer sentiment, therefore, should be the relevant driver of managerial expectations, and thus short-term resource adjustment decisions. For instance, high but constant levels of consumer sentiment over several periods should not reflect new information about future demand, which is why they are already considered in prior decisions, and consequently current levels of resources. Actual changes in consumer sentiment compared with prior reference points, instead, constitute new information that have an important psychological effect on expectations and ongoing decisions. Similarly, cost stickiness literature uses the percentage growth in GDP instead of the absolute level when controlling for managerial expectations based on current/past economic developments (e.g., Anderson et al., 2003; Dierynck et al., 2012).

17 The Michigan Consumer Sentiment Index is not seasonally adjusted (Bram & Ludvigson, 1998). Conclusions remain robust when I calculate changes between the current (q) and previous quarter (q-1).

18 Instead of using SG&A costs, I also follow alternative studies that deploy total operating costs (e.g., Banker et al., 2013; Kama & Weiss, 2013). Furthermore, instead of using more granular quarterly data to examine the effect of consumer sentiment on short-term resource adjustment decisions, I also test whether or not the results remain robust when subjected to a model, where I compare annual changes in sales, costs and consumer sentiment. For both modifications, the main conclusions remain unchanged.

19 The sample period starts in 2000 since previous quarterly financial data is available only sporadically.

20 More precisely, all continuous control variables are mean-centered (conditional on periods of increasing and decreasing sales) and all sentiment measures used throughout this study are standardized (this makes it easier to compare the strength of the coefficients). This approach generally follows prior literature (e.g., Banker et al., 2013; Chen et al., 2012) and does not affect the estimates of the relation between consumer sentiment and cost behavior.

21 As an additional, untabulated test, I use more granular data to investigate the relationship between consumer sentiment and resource adjustment decisions. For my main analyses, I use a quarterly sentiment index, but sentiment might also differ from region to region. Accordingly, I repeat my main analysis using a region-specific measure based on the four regions defined by the US Census Bureau (i.e., Northeast, Midwest, South, and West). Results remain robust for this more granular sentiment measure (${\text{β}}_3$ = −0.032 (t = −3.18); ${\text{β}}_9$ = 0.005 (t = 0.92)). I acknowledge that a division-, state-, or even county-specific sentiment measure would be more preferable but the University of Michigan provides additional information at the region-level only.

22 Calculating expected changes in SG&A costs based on the level of consumer sentiment by considering the sentiment dummy and mean changes in sales conditional on sales increasing/decreasing periods reveals the following interpretation of the economic significance of consumer sentiment: If consumer sentiment is high (low) and sales are increasing (mean sales: 0.163), SG&A costs increase by 10.845% (10.622%). When sales are decreasing (mean sales: −0.145) and sentiment is high (low), SG&A costs decrease by 4.622% (5.311%).

23 For this regression model where both subcomponents are included together, I find a negative and significant effect for the ICC on SG&A costs when demand is growing, which is difficult to interpret. Moreover, the coefficient ${\text{β}}_3$ on the incremental impact of consumer sentiment on SG&A for sales decreases is flipping signs compared to the model where only the ICC is considered (column 1). This might indicate some multicollinearity issues but the variance inflation factors (VIFs) are consistently below the critical level of 10 for this model.

24 In this and all following cross-sectional analyses, I test the equality of the coefficients across the subsamples by estimating a stacked regression model for the full sample where all independent variables in model (2) are additionally interacted with the respective binomial indicator variable (in this case the median split dummy for $FCF$).

25 In an untabulated test, I have also considered employee intensity as an alternative proxy for adjustment costs but the results are similar to the test with asset intensity.

26 In the computation of the unused resources indicator, I use seasonally adjusted SG&A costs and sales to capture quarterly changes which could not be expected due to seasonal trends.

27 I have also differentiated between the magnitudes of prior seasonal sales changes (untabulated). In line with Ciftci and Zoubi (2019), sales changes are defined as large if they have changed by more than 30%, medium if they have changed between 10% and 30%, and small if they have changed by up to 10%. Results suggest that with higher magnitudes of prior sales increases, managers seem less likely to rely on consumer sentiment-based expectations; instead, they seem to focus more on firm-specific prior sales trends to extrapolate future developments. Moreover, results suggest that managers are particularly reticent to increase resources in response to a current sales increase when firms have faced a large decrease in the prior period, which is in line with the assumption that they are more pessimistic about the future based on past sales trends even if forward-looking consumer sentiment is positive.

28 Periods with high uncertainty are for example the quarters following the September 11 attacks or the bankruptcy of Lehman Brothers in September 2008.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.