2023年4月13日伟易博治理学院行为科学和政策干预交织立异团队2023春季学期第四次（总第十二次）分享会顺遂举行。本次分享会约请到哥本哈根大学经济系的Johannes Wohlfart副教授为各人带来题为“Home Price Expectations and Spending: Evidence from a Field Experiment”的研究分享，该研究由Wohlfart教授与哥本哈根大学经济系助理教授Felix Chopra及科隆大学经济系Christopher Roth教授相助开展，这几位学者曾在宏观行为纪律的微观基础研究方面多次相助，在本次分享的研究中，三位学者对房价预期给人们当下消耗行为的影响举行了研究。
分享人 The Speaker
Johannes Wohlfart is an Associate Professor at the Department of Economics and the Center for Economic Behavior and Inequality at the University of Copenhagen and a CESifo Research Network Affiliate. In his research, Dr Wohlfart uses observational and experimental data to work on questions in behavioral economics, behavioral finance and macroeconomics. In particular, his research focuses on belief formation and the role of beliefs in shaping economics decisions.
Dr Wohlfart completed his PhD in 2019 at Goethe University Frankfurt and holds an MPhil in Economics from the University of Oxford.
分享会 The Seminar
On April 13th, the GSM Behavioral Science in Action Seminar Series welcomed Dr. Johannes Wohlfart, an Associate Professor at the Department of Economics and the Center for Economic Behavior, and also a CESifo Research Network Affiliate. Dr. Wohlfart shared his recent research, "Home Price Expectations and Spending: Evidence from a Field Experiment", co-authored with Dr. Felix Chopra and Professor Christopher Roth. Their collaboration aims to experimentally test the micro-foundations of behavioral macroeconomic models.
The study was inspired by the observed significance of housing in households’ balance sheets. It was found that in 2020, housing comprised 28% of US households' net wealth. Concurrently, expectations about future house prices tend to experience substantial fluctuations and have been suggested to be influential drivers of business cycle movements. These expectations can impact aggregate outcomes through various mechanisms, including their effects on the housing market and households' spending decisions. While extensive literature has investigated households' spending responses to realized home price changes, this study distinctively delves into the effects of expected home price fluctuations on present-day spending behaviors. Key findings reveal that homeowners remain unaffected by changes in home price expectations, but renters tend to curtail their spending in anticipation of stronger home price growths.
Methodology/Design
Linking home price expectations to changes in household spending is theoretically plausible. Yet empirically, the study of home price expectations tends to encounter significant identification challenges: real-world economic datasets often lack observations on the subjective judgements of home price expectations, which are endogenously shaped by past price shifts and individual traits. Issues such as omitted variable bias, reverse causality, and substantial measurement errors in survey data can further complicate the statistical inference between home price expectations and spending behaviors.
To address the challenges, Dr. Wohlfart and coauthors devised a field experiment to generate exogenous variations in participants’ home price expectations through an informational intervention. The experiment recruited 2,554 household heads, drawn from a pool of approximately 100,000 Nielsen panel members. After each shopping trip, participants scanned the barcodes of their purchased items with a scanner provided by the panel and received rewards for logging their purchases. As the dataset also tracks real-time price and discount information on over a million consumer products, it contains information on both the overall non-durable expenditure of households and the composition of this expenditure. The dataset also includes panelists’ self-reported socio-demographic information such as income, age, and family structure.
Participants in this study were invited to partake in an online survey in November 2019 and then a follow-up survey four weeks later in December. The main survey started by asking participants how they thought the average annual home price in the U.S. would change over the next 10 years to elicit their prior beliefs. Then, the survey randomly assigned participants to either a high forecast treatment group, where they were shown an optimistic forecast of a 6% per annum home price growth rate, or a low forecast treatment group of 1.5% annual growth rate. To mitigate the potential effects of exponential growth bias, participants were also given the implied value of a $100,000 home in ten years based on the growth rate they were shown. All participants were uniformly informed that the rate of inflation would be 2.2% for the next 10 years to avoid the confounding effect of implicit inflation expectation formation. Subsequently, the survey elicited participants’ post-treatment beliefs about future home price changes and changes in borrowing constraints, as well as background information such as real-estate ownership. Four weeks after, the follow-up survey re-elicited the expectations to test the persistence of the treatment effect and to address concerns about demand effects and numerical anchoring. Participants were also asked about their durable goods expenditure in the past four weeks and whether they had any plans for home buying and/or selling. Households' non-durable purchases from August 2019 to February 2020 (i.e., three months before and after the treatment was administered in November 2019) were collected via the Nielsen panel and used for the analysis of treatment effects.
Results
The study discovered that the information treatments led to an average 1.5 percentage point divergence in the posterior home price growth expectations across treatment groups. This difference was found to be statistically significant and persistent after four weeks. the posterior difference between the groups was one third of the difference between the two signals (a 4.5% difference between the 1.5% growth rate and the 6% growth rate), therefore a moderate learning rate of one third was determined for the treatment.
Regarding the effect of home price expectations on household expenditure, the log of monthly household expenditures was regressed on the interaction of the treatment group (high forecast = 1) and timing (post-treatment = 1) dummies. Findings show that while homeowners did not statistically significantly respond to changes in home price expectations, a 1 percentage point increase in the expected home price growth rate had led to a 7.6% decrease in renters’ non-durable expenditures, which corresponded approximately to a $35 spending reduction. The treatment effects were statistically insignificant for the entire sample, but when stratifying the sample by home ownership and moving intentions, it was found that while homeowners in general do not respond to home price expectation changes, renters who plan to move in the next 10 years (especially those who plan to move to a more expensive home) responded significantly to home price expectations.
In terms of the interpretation of the magnitudes of renters’ responses, researchers estimated that renters in the high forecast group, if increased their savings by the estimated $35 a month, would end up with an expected cumulative savings increase covering 28.6% of the changes in the higher expected down payment. However, they also noted that the estimation should be taken with a pinch of salt as (1) it was based on a set of possibly unrealistic assumptions; (2) the Nelsen panel may not record all areas of spending (e.g. cancelling a family Netflix membership could easy increase the household’s monthly savings by $20).
Mechanisms
The researchers proposed three potential channels of influence for the effect of home price expectations on household expenditure.
1. Positive perceived wealth effects for homeowners: rising home price expectations make homeowners feel wealthier, potentially leading to higher current spending.
2. Negative effect of expected housing costs: both renters and homeowners may anticipate higher future living costs due to rising home prices, leading to curtailed spending.
3. Positive effect of changes in expected future borrowing constraints for homeowners: Rising home values might result in relaxed collateral constraints, thereby increasing homeowners’ current spending.
The study revealed that renters, under the influence of rising expected housing costs, would curtail their spending in the face of rising home price expectations. The effect is found to be particularly prevalent for renters who plan to buy homes in the near future. For homeowners, the favorable wealth effect and unfavorable higher expected housing costs offset each other and lead to no significant changes in their spending decisions. It was also found that the effect of relaxed borrowing constraints is negligible for homeowners. These mechanisms and their relative magnitudes coincide with views of popular financial advice websites.
In a separate survey where respondents were asked about their expectations of their economic situation after an increase in their home price expectation, homeowners’ responses were balanced between better economic situation, no change, and worse situation, although most homeowners noted that they would not change their planned current spendings. However, the majority of renters expect a worse economic situation following the rise in expected home prices, and many plan to cut current spending as a response to this change. Answers to some open-ended qualitative questions show that following a change in people’s home prices expectations, homeowners are most worried about changes in their wealth, while the cost of buying a home is the top concern on renters’ minds.
Macroeconomic Implications
Asset price expectation shifts can widen the consumption inequality between potential buyers and sellers of the asset. However, as renters’ spending reacts inversely to home price expectations, this could act as a counter-cyclical force in business cycles.
For more information on the study, please refer to the 2023 working paper by Chopra, Roth, and Wohlfart.
