pith. sign in

arxiv: 2605.21358 · v1 · pith:OXUBEPXLnew · submitted 2026-05-20 · 💰 econ.GN · q-fin.EC

From Summer to Spring: A Shift in US Housing Market Seasonality

Yihan Hu , Cemil Selcuk This is my paper

Pith reviewed 2026-05-21 03:07 UTC · model grok-4.3

classification 💰 econ.GN q-fin.EC
keywords housing seasonalityresidential mobilitysearch and matchingUS housing marketprice cyclestransaction volumesthick market effects
0
0 comments X

The pith

A post-2021 shift in when households move explains the earlier spring peak in US housing prices and sales.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper investigates why US housing prices and sales, long peaking in summer, have shifted to strengthen in spring since 2021. It documents that residential mobility itself moved earlier in the calendar year after 2021, using SIPP survey data and supporting Google Trends evidence. The authors extend a search-and-matching framework to monthly frequency, establish equilibrium existence and uniqueness, and calibrate the model directly to the new mobility schedule. The calibrated version reproduces the observed earlier peaks in both prices and transaction volumes. This supports the conclusion that altered mobility timing alone suffices to explain the seasonality change.

Core claim

Using SIPP data the authors document a post-2021 shift in residential mobility toward spring months. They extend the search-and-matching model to monthly frequency, prove equilibrium existence and uniqueness, and calibrate it to the observed mobility patterns. The calibrated model reproduces the spring shift in both prices and transaction volumes, showing that the change in mobility timing alone accounts for the recent alteration in housing market seasonality.

What carries the argument

A monthly-frequency search-and-matching model in which higher-mobility periods generate thicker markets that raise equilibrium prices and transaction volumes.

If this is right

  • Housing seasonality can be understood as a direct reflection of mobility cycles rather than independent seasonal demand shifts.
  • Once mobility timing is accounted for, other factors such as interest-rate changes or supply constraints need not be invoked to explain the observed pattern alteration.
  • Forecasts of future housing cycles can be improved by tracking shifts in the seasonal distribution of household moves.
  • Policy interventions that affect the cost or timing of moving can be expected to alter the seasonal profile of prices and sales.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Policies that change moving costs or remote-work flexibility could indirectly reshape housing seasonality through their effect on mobility timing.
  • The same thick-market logic might generate seasonal patterns in related markets such as rental housing or local labor markets.
  • Testing the model against more granular, real-time mobility indicators would provide a sharper check on whether mobility remains the dominant driver.

Load-bearing premise

That the thick-market effects from changes in mobility timing dominate other post-2021 influences on seasonal housing patterns.

What would settle it

Observing no corresponding spring shift in mobility data while housing seasonality still moved earlier, or finding that the calibrated model fails to reproduce the price and volume changes when fed the actual post-2021 mobility schedule.

Figures

Figures reproduced from arXiv: 2605.21358 by Cemil Selcuk, Yihan Hu.

Figure 1
Figure 1. Figure 1: US housing market: real prices and sales (Zillow) [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Seasonal components of US house prices and sales (Zillow). Each series plots dm,T , the percentage deviation of month m in year T from that year’s annual mean. as dm,T = 100 ym,T − y¯T y¯T , (2) the percentage deviation of month m in year T from that year’s annual mean ¯yT . By the Frisch–Waugh–Lovell theorem, subtracting each year’s mean from yt prior to estimation is algebraically equivalent to projectin… view at source ↗
Figure 3
Figure 3. Figure 3: Seasonal components by month, pre- and post-2021 (Zillow). Each series plots the average of dm,T within the respective period. where mt denotes the calendar month of observation t, and we impose P12 m=1 µm = 0 for identification. We apply two complementary tests. The results are reported in [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Percentage of household moves by month: pre-2021 and post-2021. Figure 4a plots the average monthly share of annual moves for the pre-2021 period (2017–2019) and the post-2021 period (2021–2023). The broad hump shape is present in both series. The key difference, however, lies in the timing. Pre-2021, the distribution rises steeply into June, which accounts for 12.7% of all annual moves. Post-2021, June’s … view at source ↗
Figure 5
Figure 5. Figure 5: Raw monthly search interest and 12-month trend for eight moving-related keywords (Google Trends, US, 2010–2025). The dashed vertical line marks January 2021. 3.3 Google Trends Our third source is Google Trends search data. We obtain monthly search interest indices from Google Trends for eight moving-related keywords: moving truck, uhaul, atlas van lines, mayflower moving, moving supplies, moving help, addr… view at source ↗
Figure 6
Figure 6. Figure 6: Each cell is the post-2021 minus pre-2021 difference in average search intensity for that keyword and month. Red indicates stronger seasonal search activity post-2021, blue weaker. Summer months are consistently blue across all keywords, indicating weaker search activity in the post-2021 period relative to the pre-2021 period. Spring months, by contrast, show relatively stronger activity post-2021. and the… view at source ↗
Figure 7
Figure 7. Figure 7: Model-implied seasonal deviations from the annual mean (%), calibrated separately using SIPP monthly move shares (panels a–b) and the Google Trends moving truck search index (panels c–d). Left panels show prices, right panels show transaction volume. Each series is computed as the percentage deviation of the monthly equilibrium value from its annual mean, averaged over the pre-2021 and post-2021 periods. S… view at source ↗
Figure 8
Figure 8. Figure 8: X-13 ARIMA-SEATS seasonal components for Zillow real prices (left) and sales (right). Each series plots ˜dt = 100 St/Tt , the seasonal component as a percentage of the trend. X-13 decomposes an observed series yt into trend, seasonal, and irregular components. We apply the X-13 procedure separately to the real price and sales series, using the default specifications: additive components, no trading-day adj… view at source ↗
read the original abstract

The US housing market exhibits pronounced seasonal cycles: prices and sales rise through spring, peak in summer, and decline through autumn and winter. Since 2021, this pattern has shifted earlier in the calendar year, with spring strengthening at the expense of the traditional summer peak. A leading explanation for housing market seasonality is the search-and-matching model of Ngai and Tenreyro (2014), which links these cycles to household mobility through a thick-market mechanism. In this framework, periods with higher mobility generate thicker markets and higher prices and transaction volumes. Viewed through this lens, a shift in the seasonal cycle of prices and sales raises the question of whether the timing of household moves has changed. Did residential mobility shift earlier in the calendar year after 2021? We find that it did. Using SIPP data, and corroborating evidence from Google Trends indicators, we document a post-2021 shift in mobility toward spring. We extend the model to a monthly frequency, prove the existence and uniqueness of the equilibrium, and calibrate it to the observed mobility patterns. The calibrated model reproduces the spring shift in both prices and transaction volumes, consistent with the view that a change in the timing of household mobility alone can account for the recent shift in housing market seasonality.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper documents a post-2021 shift in US housing market seasonality toward an earlier spring peak in prices and transaction volumes. It shows a corresponding shift in residential mobility timing using SIPP data corroborated by Google Trends, extends the Ngai and Tenreyro (2014) thick-market search-and-matching model to monthly frequency with an existence and uniqueness proof for equilibrium, calibrates the model to the new mobility schedule, and reports that the calibrated model reproduces the observed spring shift in housing outcomes.

Significance. If the result holds, the paper supplies a parsimonious, mechanism-driven account of the recent seasonality change that builds directly on an established framework. The existence and uniqueness proof for the monthly equilibrium is a clear theoretical strength, and the dual data sources for mobility (SIPP and Google Trends) add credibility to the empirical premise. The work suggests mobility timing can be a first-order driver of seasonal housing fluctuations with implications for understanding thick-market effects more broadly.

major comments (2)
  1. [Calibration and Results] Calibration section: the model is calibrated directly to the post-2021 monthly mobility intensity parameters observed in SIPP and Google Trends data and then shown to reproduce the spring shift in prices and volumes. This structure renders the quantitative reproduction partly a restatement of the fitted mobility input rather than an independent test, weakening the claim that mobility timing alone accounts for the housing shift.
  2. [Empirical and Model Results] The manuscript does not report explicit robustness checks or horse-race exercises that shut down or condition on other post-2021 seasonal factors (mortgage-rate path, remote-work effects on location demand, or supply disruptions). Without such analysis, the attribution to mobility timing remains non-unique and the central claim that the thick-market mechanism is the dominant isolated driver is not fully established.
minor comments (2)
  1. [Abstract and Data] The abstract and data section would benefit from explicit statement of the exact SIPP sample construction, time window, and the precise Google Trends search terms or categories employed.
  2. [Model and Calibration] A summary table listing the monthly mobility parameters, their pre- and post-2021 values, and the calibrated matching-function parameters would improve transparency and replicability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. We address the major comments below and indicate the revisions we intend to make to strengthen the manuscript.

read point-by-point responses

  1. Referee: Calibration section: the model is calibrated directly to the post-2021 monthly mobility intensity parameters observed in SIPP and Google Trends data and then shown to reproduce the spring shift in prices and volumes. This structure renders the quantitative reproduction partly a restatement of the fitted mobility input rather than an independent test, weakening the claim that mobility timing alone accounts for the housing shift.

    Authors: We agree that the mobility schedule is an input calibrated from the data. The purpose of the exercise is to show that, given this input and the thick-market mechanism, the model generates the observed shift in housing outcomes. Other parameters are held fixed from the pre-2021 calibration or steady state. We will revise the calibration section to more clearly distinguish between the data input and the model's prediction of housing variables, emphasizing that this illustrates the quantitative importance of the mobility channel via the search-and-matching framework. revision: partial

  2. Referee: The manuscript does not report explicit robustness checks or horse-race exercises that shut down or condition on other post-2021 seasonal factors (mortgage-rate path, remote-work effects on location demand, or supply disruptions). Without such analysis, the attribution to mobility timing remains non-unique and the central claim that the thick-market mechanism is the dominant isolated driver is not fully established.

    Authors: This is a fair point. The paper's central claim is that the mobility timing shift is sufficient to generate the observed housing seasonality change through the thick-market effect, as demonstrated by the calibrated model. We do not claim it is the only or dominant factor to the exclusion of others. To address the concern, we will add a discussion section that considers alternative explanations such as changes in mortgage rates and remote work patterns, and argue based on timing and existing evidence why mobility is a key driver. A comprehensive horse-race analysis would require extending the model to include these factors explicitly, which we view as beyond the current scope but note as an avenue for future research. revision: yes

Circularity Check

1 steps flagged

Calibration to observed mobility patterns reproduces seasonality shift by construction

specific steps
  1. fitted input called prediction [Abstract]
    "We extend the model to a monthly frequency, prove the existence and uniqueness of the equilibrium, and calibrate it to the observed mobility patterns. The calibrated model reproduces the spring shift in both prices and transaction volumes, consistent with the view that a change in the timing of household mobility alone can account for the recent shift in housing market seasonality."

    Mobility timing is measured directly from post-2021 data and used as the primary exogenous seasonal input for calibration. The thick-market mechanism then maps higher mobility periods to thicker markets and higher prices/volumes by construction, so the reported reproduction of the spring shift in prices and volumes is a restatement of the fitted mobility schedule rather than an out-of-sample or independent result.

full rationale

The paper documents a post-2021 shift in residential mobility using SIPP and Google Trends, extends the Ngai-Tenreyro thick-market model to monthly frequency with an existence/uniqueness proof, calibrates the model to the new mobility schedule, and reports that it reproduces the earlier spring peak in prices and transactions. This structure makes the reproduction a direct consequence of feeding the observed mobility timing into the calibrated matching function rather than an independent prediction. The claim that mobility timing 'alone can account for' the shift therefore rests on the untested assumption that no other post-2021 seasonal factors interact with the model once mobility is conditioned on. No horse-race or explicit shutdown of alternative channels is described.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on the original Ngai-Tenreyro thick-market mechanism, the assumption that mobility timing is the primary seasonal driver, and calibration parameters chosen to match observed post-2021 mobility rates; no new entities are postulated.

free parameters (1)
  • monthly mobility intensity parameters
    Calibrated to match the post-2021 SIPP mobility schedule so that the model reproduces the observed housing seasonality shift.
axioms (2)
  • standard math Existence and uniqueness of equilibrium in the monthly-frequency search-and-matching model
    Stated as proved in the paper for the extended model.
  • domain assumption Thick-market externality links higher mobility periods to higher prices and transaction volumes
    Inherited from Ngai and Tenreyro (2014) and used to connect mobility timing to housing outcomes.

pith-pipeline@v0.9.0 · 5755 in / 1550 out tokens · 52849 ms · 2026-05-21T03:07:17.407613+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    American Economic Review , author =

    Hot and. American Economic Review , author =. 2014 , pages =. doi:10.1257/aer.104.12.3991 , language =

    work page doi:10.1257/aer.104.12.3991 2014

  2. [2]

    , journal =

    Lovell, Michael C. , journal =. 1963 , month = dec, doi =

    work page 1963

  3. [3]

    and Watson, Mark W

    Stock, James H. and Watson, Mark W. , journal =. 2008 , month = jan, doi =

    work page 2008

  4. [4]

    Journal of Economic Perspectives , year =

    Barrero, Jos\'. Journal of Economic Perspectives , year =

    work page

  5. [5]

    2025 , howpublished =

    work page 2025

  6. [6]

    Economics Letters , year =

    Selcuk, Cemil , title =. Economics Letters , year =

    work page

  7. [7]

    International Economic Review , year =

    D. International Economic Review , year =

    work page

  8. [8]

    Journal of Financial Economics , year =

    Gupta, Arpit and Mittal, Vrinda and Peeters, Jonas and Van Nieuwerburgh, Stijn , title =. Journal of Financial Economics , year =

    work page

  9. [9]

    and Wozniak, Abigail , title =

    Jia, Ning and Molloy, Raven and Smith, Christopher L. and Wozniak, Abigail , title =. Journal of Economic Literature , year =

    work page

  10. [10]

    Journal of Urban Economics , year =

    Krainer, John , title =. Journal of Urban Economics , year =

    work page

  11. [11]

    and Wozniak, Abigail , title =

    Molloy, Raven and Smith, Christopher L. and Wozniak, Abigail , title =. Journal of Economic Perspectives , year =

    work page

  12. [12]

    Mondragon, John A and Wieland, Johannes. 2022

    work page 2022

  13. [13]

    Real Estate Economics , year =

    Novy-Marx, Robert , title =. Real Estate Economics , year =

    work page

  14. [14]

    American Economic Review , year =

    Piazzesi, Monika and Schneider, Martin , title =. American Economic Review , year =

    work page

  15. [15]

    National Bureau of Economic Research , year =

    Ramani, Arjun and Bloom, Nicholas , title =. National Bureau of Economic Research , year =

    work page

  16. [16]

    and Shiller, Robert J

    Case, Karl E. and Shiller, Robert J. , title =. American Economic Review , year =

    work page

  17. [17]

    , title =

    Goodman, John L. , title =. Journal of Real Estate Research , year =

    work page

  18. [18]

    Contemporary Economic Policy , year =

    Malpezzi, Stephen , title =. Contemporary Economic Policy , year =

    work page

  19. [19]

    Real Estate , year =

    Hu, Yihan and Huang, Yifei , title =. Real Estate , year =

    work page

  20. [20]

    and Sah, Vivek and Sklarz, Michael and Pampulov, Stefan , title =

    Miller, Norman G. and Sah, Vivek and Sklarz, Michael and Pampulov, Stefan , title =. Journal of Housing Research , year =

    work page

  21. [21]

    Journal of Urban Economics , year =

    Genesove, David and Han, Lu , title =. Journal of Urban Economics , year =

    work page