EconBase
← Back to paper

Forecasted Treatment Effects

Extracted main text — title through conclusion, appendix excluded. This is what our citation measures are computed over, published so the extraction can be checked by eye.

89,911 characters · 17 sections · 25 citation commands

Rendered from LaTeX for readability, not typeset faithfully. Citation keys are highlighted; maths is left as source; figures, tables and equation environments are summarised rather than reproduced; unrecognised commands are greyed out so nothing is silently dropped. Email addresses are removed.

Forecasted Treatment Effects with Short Panels

\def\spacingset#1{ {#1}} \spacingset{1}

\if11 \fi

\if01 {

center[center omitted — 76 chars of source]

} \fi

abstractWe consider estimation and inference of the effects of a policy in the absence of an untreated or control group. We obtain unbiased estimators of individual (heterogeneous) treatment effects and a consistent and asymptotically normal estimator of the average treatment effect. Our estimator averages, across individuals, the difference between observed post‑treatment outcomes and unbiased forecasts of their counterfactuals, based on a (short) time series of pre-treatment data. The paper emphasizes the importance of focusing on forecast unbiasedness rather than accuracy when the end goal is estimation of average treatment effects. We show that simple basis function regressions ensure forecast unbiasedness for a broad class of data generating processes for the counterfactuals. In contrast, forecasting based on a specific parametric model requires stronger assumptions and is prone to misspecification and estimation bias. We show that our method can replicate the findings of some previous empirical studies but it does so without using an untreated or control group.

{\it Keywords:} Polynomial regressions; Forecast unbiasedness; Counterfactuals; Misspecification; Heterogeneous treatment effects

\spacingset{1.6}

Introduction

Evaluating the effect of a policy or a treatment usually requires comparing treated and untreated units. Standard approaches such as difference-in-differences (DiD) or two-way fixed effects (TWFE) regressions rely on outcomes of untreated units to identify the counterfactual outcomes of treated units. These methods rely on assumptions such as unconfoundedness or parallel trends (e.g., GoodmanBacon, CallawaySantAnna, SunAbraham). They cannot be applied when all units receive the treatment.

This situation is common. Examples include national tax reforms, health programs, and environmental regulations that affect all individuals or regions at the same time. Even when adoption is staggered, the last periods after full treatment do not contain untreated units, so conventional DiD or event-study designs cannot recover treatment effects for those periods. In such cases, researchers often turn to structural models or to simple before–after comparisons. These settings call for methods that make the required assumptions explicit and allow researchers to evaluate their plausibility.

This paper studies how to estimate treatment effects when no valid control group exists. We propose a transparent and practical approach based on forecasting each treated unit’s counterfactual outcome from its own pre-treatment data using basis function regressions. The key idea is that in short panels, the property required for identification is forecast unbiasedness rather than forecast accuracy. Forecasting methods in time-series and machine learning aim to minimize prediction error and often allow some bias to reduce variance. In policy evaluation, this objective is not appropriate. An unbiased forecast of the counterfactual outcome guarantees an unbiased estimate of the treatment effect, while a forecast that is accurate but biased does not. Of course, our approach relies on its own identifying assumptions, which we make explicit, but they differ from those used in designs that compare treated and untreated units.

Our approach, the Forecasted Average Treatment effect (FAT) estimator, computes the average treatment effect on the treated as the average difference between the observed post-treatment outcome and an unbiased forecast of the outcome in the absence of treatment. What matters is that the forecasting rule produces unbiased predictions on average across individuals.

The paper makes three contributions. First, we develop a simple and general framework for estimating treatment effects when no untreated units are available. Under appropriate assumptions, the FAT estimator is consistent and asymptotically normal even when the time dimension is short, the panel is unbalanced, and treatment effects are heterogeneous.

Second, we characterize a broad class of data generating processes (DGPs) under which the required forecast unbiasedness condition holds. When individual counterfactual outcomes can be generally expressed as the sum of (at most) three unobserved components - a stationary process, a unit root process, and a deterministic trend - unbiased forecasts can be obtained by regressing pre-treatment outcomes on basis functions of time, such as low-order polynomials. This result allows for DGPs with fixed effects, heterogeneous autoregressive parameters, and non-stationary trends, without requiring a detailed specification of the stochastic component. The framework also accommodates heterogeneous dynamic processes, such as unit-specific autoregressive parameters, which are difficult to handle with standard short-panel methods. We call our proposed approach “Unobserved-Components FAT”.

Third, we compare our general “Unobserved-Components FAT” approach with what we refer to as “Parametric-Model FAT”, where the researcher specifies and estimates a particular parametric model such as an AR(1) to forecast counterfactuals. This approach relies on stronger assumptions and can perform poorly in short panels because of misspecification or estimation bias. In contrast, our regression-based approach that focuses directly on forecast unbiasedness is robust to specification errors and is not affected by the incidental parameter problem.

The FAT estimator can also be used when untreated units exist but are imperfect comparators. In such cases, FAT can provide a robustness check for studies that rely on, e.g., parallel-trends assumptions. The framework complements recent work on DiD and event-study designs (e.g., deChaisemartinX2020, CallawaySantAnna, SunAbraham) by providing an alternative benchmark based on explicit assumptions about the individual-specific counterfactual process rather than on cross-group comparability.

Crucially, the approach based on FAT is advantageous when outcomes exhibit dynamics. Recent work MarxTamerTang, Klosin2024, BotosaruLiu2025, Cornwall2025 has shown that traditional TWFE and dynamic panel estimators often yield biased or negatively weighted estimates of the treatment effect in the presence of serial correlation. The solutions proposed in this literature typically rely on the homogeneity of the persistence parameters across units. By construction, FAT avoids conflating structural treatment effects with outcome persistence and is robust to feedback mechanisms Bonhomme2025, while allowing for a much richer class of models, which include heterogeneous forms of serial correlation, where the degree of outcome persistence can vary arbitrarily across individuals.

This paper relates to several strands of literature. It is thematically related to work on causal inference in settings where contemporaneous untreated comparison groups are not the primary source of identification, and where time-series structure plays a central role in constructing counterfactual outcomes from pre-treatment data. These include Bayesian approaches to causal forecasting (e.g., Brodersen2015); (comparative) interrupted time-series methods, in which untreated potential outcomes are modeled using finite-dimensional parametric mean functions in time—typically linear or piecewise-linear trends with level and/or slope changes at the intervention date (e.g., Schochet2022,Bernaletal2017,BrownWarner1985); and methods based on data-driven aggregation to purge unobserved aggregate confounders in designs with aggregate shocks (e.g., ArkhangelskyKorovkin2023).

The FAT framework differs from these strands along three dimensions. First, relative to Bayesian causal forecasting, FAT is designed to deliver forecast-unbiased counterfactuals under high-level restrictions on the untreated outcome process and does not require the researcher to commit to a fully specified likelihood or prior. Second, relative to interrupted time-series methods, FAT strictly nests the canonical segmented-regression mean specifications while allowing for richer latent evolution, including stochastic trend components and heterogeneous unit-specific dynamics. Third, relative to the aggregate-shock/aggregation literature, FAT does not take identification from exogenous aggregate shocks or instruments, nor does it rely on cross-sectional weighting to eliminate unobserved aggregate confounding; instead, it is a short-panel forecasting approach in which identification is driven by the internal time-series information in treated units’ pre-intervention histories.

In addition, the paper contributes to the panel-data literature on heterogeneous dynamic processes and random coefficients (e.g., Chamberlain1992,ArellanoBonhomme2012,GrahamPowell2012). For example, the projection estimator in ArellanoBonhomme2012 can be interpreted as a special case of FAT under specific restrictions on the latent outcome process. In particular, for a unit-root process without drift, the Arellano–Bonhomme forecast corresponds to an equal-weighted average of pre-treatment outcomes, whereas the FAT polynomial forecast assigns greater weight to more recent observations. When deterministic trends are present, the projection approach in ArellanoBonhomme2012 requires correct specification of the trend component, while FAT only requires the choice of a sufficiently rich family of basis functions, such as low-order polynomials. Therefore, FAT delivers unbiased counterfactual forecasts under weaker restrictions on the trend component, while retaining the projection-based intuition underlying existing random-coefficient panel estimators.

Our analysis also relates to work on forecast unbiasedness in time-series models (e.g., FullerHasza, Dufour1984) and to the dynamic-panel literature on bias correction and consistent estimation with short time dimensions (e.g., AngristPischke,BlundellBond1998). Our approach differs by showing that consistent treatment effect estimation in short panels can be achieved through unbiased forecasts rather than through the specification of a full stochastic model.

The rest of the paper is organized as follows. Section (ref) presents the framework and the FAT estimator. Section (ref) extends the method to include covariates and discusses forecasting based on a parametric model. Section (ref) reports simulation results. Section (ref) provides an empirical illustration. Section (ref) concludes.

Baseline case: Unobserved-Components FAT

In this section we develop our baseline “Unobserved-Components FAT” estimator for the average treatment effect on the treated (ATT) under universal treatment. We use individual pre‑treatment time series to forecast each unit’s post‑treatment counterfactual outcome and then average, across units, the difference between the observed outcome and its forecast. We first establish consistency and asymptotic normality of this estimator under a high‑level forecast‑unbiasedness condition, and then show that this condition holds for a broad class of unobserved‑components DGPs when counterfactuals are forecast using basis‑function regressions.

Parameter of interest and estimator

Consider a treatment or a policy that is implemented at a time $\tau$.\footnote{In Section (ref), we discuss individual-specific timing.} Here, the treatment affects all individuals in the population at the same time, so that the treatment indicator of individual $i$ at time $t$ is given by $$d_{it}:=\ 1(t>\tau) \text{ for all } i=1,\dots, n.$$ We adopt the potential outcomes framework with each individual $i$ having two potential outcomes at each time $t$: $y_{it}(1)$ if the individual is exposed to the treatment and $y_{it}(0)$ if the individual is not exposed to the treatment. Due to the absence of an untreated group in our setting, we will henceforth simply refer to $y_{it}(0)$ as the “counterfactual".

Under the stable unit treatment value assumption (SUTVA), the observed outcome of individual $i$ at $t$ is:

align*[align* omitted — 62 chars of source]

When all individuals are treated after time $\tau$, we have

align[align omitted — 175 chars of source]

We follow the literature on heterogeneous treatment effects in defining the ATT $h\geq 1$ periods after $\tau$ as:

align[align omitted — 208 chars of source]

where we used that $y_{i\tau+h}(1)=y_{i\tau+h}$ for $h\geq1$.\footnote{Note that with identically and independently distributed data across $i$, the right hand side of (ref) reduces to the conventional $\mathbb{E}\left[y_{i\tau+h}-y_{i\tau+h}\left(0\right)\right]$.}

The challenge in identifying and estimating ${\rm ATT}_h$ is that the counterfactual $y_{i\tau+h}\left(0\right)$ is not observed for $h \geq 1$. The conventional approach in the presence of an untreated group is to impose sufficient assumptions that identify the parameter of interest from the observed post-treatment outcomes of the untreated group. In the absence of an untreated group, we exploit pre-treatment individual time series to obtain a forecast for $y_{i\tau+h}\left(0\right)$. We denote this forecast by $\widehat{y}_{i\tau+h}(0)$.

We call our proposed estimator for ${\rm ATT}_{h}$ the Forecasted Average Treatment effect estimator (FAT), defined as:

align[align omitted — 136 chars of source]

where $\widehat{y}_{i\tau+h}(0)$ is a measurable function of past outcomes $\left\{y_{it}\right\}_{t\leq \tau}$.\footnote{In the baseline case, the individual forecast depends only on the past outcomes of the treated, in particular, there are no covariates in the information set.} We explain how to obtain the forecast $\widehat{y}_{i\tau+h}(0)$ below. For now, we note that $\widehat{y}_{i\tau+h}(0)$ uses individual-specific pre-treatment outcomes, which naturally accommodates unbalanced panels and heterogeneous treatment effects.

We make the following high-level assumptions.

assumption[Average unbiasedness] The forecast for time $\tau+h, \; h\geq1,$ is unbiased on average, in the sense that: \begin{align} \frac{1}{n}\sum_i\mathbb{E}\left(\widehat{y}_{i\tau+h}\left(0\right)-y_{i\tau+h}\left(0\right)\right)=0. \end{align}

Note that Assumption (ref) guarantees that $\mathbb{E}(\widehat{{\rm FAT}}_{h}) = {\rm ATT}_{h}$ so that ${\rm ATT}_{h}$ is identified.\footnote{The result follows trivially by writing ${\rm ATT}_{h} =\frac{1}{n}\sum_i\mathbb{E}\left[y_{i\tau+h}-\widehat{y}_{i\tau+h}(0) + \widehat{y}_{i\tau+h}(0) - y_{i\tau+h}\left(0\right)\right].$}

Let $u_{i\tau+h}:=y_{i\tau+h}-\widehat{y}_{i\tau+h}(0)$ be the forecasted individual treatment effect at $\tau+h, \; h\geq 1$.

assumption[CLT] Let $\{u_{i\tau+h}\}$ be a sequence of random variables that satisfies a CLT: \begin{equation} \frac{\frac{1}{\sqrt{n}}\sum_i \left( u_{i\tau+h} { - \mathbb{E}u_{i\tau+h}} \right) }{\bar{\sigma}_n} \Rightarrow{\cal N}\left(0,1\right), \end{equation} where $\bar{\sigma}^2_n:= {\rm Var}(\frac{1}{\sqrt{n}} \sum_i u_{i\tau+h})<\infty$.

For example, when $\{u_{i\tau+h}\}$ is a sequence of cross-sectionally independent (but not identically distributed) random variables, Theorem 5.11 in White gives an asymptotic normality result.

What Assumptions 1 and 2 fundamentally rule out is shocks that affect all individuals after the treatment and that are unforecastable.\footnote{In principle, the assumptions allow for some common shocks to be captured by the method used to forecast the counterfactuals. For example, if the common shocks are deterministic and can be modeled as polynomials, our baseline method will be able to account for it.} This is an example of the assumptions that any method for treatment effect evaluation must inevitably make in the absence of a valid control group. The assumptions are more likely to be valid the higher the frequency at which the data is measured and the closer to the treatment date one evaluates the effect (i.e., they more plausibly hold for monthly data and for effects evaluated one month after the treatment than for yearly data and effects evaluated several years after the treatment). Below we provide some discussion of how the assumption of no common shocks after the treatment could be weakened. As one would expect, the assumption can be relaxed when there is a group of untreated individuals that are subject to the same unforecastable shock, as we discuss in the Online Appendix (Section (ref)). Perhaps less obviously, the assumption could also be relaxed in the absence of an untreated group, as long as we observe another variable for the treated units that is subject to the same unforecastable shock but not subject to the treatment (see the discussion in Section (ref)).

lem[Consistency and asymptotic normality] For each $i=1,\ldots,n$, let the forecast $\widehat{y}_{i\tau+h}(0)$, $h\geq 1$, be a function of $\{y_{it}\}_{t\leq\tau}$. Let Assumptions (ref) and (ref) hold. Then\footnote{ Estimation of the variance $\bar{\sigma}_n^2$ is discussed in Appendix (ref). } \begin{align*} \frac{\sqrt{n}\left(\widehat{{\rm FAT}}_{h}-{\rm ATT}_{h}\right)}{\bar{\sigma}_n}\Rightarrow{\cal N}\left(0,1\right). \end{align*}

In the remainder of the paper, we provide low-level sufficient assumptions, including a full description of the class of DGPs for the counterfactuals $y_{it}(0)$ and the forecast methods which satisfy Assumption (ref).

Unbiased forecasts of counterfactuals

In this section, we characterize the class of DGPs for the counterfactuals. The need to discuss the DGP for counterfactuals arises because of the lack of an untreated group, which means that we must rely on forecasting counterfactuals from pre-treatment observations.

Stationary or stochastic trends DGPs

In this section, we consider a class of DGPs such that Assumption (ref) is satisfied generally, namely, by any forecast that can be written as a weighted average of pre-treatment outcomes with weights summing to 1. The DGPs in this class express the counterfactual as the sum of potentially two unobserved stochastic components. This includes a variety of processes, such as stationary and non-stationary (unit root) ARMA processes with individual-specific parameters.

assumption[Stationary or stochastic trends DGPs] The counterfactual $y_{it}(0)$ is: \begin{align} y_{it}(0) & = y_{it}^{(1)}(0)+ y_{it}^{(2)}(0), \end{align} where $y_{it}^{(1)}(0)$ is an unobserved mean-stationary process (i.e., $\mathbb{E}y_{it}^{(1)}(0)$ is constant over t) and $y_{it}^{(2)}(0)=y_{it-1}^{(2)}(0)+u_{it}(0)$ is an unobserved random walk process with $\mathbb{E}u_{it}(0)=0$ for all $t\geq2$. Either/both components could be zero.
remAssumption (ref) does not require both components to be present, which means that it accommodates stationarity as well as non-stationarity due to a stochastic trend. The user does not need to take a stance on the component(s). Our method is robust to both. When both components in Assumption (ref) are present, the assumption is equivalent to the classical trend-cycle decomposition of macroeconomic time series with stochastic trends (e.g., NelsonPlosser1982, Watson1986).
remThis class of DGPs is a plausible assumption for applications where either (1) the time series of pre-treatment outcomes does not display a trend; (2) there is a trend in pre-treatment outcomes that is plausibly stochastic (not deterministic); or (3) there is only one pre-treatment observation so a deterministic trend could never be modeled anyway. We consider processes with deterministic trends in pre-treatment outcomes in the next section.

A key insight of this paper is that a correctly specified parametric model (e.g., a specific ARIMA model) is not necessary to obtain unbiased forecasts of the counterfactuals. In fact, as the next result shows, any forecast expressed as a weighted average of pre-treatment observations satisfies the unbiasedness condition.

thm[Unbiasedness for stationary or stochastic trends DGP] Let Assumption (ref) hold. Denote by ${\cal T}_{i}=\left\{ \tau-R_{i}+1,\ldots,\tau\right\}$ the set of $R_{i}$ time periods directly preceding the treatment date. Consider a weighted average of the pre-treatment outcomes: \begin{align} \widehat{y}_{i\tau+h}(0)=\sum_{t\in{\cal T}_{i}}w_{it}y_{it}, \end{align} where $w_{it}$ are non-random weights such that $\sum_{t\in{\cal T}_{i}}w_{it}=1$. Then, \begin{align} \mathbb{E}\left[\widehat{y}_{i\tau+h}(0)-y_{i\tau+h}(0)\right] & =0. \end{align}
remNote that Theorem (ref) shows how to obtain unbiased estimates of the individual (possibly heterogeneous) treatment effects. The result in (ref) is stronger than the average unbiasedness required by Assumption (ref). This means that under the assumptions of Theorem (ref) one can not only obtain consistent and asymptotically normal estimates of the average treatment effects, but also unbiased estimates of the individual treatment effects.

There are many ways to obtain forecasts that are weighted averages of pre-treatment data. In this paper we focus on a general class of forecasts obtained via basis function regressions, such as polynomial time trends regressions.

definition[Forecasts via basis function regressions] Consider a sequence of linearly independent functions $\{b_k(t)\}_{k=0}^{q_i}$, $q_{i}\in\left\{0,1,2,\ldots,\tau-1\right\}$, on the set $\mathcal{T}_{i}=\left\{ \tau-R_{i}+1,\ldots,\tau\right\}$ with $R_{i}\in\left\{ q_i+1,\ldots,\tau\right\}$, and such that $b_0(t)=1$ for all $t$. For example, polynomial time trends set $b_k(t)=t^k$, with $q_{i}$ the order of the polynomial. For each individual $i$, we forecast the counterfactual via individual-specific regressions of pre-treatment outcomes $\{y_{it}\}_{t\in\mathcal{T}_i}$ on the basis functions $\{b_k(t)\}_{k=0}^{q_i}$: \begin{align} \widehat{y}_{i\tau+h}^{(q_i,R_i)} & :=\sum_{k=0}^{q_i}\widehat{c}_{ik}^{(q_i,R_i)}b_k(\tau+h),\\ \widehat{c_i}^{(q_i,R_i)} & :=\mathop{\mathrm{argmin}}_{c\in\mathbb{R}^{q_i+1}}\sum_{t\in{\cal T}_{i}}\left(y_{it}-\sum_{k=0}^{q_{i}}c_{k}\,b_{k}\left(t\right)\right)^{2}, \end{align} where $c_i=\left(c_{i0},\ldots,c_{iq_{i}}\right)$ is a $q_{i}+1$ vector of individual-specific coefficients.\footnote{Note that when $q_i = \tau - 1$, $R_i = \tau$, that is, all pre-treatment outcomes are used in constructing the forecast. However, fewer observations can be used. We discuss the choice of the tuning parameters $q_i$ and $R_i$ in Section (ref).}

This definition makes it clear that for any type of basis function the choice $q_i=0$ yields the sample mean of pre-treatment outcomes as the forecast. The following result shows that forecasts obtained via basis function regressions satisfy the weighted average requirement of Theorem (ref).

lemFor known basis functions $\{b_k(t)\}_{k=0}^{q_i},\,q_{i}=0,1,\dots,\tau_{i}-1$ that are linearly independent on $\mathcal{T}_{i}$ with $b_{0}\left(t\right)=1$, the forecast in equation (ref) satisfies equation (ref).

Our framework connects to the literature on random coefficient panel models, particularly Chamberlain1992, GrahamPowell2012, and ArellanoBonhomme2012. Linear panel models with unit-specific intercepts and trend coefficients, similar to specifications in this literature, belong to the class of DGPs in Assumption (ref) or Assumption (ref). However, our goals differ: that literature studies identification of the distribution of heterogeneous coefficients, whereas we focus on forecasting untreated potential outcomes to estimate average treatment effects.

Deterministic trends DGPs

In this section, we consider an expanded class of DGPs that is appropriate for applications where: 1) there is more than one pre-treatment outcome; 2) it makes sense to model the outcomes as trending over time; 3) the trend is deterministic rather than (or in addition to) stochastic. We show that the basis function regression considered in the previous section gives unbiased forecasts of the counterfactuals, under certain conditions.

The expanded class of DGPs always includes a deterministic trend component, possibly in addition to (either or both) the stochastic components considered in Assumption (ref).

assumption[Deterministic trend DGPs] The counterfactual $y_{it}(0)$ is: \begin{align} y_{it}(0) & = y_{it}^{(1)}(0)+ y_{it}^{(2)}(0)+y_{it}^{(3)}(0), \end{align} where $y_{it}^{(1)}(0)$ and $y_{it}^{(2)}(0)$ are as in Assumption (ref) and $y_{it}^{(3)}(0)$ is a deterministic time trend $y_{it}^{(3)}(0)=\sum_{k=0}^{q_{0i}}c_{ik}^{(3)}{b}_{k}(t)$ with $c_{i}^{(3)}\in\mathbb{R}^{q_{0i}+1}$ and known basis functions $\{b_k(t)\}_{k=0}^{q_{0i}}$, $q_{0i}\in\left\{0,1,2,\ldots,\tau-1\right\}$.

Theorem (ref) below clarifies when forecasts obtained via basis function regressions satisfy the unbiasedness assumption (Assumption (ref)) in the presence of deterministic trends.

thm[Unbiasedness for deterministic trend DGPs] Let Assumption (ref) hold. Then, \begin{align*} \mathbb{E}\left[\widehat{y}_{i\tau+h}^{(q_{i},R_{i})}(0)-y_{i\tau+h}(0)\right] & =0, \end{align*} where $\widehat{y}_{i\tau+h}^{(q_{i},R_{i})}(0)$ is defined in (ref). If $y_{it}^{(3)}(0)$ is not zero, it must be that $q_{i} \geq q_{0i}$.
remWhile the stochastic components of the counterfactual in Assumption (ref) are unobserved and do not have to be present, if the deterministic time trend component is present, it must be a function of the same basis functions used to obtain the forecast. This implies that the stochastic component of the DGP does not need to be correctly specified, but, if a deterministic time trend component is present, it is correctly specified up to the order (as we discuss in the next remark). This is an unusual result from the perspective of forecasting, where one typically focuses on specifying both stochastic and deterministic parts of a model.
remThe key requirement of Theorem (ref) is that $q_i$ (the number of basis functions used in estimation) be greater than or equal to $q_{0i}$ (the true number of basis functions), when the DGP has a deterministic time trend component. Intuitively, this means that, in the presence of deterministic trends, choosing a too small number of basis functions runs the risk of delivering biased forecasts of the counterfactuals. We discuss the practical implications of these findings when discussing the choice of tuning parameters in Section (ref).
remIt may be possible to weaken the assumption that there is no forecastable common shock between $\tau$ and $\tau+h, h\geq 1$. To see this, let $\widetilde{y}_{it}\left(0\right)$ follow either Assumption (ref) or (ref) and let $W_{it}$ be an observed variable that is affected by a common shock $\gamma_{t}$ but not by the treatment. Suppose that \begin{align} y_{it}\left(0\right) &= \gamma_{t}1\left(t>\tau\right)+\widetilde{{y}}_{it}\left(0\right),\\ W_{it} &= \theta_{i}^{W}\gamma_{t}1\left(t>\tau\right)+\epsilon_{it}. \end{align} Denote by $\widehat{\gamma}_t$ an estimator of $\gamma_t$ obtained from post-treatment data on $W_{it},$ $t>\tau$ via, e.g., PCA. Then \begin{align} \widehat{{\rm FAT}}_{h}&= \frac{1}{n}\sum_i \left(y_{i\tau+h}-\widehat{\widetilde{y}}_{i\tau+h}\left(0\right)\right)-\widehat{\gamma}_{\tau+h}, \end{align} where $\widehat{\widetilde{y}}_{i\tau+h}(0)$ is the counterfactual forecast obtained as before via basis function regression. Leveraging auxiliary information to partial out the effect of unobserved aggregate shocks is conceptually related to WhiteKennedy2009, Freyaldenhoven2019, BrownBW2023, among others.

Choice of basis functions and tuning parameters

Our proposed method for forecasting counterfactuals in Definition (ref) requires choosing: 1) the number of pre-treatment periods used for the estimation $R_i$; 2) the type of basis functions $b_k(t)$; and 3) the number of basis functions $q_i$. We discuss how these choices affect inference and offer some practical recommendations for empirical researchers.

The choice of estimation window, $R_i$, involves a trade-off. In our class of DGPs, a larger $R_i$ generally yields an estimator with smaller variance, but a shorter $R_i$ can guard against violations of our assumptions stemming from parameter instability in pre-treatment data. While plotting pre-treatment time series offers informal guidance on such instability, applying our method to long-T settings reveals a heightened sensitivity to the chosen estimation window. This is because the increased likelihood of structural changes and parameter instability over extended periods becomes a critical concern. This challenge resonates with discussions in the regression discontinuity literature regarding bandwidth selection (e.g., GelmanImbens2019).

Regarding the choice of basis functions, Theorems (ref) and (ref) imply that this choice only matters for unbiasedness when the DGP has a deterministic time trend, in which case the basis functions need to be correctly specified to ensure unbiasedness (up to the order). When the DGP is mean stationary or has a stochastic trend, the choice of basis functions does not matter for unbiasedness. Basis functions may be chosen based on the time series properties of pre-treatment outcomes. Polynomial time trends seem to be a natural choice of basis functions for DGPs with deterministic trends.\footnote{In applications using difference-in-differences (DiD) methods it is typical to assume the presence of time trends (mostly linear) that are common between control and treatment groups. Our results make it clear that a linear time trend can only be dealt with by either using an untreated group or, when an untreated group is not available, by using (at least two) pre-treatment time periods to model the trend (which leads to our polynomial regression).} Other basis functions could be used, e.g., periodicity could be captured by Fourier basis functions. Our practical recommendation, and what we focus on henceforth, is to consider polynomial basis functions by letting $b_k(t)=t^k$ in Definition (ref).\footnote{An advantage of polynomial basis functions is that it is in principle possible to relax the assumption that the basis functions are correctly specified by instead assuming that $y_{it}^{(3)}(0)$ in Assumption (ref) is a continuous (but otherwise unspecified) function of time. Since time in our setting is defined on a compact interval and the deterministic trend is a continuous function, $y_{it}^{(3)}(0)$ can be approximated arbitrarily well by a polynomial in time. In fact, by the Weierstrass Approximation Theorem, the approximation error approaches zero as the order of the polynomial goes to infinity. Under additional smoothness assumptions on the deterministic trend, an approximation theorem could then be used (e.g. the Polynomial Approximation Error Theorem) to derive a bound on the approximation error of $y_{it}^{(3)}(0)$ by the polynomial regression. The forecast of $y_{i\tau+h}(0)$ is biased, but we conjecture that it may be possible to do bias correction given an expression for the bias obtained via the approximation theorem.}

Regarding the choice of the order of the basis functions $q_i$ used for the estimation, this only matters for unbiasedness if the DGP has a deterministic trend component. For DGPs without such a trend, any $q_i$ ensures unbiasedness, with $q_i=0$ offering the lowest variance; however, larger $q_i$ can mitigate bias from non-stationary initial conditions. If a deterministic trend exists, $q_i$ cannot be smaller than the unknown true order, forcing a trade-off where larger $q_i$ for unbiasedness risks higher variance. This choice is inherently constrained by the number of pre-treatment periods ($T$): larger $T$ allows for more flexible polynomial forecasts and extensive placebo testing, yet estimation consumes periods from this budget; conversely, smaller $T$ drastically limits both model flexibility and the scope for placebo tests.

Standard cross-validation methods typically target predictive accuracy, not the unbiasedness crucial for our context. While bias-targeting cross-validation is a promising avenue for selecting $q_i$, its implementation faces challenges in our short-T panel context. It inherently requires long pre-treatment series for data splitting, which is often not feasible. Moreover, even with long-T data, this can heighten concerns about parameter instability (as stationarity becomes less defensible over extended periods). Therefore, our practical recommendation is to report results for a small range of $q_i$ values (e.g., $0,1,2,3$), with pre-treatment time series plots providing informal guidance.

Individual treatment timing and limited anticipation

The treatment timing $\tau_i$ can be individual-specific. While staggered adoption designs typically allow for identification via comparisons with not-yet-treated units, such strategies are inherently limited to the window before the last unit becomes treated. In settings where all units eventually adopt the treatment, such as the universal rollout of a national policy, standard event-study designs cannot recover treatment effects for the periods following full adoption, as no valid comparison group remains.

Our approach overcomes this limitation. Because FAT constructs counterfactuals using only the treated unit's own pre-treatment history, it does not rely on the existence of a contemporaneous control group. Consequently, it allows researchers to estimate treatment effects over longer horizons, extending into periods after all units have been treated.

We require that the treatment timing be exogenous with respect to the counterfactual outcome path $y_{it}(0)$, conditional on the unit's history used for forecasting. The basis functions $\{b_k(t)\}$ in Definition (ref) are then functions of the time to adoption $t - \tau_i$. Additionally, it is possible to allow for treatment anticipation, as long as it is limited. In this case, one can simply modify the pre-treatment estimation window ${\cal T}_{i}$ in Definition (ref) to include observations only up to the time $\tau_i-\delta_i$ at which it is still reasonable to assume that there was no treatment anticipation, that is, ${\cal T}_{i} \equiv \left\{ \tau_i-\delta_i-R_{i}+1,\ldots,\tau_i-\delta_i\right\}$ (and $h$ is adjusted accordingly).

Balanced panels and pooled estimation

For balanced panels, our proposed estimator is algebraically equivalent to two simpler pooled estimation strategies. These alternative approaches are also suitable for repeated cross-section data or when cohort of birth plays the role of time. However, in unbalanced panels, these simpler strategies may yield inconsistent estimates for ATT. Therefore, our exposition primarily focuses on individual-level estimators, given their ability to accommodate unbalanced panels; these same pooled estimators, however, are how the FAT is obtained in balanced panels.

Assume that $R=R_{i}$ and $q=q_{i}$ are constant across $i$ and focus on polynomial basis functions in Definition (ref). The first alternative way to obtain our proposed estimator is to consider the cross-sectional averages $\overline{y}_{t}=\frac{1}{n}\sum_{i=1}^{n}y_{it}$ of the observed outcomes in time period $t$. Due to linearity of the forecasting procedure, we can rewrite:

align[align omitted — 294 chars of source]

where ${\cal T}=\{\tau-R+1,\ldots,\tau\}$, and we suppress the dependence on $\tau$, $q$, $R$. Here, the cross-sectional averages for $t\leq\tau$ are used to obtain a forecast of the average counterfactual for $t=\tau+h$, which is then subtracted from the cross-sectional average observed at that time.

The second alternative is to consider a pooled regression estimator, $\widehat{{\rm FAT}}_{h}=\widehat{\beta}_{h}$, where

align[align omitted — 323 chars of source]

which is the OLS estimator obtained from regressing $y_{it}$ on a set of time dummies $1(t=\tau+k)$, for $k\in\{1,\ldots,h\}$, and individual-specific time trends.\footnote{It actually does not matter for $\widehat{\beta}$ whether the coefficients $\alpha$ on the time trend are individual-specific.}

Parametric-Model FAT

In this section, we consider an alternative approach that explicitly incorporates covariates, including lagged outcomes, in the estimation of FAT. Compared to the baseline case, this involves specifying a parametric model for the counterfactuals and imposing additional assumptions.

Homogeneous parameters

Consider the following parametric model for the counterfactual $y_{it}(0)$:

align[align omitted — 109 chars of source]

where $x_{it}\in\mathbb{R}^{\dim x_{it}}$ is a vector of covariates, $\beta\in\mathbb{R}^{\dim x_{it}}$ is restricted to be homogeneous across individuals, $c_{ik} \in \mathbb{R}$ are unknown parameters and $\varepsilon_{it}\in\mathbb{R}$ is such that

align[align omitted — 142 chars of source]

Assuming that we have estimates $\widehat{\beta}$ for the common parameter that are consistent as $n\rightarrow\infty$ under correct model specification\footnote{For example, when $q_{i}=0$ and $x_{it}=(y_{it-1},z'_{it})'$, a consistent estimator for $\beta=(\rho,\theta')'$ can be obtained by applying an IV regression to the first-differenced model

align*[align* omitted — 142 chars of source]

using, for example, $y_{it-2}$ and $z_{it-1}$ as instruments. In the Monte Carlo simulations, we further extend this case to $q_{i}=1$.} leads to the parametric-model FAT, defined as:

align[align omitted — 159 chars of source]

where $\widehat{y}_h(\widehat{\beta},y_i,x_i)$ is the forecast obtained as:

align[align omitted — 411 chars of source]

where $c_{i}=(c_{i,0},\ldots,c_{i,q_i})$ is a $q_i+1$ vector, and ${\cal T}_i=\{\tau-R_i+1,\ldots,\tau\}$ is the set of the $R_i$ time periods directly preceding the treatment date. The parameters $q_i$ and $R_i\in\{q_i+1,\ldots,\tau\}$ are chosen by the researcher.

thmConsider $\widehat{{\rm FAT}}_{h}^{\rm MB}$ in (ref). Assume that \begin{enumerate}[(i)] • The forecast is unbiased when evaluated at the true parameter value $\beta_0$, i.e., $$\mathbb{E}\left[\widehat{y}_h(\beta_0,y_i,x_i) - y_{i\tau+h}\left(0\right) \right] = 0.$$ • The function $\widehat{y}_h(\beta,y_i,x_i)$ is twice continuously differentiable such that $\frac{\partial \widehat{y}_h(\beta_0,y_i,x_i)} {\partial \beta}$ has finite second moments, and for some $\delta>0$ we have $$R_n := \sup_{\left\{ \beta \, : \, \left\| \beta - \beta_0 \right\| \leq \delta \right\}} \left\| \frac{1}{n} \sum_{i=1}^{n} \frac{\partial^2 \widehat{y}_h(\beta,y_i,x_i)} {\partial \beta \partial \beta'} \right\| = o_P(n^{1/2}).$$ • The estimator $\widehat{\beta}$ satisfies \begin{equation} \widehat{\beta} - \beta_0 = \frac 1 n \sum_{i=1}^n \psi(y_i,x_i) + r_n, \end{equation} where $\psi(y_i,x_i) $ has zero mean and finite variance, and $r_n= o_P(n^{-1/2})$. Together with assumption (i) this implies that $\widehat{\beta} - \beta_0 =O_P(n^{-1/2})$. • The sequence of random variables \begin{equation} u^*_{i\tau+h} := y_{i\tau+h} -\widehat{y}_h(\beta_0,y_i,x_i) - \frac{1}{n}\sum_{j=1}^n\mathbb{E}\left[\frac{\partial \widehat{y}_h(\beta_0,y_j,x_j)} {\partial \beta'}\right] \psi(y_i,x_i) \end{equation} satisfies a CLT in the sense that $$\frac{\frac{1}{\sqrt{n}}\sum_i \left(u^*_{i\tau+h}-\mathbb{E}u^*_{i\tau+h}\right)}{\bar{\sigma}^{*}_n}\Rightarrow{\cal N}\left(0,1\right), \quad \bar{\sigma}^{*2}_n := {\rm Var}\left(\frac{1}{\sqrt{n}}\sum_i u^*_{i\tau+h} \right)<\infty.$$ \end{enumerate} Then we have that $\sqrt{n}\frac{\widehat{{\rm FAT}}_{h}^{\rm MB}-{\rm ATT}_{h}}{\bar{\sigma}^{*}_n}\Rightarrow{\cal N}\left(0,1\right)$.\footnote{ Estimation of the variance $(\bar{\sigma}^{*}_n)^2$ is discussed in Appendix (ref). }

When the covariate vector includes lagged outcomes (e.g., $x_{it} = y_{i,t-1}$), the forecast in (ref) requires these lagged values to be observed. For $h=1$, this poses no problem since $y_{i\tau}$ is observed. For $h \geq 2$, however, the required intermediate counterfactuals $y_{i,\tau+1}(0), \ldots, y_{i,\tau+h-1}(0)$ are not observed under treatment, so iterated forecasting is not directly feasible. One solution is to adopt a local-projection approach that directly models the $h$-step-ahead relationship between $y_{i,\tau+h}(0)$ and $x_{i\tau}$, analogous to Jorda2005. Alternatively, one may use the Unobserved-Components FAT from Section (ref), which does not require dynamic covariates and remains feasible at all horizons.

Heterogeneous parameters

If the model for the counterfactuals is an AR(p) with heterogeneous parameters, the time series literature (e.g., FullerHasza,Dufour1984) has derived conditions under which forecasts from an individual AR(p) model are unbiased. The assumptions are stationarity of the initial condition and symmetry of the error term.

A second example is strictly exogenous covariates with heterogeneous coefficients. For example, suppose that the $h=1$ period forecast of $y_{i\tau+1}\left(0\right)$ is given by

align[align omitted — 517 chars of source]

The forecast (ref) is unbiased provided that a Vandermonde matrix which includes functions of $\left(\tau-R_{i}+s\right)^{j},j=0,\dots,q_{i},s=1,2,\dots,R_{i}$ and the covariates $x_{i\tau-R_{i}+s},s=1,2,\dots,R_{i},$ is invertible. This condition constrains how the covariates can change over time.

Simulation studies

In this section, we study the finite-sample performance of our estimators. Although we focus on a universal-treatment setting, the insights from the simulations carry over unchanged to staggered adoption, as linearity of the forecasting operator ensures unbiased individual forecasts aggregate to unbiased cohort forecasts.

In Section (ref), we compare the Unobserved-Components FAT (UC FAT) to the Parametric-Model FAT (PM FAT) across forecast horizons, sample sizes, autoregressive persistence, and alternative initial-condition regimes (stationary vs.\ nonstationary). PM FAT can inherit small-sample biases from first-stage estimation of the AR parameter, whereas UC FAT avoids these by construction; the contrast is most pronounced under high persistence and small $n$. When initial conditions are stationary, finite-sample biases are generally small across forecast horizons.

Section (ref) focuses on UC FAT and evaluates how tuning parameter choices - polynomial order $q$ and pre-treatment window $R$ - trade off bias and variance under DGPs satisfying Assumptions (ref) and (ref). In the small-$T$ setting we consider, the simulations support the recommendation to use the lowest polynomial order that accommodates any deterministic trend in the outcome and, conditional on that choice, select the largest feasible pre-treatment window. When the trend order is underfit, bias emerges; in that case, shortening the pre-treatment window can mitigate misspecification. Once the polynomial order is rich enough, bias is largely insensitive to $(q,R)$, and precision improves with lower order and longer windows. These patterns are robust across DGPs with heterogeneous autoregressive parameters and individual-specific trends, and they provide clear guidance for empirical implementation.

Performance across forecast horizons

We consider the following DGP for the counterfactual outcome: \[ y_{it}(0) \;=\; y^{(1)}_{it}(0) + y^{(3)}_{it}(0), \qquad t=1,\dots,T, \] where the autoregressive component has a homogeneous AR parameter \[ y^{(1)}_{it}(0) = \mu_i + \rho\, y^{(1)}_{i,t-1}(0) + u_{it}, \quad \mu_i \sim \mathcal U[-1,1],\quad u_{it}\sim\mathcal N(0,1),\quad \rho\in\{0.2,0.9\}, \] and the deterministic trend is homogeneous linear \[ y^{(3)}_{it}(0) = \delta_i t,\qquad \delta_i=1. \] We consider two regimes for $y^{(1)}_{i0}(0)$:

align*[align* omitted — 183 chars of source]

We work with a balanced panel, $T=8$ and $\tau=5$ (five pre-treatment periods), forecast horizons $h\in\{1,2,3\}$, and sample sizes $n\in\{50,1000\}$. Treatment effects at $\tau+h$ are zero by construction.

\paragraph{Estimators.} For $b_k(t)=t^k$, the UC forecast for unit $i$ at $\tau+h$ is \[ \widehat y^{\text{UC}}_{i,\tau+h}(0)=\sum_{k=0}^{q} \widehat c_{ik}(\tau+h)^k,\quad \widehat c_i=\arg\min_{c\in\mathbb R^{q+1}}\sum_{t=1}^{\tau}\Big(y_{it}-\sum_{k=0}^q c_k t^k\Big)^2, \] and our UC FAT estimator is

equation[equation omitted — 149 chars of source]

We compute (ref) using the fatEstimator package in R, with $R=q+1$ and $q\in\{0,1,2,3\}$.

The PM forecast first estimates $\rho$ by Anderson–Hsiao using $y_{i,t-2}$ as an instrument for $\Delta y_{i,t-1}$ in $\Delta y_{it}=\rho\,\Delta y_{i,t-1}+\Delta u_{it}$, then fits a polynomial to the residuals $y_{it}-\widehat\rho\,y_{i,t-1}$, so that: \[ \widehat y^{\text{PM}}_{i,\tau+h}(0)=\widehat\rho\,y_{i,\tau}+\sum_{k=0}^{q}\widehat c_{ik}(\tau+h)^k,\quad \widehat c_i=\arg\min_{c}\sum_{t=1}^{\tau}\Big(y_{it}-\widehat\rho\,y_{i,t-1}-\sum_{k=0}^{q} c_k t^k\Big)^2, \] with the PM FAT estimator given by:

equation[equation omitted — 148 chars of source]

We estimate $\rho$ using the ivreg package in R and then use the fatEstimator package on the residuals $y_{it}-\widehat\rho\,y_{i,t-1}$. The specification fits polynomials of degree $q\in\{0,1,2,3\}$ to $R=q+1$ pre-treatment periods.

\paragraph{Simulation scenarios.} We consider four scenarios, varying the cross-sectional size and the autoregressive parameter: \[ (n,\rho) \in \{(50,0.2),\ (50,0.9),\ (1000,0.2),\ (1000,0.9)\}. \]

We report bias and RMSE for UC and PM FAT across polynomial orders $q\in\{0,1,2,3\}$, forecast horizons $h\in\{1,2,3\}$, AR parameter magnitude $\rho\in\{0.2,0.9\}$, and sample size $n\in\{50,1000\}$, based on 500 Monte Carlo replications.

Tables (ref) and (ref) report bias and RMSE for the UC FAT for a DGP with initial condition drawn from the nonstationary distribution (Table (ref)) and from the stationary distribution (Table (ref)). With a nonstationary initial condition (Table (ref)), the $q=0$ specification - which omits the linear trend - exhibits large upward bias that grows roughly linearly with the forecast horizon $h$. Including a linear term ($q=1$) effectively removes this bias: the remaining deviations from zero are small across all horizons and values of $\rho$. Higher-order specifications ($q=2,3$) do not reduce asymptotic bias in this DGP (the true trend is linear), but they affect small-sample behavior: they can attenuate some initialization effects at the cost of substantially larger RMSE because of polynomial extrapolation. In all cases, RMSE increases with the forecast horizon.

With a stationary initial condition (Table (ref)), the $q=0$ specification still omits the linear time trend and exhibits a systematic upward bias. Once a linear term is included ($q=1$), the forecast specification matches the deterministic component of the DGP, and bias is essentially zero across horizons and values of $\rho$; the small nonzero entries in the table reflect finite-sample variability. Higher-order specifications ($q=2,3$) remain asymptotically unbiased but exhibit larger RMSE due to higher-order polynomial extrapolation. As expected, RMSE increases with $h$, while bias patterns are stable across $\rho$.

Tables (ref) and (ref) report bias and RMSE of PM FAT for a DGP with initial condition drawn from the nonstationary distribution (Table (ref)) and from the stationary distribution (Table (ref)). With a nonstationary initial condition (Table (ref)), the $q=0$ specification again omits the linear trend and generates a large upward bias that increases with the forecast horizon $h$. Introducing a linear term ($q=1$) removes this trend-misspecification bias at the population level, so that any remaining bias is driven by first-stage estimation of the AR parameter $\rho$. In finite samples, PM FAT inherits a small Nickell-type bias through $\hat{\rho}$, which propagates into the $h$-step forecast and yields modest biases even when $q=1$. Higher-order specifications ($q=2,3$) do not further reduce this asymptotic bias but amplify RMSE because polynomial extrapolation increases forecast variance, especially at longer horizons. As with UC FAT, RMSE rises with $h$, while the qualitative bias pattern is similar across values of $\rho$.

With a stationary initial condition (Table (ref)), the $q=0$ model still omits the linear trend and exhibits a systematic upward bias, though somewhat attenuated relative to the nonstationary case. Once a linear term is included ($q=1$), the forecasting specification matches the deterministic component of the DGP, and PM FAT is asymptotically unbiased; the small residual biases reflect the Nickell-type distortion from estimating $\rho$ in the first step. As before, higher-order specifications ($q=2,3$) remain asymptotically unbiased but exhibit larger RMSE due to variance inflation from higher-order polynomials. Compared to the nonstationary case, the stationary initial condition reduces the magnitude of the finite-sample biases, while RMSE continues to increase with $h$ and bias patterns remain stable across $\rho$.

table[table omitted — 2,569 chars of source]
table[table omitted — 2,718 chars of source]
table[table omitted — 2,610 chars of source]
table[table omitted — 2,646 chars of source]

Choice of tuning parameters for Unobserved-Components FAT

In this section, we compare the finite-sample performance of the Unobserved-Components FAT estimator across different tuning parameters: the polynomial order, $q$, and the estimation window, $R$, for different specifications of the DGP. All specifications satisfy Assumption (ref), with the counterfactual process specified as the sum of up to three different components.

Let $I_j, j=1,2,3$ be an indicator that equals to one whenever the associated components is present in the specification of the counterfactual process. For each $i=1,\dots,n$:

align*[align* omitted — 676 chars of source]

Note that the initial observation ${y}_{i0}^{(1)}$ is drawn from the stationary distribution and the time trend is linear and homogeneous, so it can be interpreted as a common shock.

Table (ref) shows results for the bias and standard error of $\widehat{\rm FAT}^{\rm UC}_1$ across different tuning parameters. The table shows that, when the DGP is mean stationary (first panel) or when it is the sum of a mean stationary and a random walk (second panel), the bias does not vary much across different values of the tuning parameters, while the standard errors are smaller for smaller $q$ and larger $R$. When the DGP contains a time trend component, we observe bias when the polynomial-order $q$ is less than the true order of the time trend, as the theory predicts. In this case, however, a smaller estimation window $R$ gives a smaller bias. When $q\geq 1$, the performance of the estimator in terms of bias is again robust to the choice of tuning parameters, with smaller standard errors for smaller $q$ and larger $R$.\footnote{We observe the same behavior with a sample size of $N=30$, see Table (ref) in the Online Appendix.}

table[table omitted — 4,411 chars of source]

\spacingset{1.9}

Empirical illustration

We consider an empirical exercise showing that FAT, despite not requiring an untreated control group, yields conclusions consistent with established methods that rely on a control group. This positions FAT as a valuable complementary robustness check when a control group is available, rather than a replacement for existing methods such as TWFE or DiD.\footnote{Additional replications are included in the Online Appendix (ref) and (ref).}

We consider the application in GoodmanBacon on the effect of no-fault laws on female suicide. The data are on U.S. states that adopted no-fault divorce laws from 1969 to 1985. The outcome is the age-adjusted average suicide mortality rate per million women (ASMR) in the state. We work with a balanced panel of $n=36$ treated states and 10 time periods, 5 of which are pre-treatment.

We assume the same tuning parameters $q$ and $R$ for FAT across states. To help select the basis functions and the tuning parameters, Figure (ref) plots the outcome variable averaged across states, as a function of time to adoption. The visible trending behaviour (plausibly reflecting a deterministic component) in the pre-treatment data suggests using a polynomial basis function and choosing orders $q>0$. There is no strong indication of structural instability in the pre-treatment data, so we use all available data for estimation, that is, we set $R=5$. We compute the forecast $\widehat{y}^{UC}_{i\tau+h}$ in (ref) for a number of forecast horizons $h$ and orders $q$. The Unobserved-Components FAT estimates at $\tau+h$ are shown in the top panel of Figure (ref) as a function of $h\in\{1,\dots,5\}$ and for $q\in\{1,2\}$. The variance of the estimator is computed as explained in Section (ref). The results show a statistically insignificant decrease in the suicide rate after adoption of the no-fault divorce laws across different forecast horizons. This replicates the findings in GoodmanBacon, ClarkeTapiaSchythe, but without using a control group. The standard error of our estimator is comparable to that obtained in the above-mentioned studies that employ event-study design approaches and, as expected, our $95\%$ confidence intervals are greater for longer forecast horizons (see, e.g., the results in Figure 2 in ClarkeTapiaSchythe).\footnote{To see how our estimator performs in terms of standard error with a small sample size, as in this application, see Table (ref) in the Online Appendix for simulation results with $N=30$.}

To validate the procedure, in Figure (ref) we report placebo FAT at lag $j$, $j\in\{0,\dots,3\}$, calculated as if the law was adopted $j$ years earlier than its actual adoption date (for each state). The forecast horizon is $h=1$. The figure shows that the placebo FATs for both polynomial orders are close to zero. This offers suggestive evidence that FAT estimates may be interpreted as ATT estimates.

figure[figure omitted — 234 chars of source]
figure[figure omitted — 507 chars of source]

Conclusion

This paper proposes an estimator of the average treatment effects in the absence of an untreated or control group, based on forecasting individual counterfactuals using basis function regressions over a (short) time series of pre-treatment data. Forecast unbiasedness is a key requirement that is satisfied by our approach under a broad class of DGPs that express the individual counterfactuals as the sum of up to three unobserved components: a stationary process, a stochastic trend, and a deterministic trend. The approach is robust, general, and flexible, allowing for unbalanced panels, heterogeneous treatment effects, and staggered treatment timing. Forecasting counterfactuals using a parametric model instead requires stronger assumptions and can perform poorly due to misspecification and estimation bias in small samples (e.g., incidental parameter problem).

\FloatBarrier

thebibliography\bibitem[\citeauthoryear{Abadie}{Abadie}{2021}]{Abadie2021} Abadie, A. (2021). \newblock Using synthetic controls: Feasibility, data requirements, and methodological aspects. \newblock {\em Journal of Economic Literature\/} {\em 59\/}(2), 391--425. \bibitem[\citeauthoryear{Angrist and Pischke}{Angrist and Pischke}{2009}]{AngristPischke} Angrist, J. and J. Pischke (2009). \newblock {\em Mostly harmless econometrics}. \newblock Princeton University Press. \bibitem[\citeauthoryear{Arellano and Bond}{Arellano and Bond}{1991}]{ArellanoBond1991} Arellano, M. and S. Bond (1991). \newblock Some tests of specification for panel data: Monte carlo evidence and an application to employment equations. \newblock {\em Review of Economic Studies\/} {\em 58\/}(2), 277--297. \bibitem[\citeauthoryear{Arellano and Bonhomme}{Arellano and Bonhomme}{2012}]{ArellanoBonhomme2012} Arellano, M. and S. Bonhomme (2012). \newblock Identifying distributional characteristics in random coefficients panel data models. \newblock {\em The Review of Economic Studies\/} {\em 79\/}(3), 987--1020. \bibitem[\citeauthoryear{Arkhangelsky and Korovkin}{Arkhangelsky and Korovkin}{2024}]{ArkhangelskyKorovkin2023} Arkhangelsky, D. and V. Korovkin (2024). \newblock On policy evaluation with aggregate time-series shocks. \newblock arXiv preprint arXiv:1905.13660. \bibitem[\citeauthoryear{Athey, Bayati, Doudchenko, Imbens, and Khosravi}{Athey et al.}{2021}]{athey2021matrix} Athey, S., M. Bayati, N. Doudchenko, G. Imbens, and K. Khosravi (2021). \newblock Matrix completion methods for causal panel data models. \newblock {\em Journal of the American Statistical Association\/} {\em 116\/}(536), 1716--1730. \bibitem[\citeauthoryear{Bachhuber, Saloner, Cunningham, and Barry}{Bachhuber et al.}{2014}]{Bachuber} Bachhuber, M., B. Saloner, C. Cunningham, and C. Barry (2014). \newblock Medical cannabis laws and opioid analgesic overdose mortality in the united states, 1999-2010. \newblock {\em JAMA Intern Med.\/} {\em 174\/}(10), 1668--1673. \bibitem[\citeauthoryear{Bai and Ng}{Bai and Ng}{2021}]{bai2019matrix} Bai, J. and S. Ng (2021). \newblock Matrix completion, counterfactuals, and factor analysis of missing data. \newblock {\em Journal of the American Statistical Association\/} {\em 116\/}(536), 1746--1763. \bibitem[\citeauthoryear{Bernal, Cummins, and Gasparrini}{Bernal et al.}{2017}]{Bernaletal2017} Bernal, J. L., S. Cummins, and A. Gasparrini (2017). \newblock Interrupted time series regression for the evaluation of public health interventions: a tutorial. \newblock {\em International Journal of Epidemiology\/} {\em 46\/}(1), 348--355. \bibitem[\citeauthoryear{Blundell and Bond}{Blundell and Bond}{1998}]{BlundellBond1998} Blundell, R. and S. Bond (1998). \newblock Initial conditions and moment restrictions in dynamic panel data models. \newblock {\em Journal of econometrics\/} {\em 87\/}(1), 115--143. \bibitem[\citeauthoryear{Bonhomme}{Bonhomme}{2025}]{Bonhomme2025} Bonhomme, S. (2025). \newblock Back to feedback: Dynamics and heterogeneity in panel data. \newblock arXiv preprint arXiv:2512.17576. \bibitem[\citeauthoryear{Borusyak, Jaravel, and Spiess}{Borusyak et al.}{2023}]{Borusyaketal2021} Borusyak, K., X. Jaravel, and J. Spiess (2023). \newblock Revisiting event study designs: Robust and efficient estimation. \newblock https://arxiv.org/pdf/2108.12419.pdf. \bibitem[\citeauthoryear{Botosaru and Liu}{Botosaru and Liu}{2025}]{BotosaruLiu2025} Botosaru, I. and L. Liu (2025). \newblock Time-varying heterogeneous treatment effects in event studies. \newblock arXiv preprint arXiv:2509.13698. \bibitem[\citeauthoryear{Brodersen, Gallusser, Koehler, Remy, and Scott}{Brodersen et al.}{2015}]{Brodersen2015} Brodersen, K., F. Gallusser, J. Koehler, N. Remy, and S. L. Scott (2015). \newblock Inferring causal impact using bayesian structural time-series models. \newblock {\em Annals of Applied Statistics\/} {\em 1\/}(9), 247--274. \bibitem[\citeauthoryear{Brown, Butts, and Westerlund}{Brown et al.}{2021}]{BrownBW2023} Brown, P. J., C. T. Butts, and J. Westerlund (2021). \newblock Difference-in-differences via common correlated effects. \newblock arXiv preprint arXiv:2109.12175. \bibitem[\citeauthoryear{Brown and Warner}{Brown and Warner}{1985}]{BrownWarner1985} Brown, S. and J. Warner (1985). \newblock Using daily stock returns: The case of event studies. \newblock {\em Journal of Financial Economics\/} {\em 14\/}(1), 3--31. \bibitem[\citeauthoryear{Callaway and Sant{'}Anna}{Callaway and Sant{'}Anna}{2021}]{CallawaySantAnna} Callaway, B. and P. H. C. Sant{'}Anna (2021). \newblock Difference-in-differences with multiple time periods. \newblock {\em Journal of Econometrics\/} {\em 225\/}(2), 200--230. \bibitem[\citeauthoryear{Chamberlain}{Chamberlain}{1992}]{Chamberlain1992} Chamberlain, G. (1992). \newblock Efficiency bounds for semiparametric regression. \newblock {\em Econometrica: Journal of the Econometric Society\/}, 567--596. \bibitem[\citeauthoryear{Clarke and Tapia-Schythe}{Clarke and Tapia-Schythe}{2021}]{ClarkeTapiaSchythe} Clarke, D. and K. Tapia-Schythe (2021). \newblock Implementing the panel event study. \newblock {\em The Stata Journal\/} {\em 21\/}(4), 853--884. \bibitem[\citeauthoryear{Cornwall and Wentland}{Cornwall and Wentland}{2025}]{Cornwall2025} Cornwall, G. and S. Wentland (2025). \newblock It's about time (series): A simple correction for difference-in-differences estimators. \newblock ssrn:5991734. \bibitem[\citeauthoryear{de Chaisemartin and D'Haultfoeuille}{de Chaisemartin and D'Haultfoeuille}{2020}]{deChaisemartinX2020} de Chaisemartin, C. and X. D'Haultfoeuille (2020). \newblock Two-way fixed effects estimators with heterogeneous treatment effects. \newblock {\em American Economic Review\/} {\em 110\/}(9), 2964--2996. \bibitem[\citeauthoryear{Dinas, Matakos, Xefteris, and Hangartner}{Dinas et al.}{2019}]{Dinas} Dinas, E., K. Matakos, D. Xefteris, and D. Hangartner (2019). \newblock Waking up the golden dawn: Does exposure to the refugee crisis increase support for extreme-right parties? \newblock {\em Political Analysis\/} {\em 27\/}(2), 244--254. \bibitem[\citeauthoryear{Dufour}{Dufour}{1984}]{Dufour1984} Dufour, J.-M. (1984). \newblock Unbiasedness of predictions from estimated autoregressions when the true order is unknown. \newblock {\em Econometrica\/} {\em 52\/}(1), 209--216. \bibitem[\citeauthoryear{Fern{\'a}ndez-Val, Freeman, and Weidner}{Fern{\'a}ndez-Val et al.}{2021}]{fernandez2020low} Fern{\'a}ndez-Val, I., H. Freeman, and M. Weidner (2021). \newblock Low-rank approximations of nonseparable panel models. \newblock {\em The Econometrics Journal\/} {\em 24\/}(2), C40--C77. \bibitem[\citeauthoryear{Freyaldenhoven, Hansen, and Shapiro}{Freyaldenhoven et al.}{2019}]{Freyaldenhoven2019} Freyaldenhoven, S., C. Hansen, and J. M. Shapiro (2019). \newblock Pre-event trends in the panel event-study design. \newblock {\em American Economic Review\/} {\em 109\/}(9), 3307--3338. \bibitem[\citeauthoryear{Fuller and Hasza}{Fuller and Hasza}{1980}]{FullerHasza} Fuller, W. and D. Hasza (1980). \newblock Predictors for the first-order autoregressive process. \newblock {\em Journal of Econometrics\/} {\em 13\/}(2), 139--157. \bibitem[\citeauthoryear{Gelman and Imbens}{Gelman and Imbens}{2019}]{GelmanImbens2019} Gelman, A. and G. Imbens (2019). \newblock Why high-order polynomials should not be used in regression discontinuity designs. \newblock {\em Journal of Business & Economic Statistics\/} {\em 37\/}(3), 447--456. \bibitem[\citeauthoryear{Goodman-Bacon}{Goodman-Bacon}{2021}]{GoodmanBacon} Goodman-Bacon, A. (2021). \newblock Difference-in-differences with variation in treatment timing. \newblock {\em Journal of Econometrics\/} {\em 225\/}(2), 254--277. \bibitem[\citeauthoryear{Graham and Powell}{Graham and Powell}{2012}]{GrahamPowell2012} Graham, B. S. and J. L. Powell (2012). \newblock Identification and estimation of average partial effects in “irregular” correlated random coefficient panel data models. \newblock {\em Econometrica\/} {\em 80\/}(5), 2105--2152. \bibitem[\citeauthoryear{Jord{\`a}}{Jord{\`a}}{2005}]{Jorda2005} Jord{\`a}, {\`O}. (2005). \newblock Estimation and inference of impulse responses by local projections. \newblock {\em American Economic Review\/} {\em 95\/}(1), 161--182. \bibitem[\citeauthoryear{Klosin}{Klosin}{2024}]{Klosin2024} Klosin, S. (2024). \newblock Dynamic biases of static panel data estimators. \newblock arXiv preprint arXiv:2410.16112. \bibitem[\citeauthoryear{Marx, Tamer, and Tang}{Marx et al.}{2024}]{MarxTamerTang} Marx, P., E. Tamer, and X. Tang (2024). \newblock Parallel trends and dynamic choices. \newblock {\em Journal of Political Economy: Microeconomics\/} {\em 2\/}(1), 129--171. \bibitem[\citeauthoryear{Nelson and Plosser}{Nelson and Plosser}{1982}]{NelsonPlosser1982} Nelson, C. and C. Plosser (1982). \newblock Trends and random walks in macroeconomic time series: Some evidence and implications. \newblock {\em Journal of Monetary Economics\/} {\em 10\/}(2), 139--162. \bibitem[\citeauthoryear{Schochet}{Schochet}{2022}]{Schochet2022} Schochet, P. Z. (2022). \newblock Statistical power for estimating treatment effects using difference-in-differences and comparative interrupted time series estimators with variation in treatment timing. \newblock {\em Journal of Educational and Behavioral Statistics\/} {\em 47\/}(4), 367--405. \bibitem[\citeauthoryear{Shover, Davis, Gordon, and Humphreys}{Shover et al.}{2019}]{Shover} Shover, C., C. Davis, S. Gordon, and K. Humphreys (2019). \newblock Association between medical cannabis laws and opioid overdose mortality has reversed over time. \newblock {\em PNAS\/} {\em 116\/}(26), 12624--12626. \bibitem[\citeauthoryear{Sun and Abraham}{Sun and Abraham}{2020}]{SunAbraham} Sun, L. and S. Abraham (2020). \newblock Estimating dynamic treatment effects in event studies with heterogeneous treatment effects. \newblock {\em Journal of Econometrics\/} {\em 225\/}(2), 175--199. \bibitem[\citeauthoryear{Watson}{Watson}{1986}]{Watson1986} Watson, M. (1986). \newblock Univariate detrending methods with stochastic trends. \newblock {\em Journal of Monetary Economics\/} {\em 18}, 49--75. \bibitem[\citeauthoryear{White}{White}{2001}]{White} White, H. (2001). \newblock {\em Asymptotic theory for econometricians}. \newblock Academic Press. \bibitem[\citeauthoryear{White and Kennedy}{White and Kennedy}{2009}]{WhiteKennedy2009} White, H. and J. Kennedy (2009). \newblock Retrospective estimation of causal effects through time. \newblock In J. Castle and N. Shephard (Eds.), {\em The Methodology and Practice of Econometrics: Festschrift in Honour of David F. Hendry}, pp.\ 409--420. Oxford: Oxford University Press. \bibitem[\citeauthoryear{Xu}{Xu}{2017}]{Xu} Xu, Y. (2017). \newblock Generalized synthetic control method: Causal inference with interactive fixed effects models. \newblock {\em Political Analysis\/} {\em 25\/}(1), 1--20.

\spacingset{1.9}