Kripfganz Schneider 2023 Ardl Estimating Autoregressive Distributed Lag and Equilibrium Correction Models

The Stata Journal (2023)
23, Number 4, pp. 983–1019 DOI: 10.1177/1536867X231212434
ardl: Estimating autoregressive distributed lag

and equilibrium correction models
Sebastian Kripfganz
University of Exeter Business School
Exeter, U.K.
[email protected]
Daniel C. Schneider
Max Planck Institute for Demographic Research
Rostock, Germany
[email protected]
Abstract. We present a command, ardl, for the estimation of autoregressive

distributed lag (ARDL) models in a time-series context. The ardl command can
be used to fit an ARDL model with the optimal number of autoregressive and
distributed lags based on the Akaike or Bayesian (Schwarz) information criterion.
The regression results can be displayed in the ARDL levels form or in the error-
correction representation of the model. The latter separates long-run and short-run
effects and is available in two different parameterizations of the long-run (cointe-
grating) relationship. The popular bounds-testing procedure for the existence of a
long-run levels relationship is implemented as a postestimation feature. Compre-
hensive critical values and approximate p-values obtained from response-surface
regressions facilitate statistical inference.
Keywords: st0734, ardl, ardl postestimation, autoregressive distributed lag model,
error-correction model, bounds test, long-run relationship, cointegration, time-
series data
1 Introduction
Real-world phenomena are often characterized by complex relationships. Some observed
variables might exhibit erratic behavior in the short run but tend to comove in a stable
and predictable way over longer time horizons. Attempting to empirically uncover such
long-run equilibrium relationships is tantamount to separating them from the overlaid
short-run dynamics. This separation allows one to find evidence for or against an
equilibrium relationship, which is often at the heart of a research question. It also
allows analysis of the short-term fluctuations around the equilibrium, which can be
valuable in its own right, for example, when conducting forecasting exercises or dynamic
simulations.
When we observe the variables of interest over a sufficiently long stretch of consecu-
tive time periods, multiequation vector autoregressive (VAR) and vector error-correction
(VEC) models are commonly used to assess their dynamic relationships. When we have
© 2023 StataCorp LLC st0734

984 Autoregressive distributed lag model estimation
reasons to assume that there is a natural ordering of the variables such that there is no
contemporaneous feedback from a response variable to the other variables in the system,
a single-equation autoregressive distributed lag (ARDL) model can simplify the analysis
and facilitate more efficient inference.1
ARDL models have many possible applications. They are extensively used in studies
analyzing linkages of pollution and energy consumption to economic growth (Fatai,
Oxley, and Scrimgeour [2004]; Narayan and Smyth [2005]; Wolde-Rufael [2006]; Ang
[2007]; Halicioglu [2009]; Jalil and Mahmud [2009]; Zhang et al. [2015]; Ntanos et al.
[2018]; Bekun, Emir, and Sarkodie [2019]; Kirikkaleli, Güngör, and Adebayo [2022];
and many more). Relationships with economic growth have also been investigated for
foreign direct investment and trade (Oteng-Abayie and Frimpong 2006; Belloumi 2014),
infrastructure (Fedderke, Perkins, and Luiz 2006), immigration (Morley 2006), tourism
(Katircioglu 2009; Wang 2009; Song et al. 2011), stock market development (Enisan
and Olufisayo 2009), and health expenditures (Murthy and Okunade 2016).
Other examples include the nexus between viral infections and meteorological fac-
tors (He et al. 2017; Doğan et al. 2020), childcare availability, fertility, and female labor
force participation (Lee and Lee 2014), wages, productivity, and unemployment (Pe-
saran, Shin, and Smith 2001), savings and investment (Narayan 2005), exchange rates
and trade (Bahmani-Oskooee and Brooks 1999; De Vita and Abbott 2004), exchange
rates and monetary policy (Frankel, Schmukler, and Servén 2004; Shambaugh 2004;
Obstfeld, Shambaugh, and Taylor 2005), financial development and inequality (Ang
2010), bank lending and property prices (Davis and Zhu 2011), financial reforms and
credit growth (Adeleye et al. 2018), stock market efficiency and fiscal policy (Stoian
and Iorgulescu 2020), democracy and the shadow economy (Esaku 2022), and the inter-
dependencies among stock price indices and commodity prices (Narayan, Smyth, and
Nandha 2004; Sari, Hammoudeh, and Soytas 2010; Büyükşahin and Robe 2014), as well
as cryptocurrencies (Ciaian, Rajcaniova, and Kancs 2016, 2018), to list only a few.
Recently, the ARDL methodological toolkit was used extensively to analyze adjust-
ment processes during the COVID-19 pandemic, including tourism demand forecasts
(Zhang et al. 2021) and the effects on macroeconomic activity (Varona and Gonzales
2021) or energy consumption (Aruga, Islam, and Jannat 2020).
The ARDL model can be conveniently reparameterized in so-called error-correction
(EC) form, which disentangles the long-run relationship from the short-run dynamics.
When the variables are nonstationary—to be precise, integrated of order 1—the long-
run relationship embedded in an EC model corresponds to a cointegrating relationship
(Engle and Granger 1987; Hassler and Wolters 2006). Testing for cointegration in such
a setup therefore equals testing for the existence of a long-run relationship. However,
the latter concept retains its relevance when some of or all the variables are stationary.
Pesaran and Shin (1998) and Hassler and Wolters (2006) highlight some advan-
tages of the ARDL approach over alternative strategies for cointegration analysis—such
as the Engle and Granger (1987) two-step procedure implemented in the community-
1. Occasionally, the abbreviation ADL is used in the literature instead of ARDL.

S. Kripfganz and D. C. Schneider 985
contributed command egranger (Schaffer 2010) or the Phillips and Hansen (1990) fully
modified ordinary least-squares approach implemented in cointreg (Wang 2012). First
of all, it can accommodate a mixture of stationary and nonstationary variables without
the need for pretesting the order of integration. Moreover, the short-run and long-run
coefficients can be consistently estimated in one step, and the estimator’s asymptotic
normality eases statistical inference.2
Compared with a system-based Johansen (1995) cointegration analysis, which is im-
plemented in Stata’s vec command suite, the single-equation approach can be more
efficient if the focus is on one outcome variable, in addition to the aforementioned flex-
ibility regarding the integration orders. However, in the ARDL framework, the outcome
variable is not allowed to simultaneously determine the long-run equilibrium of other
explanatory variables, which would cause an endogeneity problem. The VAR or VEC ap-
proach can be more suitable for impulse–response analysis or dynamic forecasts because
feedback from the dependent variable to the weakly exogenous variables is explicitly
modeled.
Despite its advantages, testing for the existence of a long-run (cointegrating) rela-
tionship with the ARDL framework still requires a bit of effort. The test statistic has
a nonstandard distribution that depends on various characteristics of the model and
the data, including the integration order of the variables. Pesaran, Shin, and Smith
(2001) propose a “bounds test”, which involves comparing the values of conventional F
and t statistics with pairs of critical values (CV). Outside these bounds, the test either
conclusively rejects or does not reject the null hypothesis. Within the bounds, the test
is inconclusive.
This bounds test is implemented as a postestimation feature in our ardl pack-
age for the estimation of single-equation ARDL and EC models. Improved CV bounds
and approximate p-values have been obtained by Kripfganz and Schneider (2020) with
response-surface regressions using billions of simulated test statistics. These CVs are
more precise and exhaustive than earlier ones tabulated by Pesaran, Shin, and Smith
(2001) and Narayan (2005). A key feature of ardl is the automatic selection of the
optimal lag order with the Akaike information criterion (AIC) or Bayesian (Schwarz)
information criterion (BIC). With an increasing number of independent variables, the
number of candidate models—which are characterized by all possible combinations of
lag orders—is quickly in the tens or even hundreds of thousands. A computationally
efficient implementation of this procedure ensures that the optimal model is still found
within seconds.
Closely related, Jordan and Philips (2018) recently introduced the dynardl com-
mand for dynamic simulations of ARDL models. Their pssbounds command also pro-
vides an interface to display the original Pesaran, Shin, and Smith (2001) and Narayan
(2005) asymptotic and finite-sample CVs for the bounds test. As argued above, those
2. Shin, Yu, and Greenwood-Nimmo (2014) extend the ARDL framework by introducing nonlinearities
that allow for asymmetric long-run effects. Such a nonlinear ARDL model can be fit in Stata using
the command nardl, implemented by M. Sunder (http://www.marco-sunder.de/stata/). Here we
restrict our attention to the symmetric case.
CVs are now largely superseded. Moreover, their commands do not perform an auto-
matic lag-order selection, which is a key feature of our ardl command. Once the optimal
model specification is obtained with the ardl command, the dynardl command can still
be a useful complement if a visualization of the dynamic effects is desired.
This article is concerned only with time-series data. For the estimation of ARDL mod-
els in a large-T panel-data context, see the community-contributed commands xtpmg
(Blackburne and Frank 2007), xtdcce2 (Ditzen 2018, 2021), and xtivdfreg (Kripf-
ganz and Sarafidis 2021). The command xtwest (Persyn and Westerlund 2008) enables
cointegration tests based on panel-data EC models.
In section 2, we outline the econometric background for the ARDL approach to the
analysis of long-run equilibrium relationships, and we provide guidance for the model
specification and bounds-testing procedure. In sections 3 and 4, we describe the syntax
and options for the ardl package. In section 5, we illustrate the approach with an
empirical example from the realm of cryptocurrencies. Section 5 concludes.
2 Econometric model and methods

2.1 ARDL model
Suppose we expect the existence of an equilibrium relationship between an outcome
variable yt and a set of K explanatory variables xt = (x1t , x2t , . . . , xKt )0 :
yt = b0 + b1 t + x0t θ + et (1)
b0 is the intercept of the regression line, and b1 is the slope coefficient of a linear time
trend. The data are observed at consecutive time points t = 1, 2, . . . , T . Estimating
the regression coefficients in such a static model by ordinary least squares (OLS) might
result in spuriously large coefficient estimates even if there is no underlying relationship
among the variables. This is known to happen when the error term et is nonstationary
because of the nonstationarity of yt and xt (after accounting for the possibility of a
deterministic time trend).
Equation (1) remains a valid regression model if yt and some of or all the variables
xt are cointegrated, that is, when yt and xt are individually integrated of order 1,
I(1), but there exists a linear combination among them such that et is integrated of
order 0, I(0). Equation (1) reflects a conditional long-run equilibrium relationship—if it
exists—to which a process reverts over time. In the short run, the process might divert
from this equilibrium, but the above equation is silent about the dynamic evolution
of the process when it is off the equilibrium path. Such deviations are transitory, and
the elements in the data-generating process (DGP) governing them are therefore I(0).
These neglected I(0) components in the DGP affect the finite-sample (and possibly the
asymptotic) distributions of test statistics and thus invalidate conventional hypothesis
tests and regression diagnostics.3
To circumvent the problems associated with fitting a static model, we can augment
the regression equation with lags of the dependent and independent variables. We can
even include another set of L exogenous variables zt , which may have predictive power
to explain the short-term fluctuations of yt but do not affect its equilibrium path. We
assume that all variables in zt are (trend) stationary. Augmenting the model in this
way aims at obtaining a dynamically complete model in which the regression error term
ut is free of serial correlation:
p q
(2)
X X
yt = c0 + c1 t + φi yt−i + βi0 xt−i + γ 0 zt + ut
i=1 i=0
t = 1 + p∗ , . . . , T . Leaving aside the variables zt , this is a general ARDL (p, q, . . . , q)

model with intercept c0 , linear trend c1 t, and lag orders p ∈ [1, p∗ ] and q ∈ [0, p∗ ].4 To
ensure that there are enough degrees of freedom available to fit the model’s coefficients
with sufficient precision, we may need to choose the maximum admissible lag order
p∗ conservatively. This is especially relevant when the number of observations in the
dataset (T ) is relatively small, the number of variables in xt (K) is relatively large, or
both.5
Given the initial observations y1 , y2 , . . . , yp∗ and the time paths of xt and zt , (2)
describes the dynamic evolution of yt over time, irrespective of whether an equilibrium
relationship—as postulated in (1)—exists. The intercept c0 and the linear time trend c1 t
may or may not be included in the model, depending on the nature of the variables under
consideration.6 We assume that enough lags have been included in the ARDL model (2)
to purge the error term from any remaining serial correlation and to ensure that the
3. We provide an extended introduction to these topics in the working-paper version of this article
(Kripfganz and Schneider 2022). For further background reading, interested readers are referred to
any textbook on time-series econometrics of their choice.
4. Allowing for different lag orders among the components of xt is straightforward and can be treated
as a special case of the general model by restricting some coefficients to be zero.
5. The importance of the maximum lag order p∗ is explained further below. In practice, the data
frequency often guides this choice. For instance, it is customary to allow for up to 12 lags with
monthly data and up to 4 or 8 lags with quarterly data.
6. In general, other deterministic model components—such as quadratic time trends or impulse dummy
variables—can be added. We abstract from them here but note that their inclusion may affect the
applicability of the CVs for the bounds test, which is presented further down in this article.
variables xt are weakly exogenous/long-run forcing—ruling out any contemporaneous

feedback from yt to xt . If there exists a stable long-run relationship, conventional
asymptotic theory can be applied for statistical inference on any of the coefficients even
if some of the variables are nonstationary (Pesaran and Shin 1998). This highlights
the importance of testing for the existence of such a long-run relationship, which we
consider in section 2.3.
While the inclusion of further lags improves the regression fit, this comes at the cost
of a higher variance of the coefficient estimates. To balance this tradeoff, we can base a
data-driven approach to optimal lag selection on the AIC or the BIC,
AIC = −2 ln(L) + 2K ∗
BIC = −2 ln(L) + ln(T ∗ )K ∗
where ln(L) is the value of the log-likelihood function from the fitted regression model,
T ∗ = T − p∗ is the effective sample size, and K ∗ = 2 + p + K(q + 1) + L is the number
of estimated coefficients in (2). These criteria balance the desire for a better fit of the
model—higher values of ln(L)—against the temptation of creating ever larger models.
The BIC has a larger penalty term than the AIC (for T ∗ ≥ 8) and therefore tends to select
more parsimonious models. The optimal lag orders are then found by fitting model (2)
for all possible combinations of p and q and choosing the model that minimizes the AIC
or BIC.
For the comparability of the model-selection criteria, we must base all regressions on
the same estimation sample. This is the reason for initially choosing a fixed maximum
lag order p∗ . When both p and q are smaller than p∗ , the estimation of model (2) does
not use all the available observations. This is the price we need to pay for consulting
the model-selection criteria. Once the optimal lag orders p and q have been found, we
can subsequently refit the model, utilizing all available observations by setting p∗ =
max(p, q).
2.2 Error-correction representation

To gain a better interpretability of the model’s coefficients, we can reformulate the ARDL
model in EC representation (Hassler and Wolters 2006):7
p−1 q−1
∆xt−i +γ 0 zt +ut (3)
X X
∆yt = c0 +c1 t−α(yt−1 −θxt−1 )+ ψyi ∆yt−i +ω 0 ∆xt + 0
ψxi
i=1 i=1
7. By convention, the summations evaluate to zero if the upper limit is smaller than the lower limit.
The coefficients in (3) can be mapped in a straightforward algebraic way to the coeffi-
cients in (2):
p Pq
X j=0 βj
α=1− φi , θ=
i=1
α
p
X q
X
ψyi = − φj , ω = β0 , ψxi = − βj
j=i+1 j=i+1
Now recall the hypothesized long-run equilibrium relationship between yt and xt in

(1). Ignoring the intercept and linear time trend for the moment, we see the deviations
from this equilibrium, et−1 = yt−1 − θxt−1 , can be found again in the EC model (3).
Because of the nonlinear interaction between the coefficients α and θ, we cannot directly
fit (3) with OLS. However, given the mapping above, we can recover consistent estimates
of all coefficients from the ARDL model (2). Yet a computationally more convenient
approach is to instead fit the following model:8
p−1 q−1
∆xt−i +γ 0 zt +ut (4)
X X
0 0
∆yt = c0 +c1 t+πy yt−1 +πx xt−1 + ψyi ∆yt−i +ω ∆xt + ψxi
i=1 i=1
From the above model, we can easily recover the so-called speed-of-adjustment coeffi-
cient α = −πy and the long-run coefficients θ = πx /α. The corresponding standard
errors can be computed with the delta method (Pesaran and Shin 1998). Note that (4)
collapses to the well-known augmented Dickey and Fuller (1979) regression for unit-root
testing when no explanatory variables xt and zt are present (K = L = 0).
The speed-of-adjustment coefficient α tells us how fast the process for yt reverts to
its long-run relationship when this equilibrium is distorted. α = 1 would imply that—
in the absence of any other short-run fluctuations—any deviation from the equilibrium
is fully corrected immediately in the period after the distortion occurs. In contrast,
α = 0 would imply that the process never returns to its equilibrium path. Values of
α between these two boundaries reflect a partial-adjustment process, where the gap to
the equilibrium is gradually closed over time.9
Clearly, θ 6= 0 is not a sufficient condition for the existence of a conditional long-run
relationship between the levels of yt and xt . When α = 0, then yt is I(1) and no such
relationship exists. In the opposite scenario, when θ = 0 and α ∈ (0, 2), then yt is
(trend) stationary, irrespective of the integration order of the components in xt . For a
long-run level relationship to exist, we need both θ 6= 0 and α > 0. In this case—as long
as the elements of xt are not cointegrated among themselves—the integration properties
8. When called with the option ec1, the ardl command estimates (4) but reports the coefficients for
(3).
9. While we allow α to fall into the interval [0, 2) in the following, we do not pay particular attention
to the oscillating or overshooting case α > 1 in this article. We also rule out explosive processes,
which result under α < 0. An estimate of α outside the reasonable region [0, 1] should be seen as a
warning signal for potential model misspecification.
of xt determine the integration order of yt . If the variables in xt with nonzero long-run

coefficient are I(1), then yt is I(1) as well, and the conditional long-run relationship
corresponds to a cointegrating relationship.
In this context, note that the assumption of xt being long-run forcing for yt implies
that there can exist at most one cointegrating relationship that involves yt . Consider
the VEC model
∗
0
pX −1
∆yt πyy πyx yt−1 ∆yt−i ε
= a0 + a1 t + + Ψi + Γzt + y,t (5)
∆xt πxy Πxx xt−1 ∆xt−i εx,t
i=1
xt is long-run forcing for yt if it obeys the restriction πxy = 0; that is, there is no
level effect of yt−1 on ∆xt .10 This does not rule out further cointegrating relationships
among the elements of xt , Πxx 6= 0. Thus, without further inspection, a cointegration
rank larger than one for the entire system (yt , x0t )0 does not necessarily imply a viola-
tion of this assumption. However, if there is reason to suspect multiple cointegrating
relationships involving yt , πxy 6= 0, then a single-equation ARDL or EC model is inap-
propriate.11 Instead, this would call for a multivariate cointegration analysis within the
Johansen (1995) framework by fitting a VAR or VEC model.12 In contrast, if πxy = 0
is indeed satisfied and the interest is primarily on the long-run relationship between
yt and xt , then fitting a single-equation model is more efficient and computationally
straightforward.
The remaining coefficients ψyi , ω, ψxi , and γ in (3) capture the short-run dynamics
that are not prescribed by the equilibrium-reverting forces.13 They not only are relevant
for making dynamic forecasts but also play a role for choosing appropriate CVs when
testing for the existence of a long-run relationship, which we explore in section 2.3.
A complication arises if q = 0 for some of or all the long-run forcing variables.
In that situation, πx = ω, which implies that the corresponding variance–covariance
matrix of the coefficient estimates in (4) is rank deficient. To avoid this complication, we
can equivalently formulate the EC representation with the levels of the long-run forcing
variables expressed in period t instead of t − 1:
p−1 q−1
(6)
X X
0
∆yt = c0 + c1 t − α(yt−1 − θxt ) + ψyi ∆yt−i + ψxi ∆xt−i + γ 0 zt + ut
i=1 i=0
10. See Pesaran, Shin, and Smith (2001) for a detailed discussion of the relationship between the
single-equation EC model (3) and the VEC model (5).
11. Consequently, it is not permissible to run several ARDL regressions involving the variables (yt , x0t )0 ,
in which the dependent variable of one regression becomes an independent variable in other regres-
sions.
12. See var, vec, and related Stata commands.
13. Strictly speaking, the error correction governed by the coefficient α is a short-run adjustment as
well.
It has the same parameter restrictions as defined above. Note that ω 0 ∆xt is replaced
by ψx0
0
∆xt . The interpretation of the long-run coefficients θ does not change because
the time subscript does not matter when the process is in equilibrium. The equation to
be estimated in this case becomes
p−1 q−1
(7)
X X
0
∆yt = c0 + c1 t + πy yt−1 + πx xt + ψyi ∆yt−i + ψxi ∆xt−i + γ 0 zt + ut
i=1 i=0
where the coefficients πx are identical to the corresponding coefficients in (4), despite
the change in the time subscript.14
2.3 Bounds test

Although we can consistently estimate all coefficients in the ARDL model (2) or its EC
representations, testing for the existence of a long-run relationship involves a bit more
effort. This is because the process for yt contains a unit root under the null hypothesis
of no long-run relationship; therefore, the test statistics have nonstandard distribu-
tions. Moreover, the tests depend on the choice of deterministic model components. In
the ARDL model (2)—and its EC representations (3) and (6)—we have allowed for an
intercept c0 and a linear time trend c1 t. We can distinguish the following five cases:
1. No deterministic model components are included (c0 = c1 = 0).

2. A restricted intercept is included (c0 = αb0 ) but no time trend (c1 = 0).
3. An unrestricted intercept is included (c0 6= 0) but no time trend (c1 = 0).
4. An unrestricted intercept is included (c0 6= 0) and a restricted time trend
(c1 = αb1 ).
5. Both deterministic model components are unrestricted (c0 6= 0 and c1 6= 0).
A decision about the relevant case can often be guided by a visual inspection of the
time series. Cases 1 and 2 are in line with a process yt , which could reasonably be an
I(1) process without drift under the null hypothesis of no long-run level relationship.
Under the alternative hypothesis, yt would either be I(0) or cointegrated with xt . Case 1
is most appropriate if yt and xt fluctuate around a zero mean or if any nonzero means
cancel out in the long-run level relationship; that is, b0 = b1 = 0 in (1). The latter
condition is hard to verify ex ante, such that case 2 is often the safer option whenever
some variables have a nonzero mean.
14. When called with the option ec, the ardl command estimates (7) but reports the coefficients for (6).
As an aside, when option ec1 is specified—which normally commands time subscripts t − 1 for the
long-run forcing variables—a subscript t is used in the estimation equation for those variables whose
lag order is q = 0. However, the reported results are still reparameterized as in (3), incorporating
the constraint on ω.
If yt appears to be trending, it could be an I(1) process with drift under the null
hypothesis. This calls for case 3 or 4. Under the alternative hypothesis, yt would either
be trend stationary or cointegrated with xt . Case 3 is most appropriate if the trend
in yt is entirely attributable to a trend in xt ; that is, b1 = 0 in (1). Again, this may
be difficult to justify ex ante. Despite the fact that case 3 is most commonly applied
in the empirical practice, case 4 is generally the safer option when there is insufficient
knowledge about the source of the observed time trend.
Especially when the sample size is relatively small, it might be difficult to distinguish
visually between a mildly drifting unit-root process under the null hypothesis and a
stationary process which is fluctuating around a constant mean under the alternative
hypothesis. This can be another relevant situation for case 3. Similarly, case 5 could
be used to statistically discriminate between a unit-root process with faster—although
hardly noticeable—than linear growth (or decline) and a trend-stationary process. For
most practical applications, this might be a rather irrelevant scenario.
Note that the restrictions on the intercept or linear trend under cases 2 and 4 do not
affect the estimation of the ARDL model because it is irrelevant whether we treat c0 (c1 )
or b0 (b1 ) as a free parameter to be estimated. Under case 1, (2) is estimated without
intercept and trend. Under cases 2 and 3, an intercept is included in the regression.
Under cases 4 and 5, an intercept and linear time trend are included. However, the
restrictions are incorporated into step 1 of the bounds testing procedure, which we
describe in the following:
1. First, we test the joint null hypothesis
case 1, 3, or 5

(πy = 0) ∩ (πx = 0),

H0 : (πy = 0) ∩ (πx = 0) ∩ c0 = 0, case 2
case 4

(πy = 0) ∩ (πx = 0) ∩ c1 = 0,

versus the alternative hypothesis
case 1, 3, or 5

(πy 6= 0) ∪ (πx =
 6 0),
H1 : (πy 6= 0) ∪ (πx = 6 0) ∪ c0 =6 0, case 2
case 4

(πy 6= 0) ∪ (πx = 6 0) ∪ c1 =6 0,

The hypotheses are not directly formulated in terms of the long-run coefficients
θ, because they are not well defined when πy = 0. Instead, the test is formulated
as a test for valid exclusion of the level terms yt−1 and xt−1 (or xt ) in (4) or (7).
The test statistic is a conventional F statistic for joint validity of the K + 1 (or
K + 2) restrictions imposed under the null hypothesis. However, the nonstandard
distribution requires the use of different CVs, which we discuss further below. If
the null hypothesis is not rejected, we conclude that there is no statistical evidence
in favor of a long-run level relationship between yt and xt . Otherwise, we should
proceed with the following steps because of the possibility of degenerate cases,
which are not ruled out by the alternative hypothesis of this first step.
2. If the null hypothesis from step 1 is rejected, we need to rule out the special case
that yt is I(1) but not cointegrated with any variable in xt . This is done by testing
H0 : π y = 0 versus H1 : πy < 0
The test statistic is a conventional t statistic for statistical insignificance of the

negative speed-of-adjustment estimate with a one-sided rejection region. As in
step 1, the distribution is nonstandard, and the usual CVs do not apply. If the null
hypothesis is not rejected, we conclude again that there is no statistical evidence
of a long-run level relationship. Otherwise, we proceed with step 3.
3. If the null hypotheses in steps 1 and 2 are both rejected, we eventually consider the
degenerate case that yt is (trend) stationary but not part of a long-run relationship
with xt . For this purpose, we can use conventional Wald tests for the joint (or
individual) statistical insignificance of the long-run coefficients:
H0 : θ = 0 versus H1 : θ 6= 0
We base this test on the long-run coefficients θ rather than πx because the OLS
estimator of θ is asymptotically normally distributed (Pesaran and Shin 1998),
irrespective of the integration orders of xt , assuming that α > 0 as indicated by
the test result from step 2. Thus, conventional CVs can be used.
The rejection of the null hypotheses from all three steps is necessary to conclude that
there is statistical evidence in favor of a long-run relationship; that is, (α > 0)∩(θ 6= 0).
It is clear that the alternative hypothesis in step 1 does not rule out the two degenerate
cases, which are the subject of steps 2 and 3. Yet we should still start with step 1
because it is carried out under less restrictive assumptions on the DGP than step 2.15
For the test statistics in steps 1 and 2, Pesaran, Shin, and Smith (2001) derive the
asymptotic distributions under two scenarios. In the first scenario, all long-run forcing
variables xt are individually I(0). In the second scenario, all of them are I(1) and not
mutually cointegrated. When the (co)integration properties of xt are unknown, the
corresponding CVs form lower and upper bounds. Conclusive evidence is possible when
the value of the test statistic falls outside these bounds. The region for not rejecting
the null hypothesis is below the lower bound (closer to zero), and the rejection region is
above the upper bound. The test is inconclusive if the test statistic falls between the two
bounds. Because the distributions have nonstandard forms, CVs have to be obtained
by simulations. This is complicated by the fact that the distributions depend on the
number of variables in xt . For K ≤ 10, Pesaran, Shin, and Smith (2001) tabulated
near-asymptotic CVs for the F statistic in step 1 and the t statistic in step 2. However,
the asymptotic distributions might be poor approximations when the sample size is
relatively small.16
15. For technical details and a full set of assumptions, see Pesaran, Shin, and Smith (2001).
16. Because the restrictions on the deterministic components do not alter the underlying DGP, the
CVs for the single-hypothesis test in step 2 are the same for cases 2 and 3 and similarly for cases 4
and 5; see again Pesaran, Shin, and Smith (2001) for further discussion.
Note that the distributions and CVs are obtained under the assumption of indepen-
dent and identically normally distributed errors ut . As mentioned earlier, a standard
procedure for dealing with suspected serial correlation is to increase the lag orders p, q,
or both in the ARDL model. While the p + Kq short-run terms in the EC representation
do not affect the asymptotic distributions of the test statistics, they are relevant for the
finite-sample distributions. Consequently, different CVs are needed for each combination
of T ∗ , K, and p + Kq, separately for the lower and upper bounds. Instead of tabulating
vast amounts of CVs, Kripfganz and Schneider (2020) estimated response-surface regres-
sions, which can predict CVs for any desired sample size, number of long-run forcing
variables, and lag order. This includes asymptotic CVs. Another important advantage of
this approach is the ability to compute approximate p-values, which facilitate statistical
inference.
2.4 Practical guidelines

The following stages characterize a stylized ARDL approach to testing for the existence
of a conditional long-run level relationship:
1. Decide about the candidate variables xt that are assumed to be long-run forcing
for yt . These variables can be either I(0) or I(1). No pretesting is necessary
unless we suspect that a variable might be I(2). Stationary variables zt that are
suspected to affect the short-run dynamics—but not the long-run equilibrium—
can be added to the ARDL model as well. If there is doubt about the (trend)
stationarity of zt , unit-root tests can be carried out.
2. Decide about the deterministic model components to be included in the model
and whether the constant or linear trend coefficient should be restricted; that is,
choose one of the five cases above. If in doubt, choose a more flexible model.17
3. Choose a maximum lag order p∗ , ensuring that sufficiently many degrees of free-
dom are available.18 Keeping the estimation sample fixed, use the AIC or BIC to
obtain the optimal lag orders p and q. To assert that the model is dynamically
complete, a serial-correlation test could be of assistance. If there is concern about
remaining serial correlation, the AIC might be preferred over the BIC because it
tends to select less parsimonious models. Additional specification tests—for ex-
ample, tests for heteroskedasticity and normality of the errors—could be used to
check whether the assumptions underlying the bounds test are met.
4. Check the plausibility of the coefficient estimates in the EC representation. For
example, an implausible estimate of α, which is clearly outside of the interval [0, 2),
might give rise to concern about the correct model specification or a potential
overparameterization of the model.
17. The higher the case number, the less restrictive the model specification.
18. The Kripfganz and Schneider (2020) CVs are available only if there are at least twice as many
observations T ∗ than coefficients K ∗ . For reliable inference, a much higher ratio is usually recom-
mended.
5. Follow the three steps of the bounds test procedure. For steps 1 and 2, do not
reject the null hypothesis if the value of the test statistic is below—that is, closer
to zero—the lower bound of the Kripfganz and Schneider (2020) CVs. Reject the
null hypothesis (and proceed with the next testing step) if the test statistic exceeds
the upper-bound CV.
6. If there is conclusive statistical evidence in favor of a long-run relationship, con-
sider refitting a more parsimonious model with lag orders selected by the BIC. If
there is evidence against a long-run level relationship, consider refitting an ARDL
model in first differences to obtain more efficient estimates,
p−1
X q−1
X
0
∆yt = c0 + c1 t + ψyi ∆yt−i + ψxi ∆xt−i + γ 0 zt + ut
i=1 i=0
which is a restricted version of (7) with πy = 0 and πx = 0. In both cases, it

might be worth removing variables that do not help to improve the model fit.
This reestimation stage can be skipped if there is no interest in further statistical
analysis—for example, forecasting—beyond the exploration of a levels relation-
ship.
To avoid pretesting problems, keep model simplifications—like those at stage 6—to

a minimum before the bounds test is performed. Also note that there is no need to
separately fit a static model in levels if the bounds test provides evidence in favor of a
long-run relationship. As discussed earlier, the respective long-run coefficients can be
inferred directly from the EC representation (3) or (6).
3 The ardl command

3.1 Syntax
ardl depvar indepvars if in

, lags(numlist) exog(exogvars) ec ec1

noconstant trendvar (trendvarname) restricted regstore(storename)

perfect maxlags(numlist) aic bic maxcombs(combnum)

matcrit(lagcombmat) nofast dots noctable noheader display_options

3.2 Options
lags(numlist) specifies the number of lags for some or all regressors. The first number
specifies the lag length p for depvar (yt ), which must be larger than zero. The
following numbers specify the lag lengths q for the independent variables in the
order they appear in indepvars (xt ), which can be zero or higher. Missing values are
allowed; they indicate that the respective lag order is not prespecified but instead
determined with information criteria. If numlist contains only one element, the same
lag order is applied to all variables. Otherwise, the number of elements in numlist
must equal the number of variables in depvar and indepvars.
exog(exogvars) specifies additional variables (zt ) to be added to regression. An auto-
matic lag order selection is not performed for these variables.
ec displays the results in error-correction form. indepvars enter the long-run relationship
with time subscript t, as in (6).
ec1 displays the results in error-correction form. indepvars enter the long-run relation-
ship with time subscript t − 1, as in (3).
noconstant suppresses the constant term. Specifying this option implies that the
bounds test uses CVs for case 1.
trendvar (trendvarname) specifies a linear time trend to be added to the regression.

trendvarname must be a variable that is collinear with timevar, the variable that is
used with tsset to declare the data to be time-series data. Specifying trendvar
is equivalent to trendvar(timevar). Specifying this option implies that the bounds
test uses CVs for case 4 or 5.
restricted specifies that the constant term or the time trend, if specified, will be
restricted for the purpose of the bounds test. The restricted deterministic component
will be displayed in the long-run section of the error-correction output. Specifying
this option implies that the bounds test uses CVs for case 2 or 4.
regstore(storename) stores the estimation results from the underlying regress com-
mand. These are the OLS estimates of (4) or (7) when option ec1 or ec0 is specified,
respectively, and (2) otherwise.
perfect omits the collinearity check among the regressors.
maxlags(numlist) specifies the maximum lag order p∗ for the optimal lag order selection.
The first number specifies the maximum lag length for depvar (yt ), which must be
larger than zero. The following numbers specify the maximum lag lengths for the
independent variables in the order they appear in indepvars (xt ), which can be zero
or higher. Missing values are allowed; they indicate that the default maximum lag
order 4 is to be used. If numlist contains only one element, the same maximum lag
order is applied to all variables. Otherwise, the number of elements in numlist must
equal the number of variables in depvar and indepvars.
aic requests that the optimal lag lengths be determined with the AIC.
bic, the default, requests that the optimal lag lengths be determined with the BIC.
maxcombs(combnum) restricts the maximum number of lag permutations for the auto-
matic lag selection. The default is maxcombs(100000), or maxcombs(500) if option
nofast is specified. Higher values are possible.19
19. The purpose of this option is to prevent the optimal lag order selection from taking a lot of time
without explicit user consent.
matcrit(lagcombmat) saves the lag permutations and the respective information crite-
rion in a matrix named lagcombmat.
nofast uses the regress command instead of dedicated Mata code to run the auxiliary
regressions for the optimal lag order selection. This is much slower but might be
numerically more robust in rare cases.
dots displays a progress bar for the optimal lag order selection. This is useful when
there are many permutations because of many variables and high maximum lag
orders. Each dot represents a 1% progress in the evaluation of candidate models.
noctable suppresses the display of the coefficient table.
noheader suppresses the display of the coefficient table header.
display_options: noomitted, vsquish, cformat(%fmt), pformat(%fmt),
sformat(%fmt), and (Stata 12+ only) nolstretch; see [R] Estimation options.
3.3 Stored results

ardl stores the following results in e():20
Scalars
e(N) number of observations e(ll_0) log likelihood, constant only
e(df_m) model degrees of freedom e(rank) rank of e(V)
e(df_r) residual degrees of freedom e(F) F statistic ∗
e(mss) model sum of squares e(case) case number for deterministic
e(rss) residual sum of squares components ∗
e(rmse) root mean squared error e(F_pss) bounds-test F statistic ∗
e(r2) R2 e(t_pss) bounds-test t statistic ∗
e(r2_a) adjusted R2 e(numcombs) number of lag combinations ∗
e(ll) log likelihood under assumption e(N_gaps) number of gaps in sample
of independent and e(tmin) first time period in sample
identically distributed e(tmax) last time period in sample
normal errors
Macros
e(cmd) ardl e(tmaxs) formatted maximum time
e(cmdline) command as typed e(regressors) full set of regressors
e(cmdversion) version of the ardl command e(det) deterministic components ∗
e(depvar) name of dependent variable e(exogvars) exogenous variables ∗
e(title) title in estimation output e(srvars) short-run regressors ∗
e(estat_cmd) ardl_estat e(lrdet) long-run deterministic
e(predict) ardl_p component ∗
e(tsfmt) format for the time variable e(lrxvars) long-run regressors ∗
e(tvar) time variable e(properties) b V
e(tmins) formatted minimum time e(model) level or ec
Matrices
e(b) coefficient vector e(maxlags) maximum lag lengths
e(V) variance–covariance matrix e(lags) lag lengths in ARDL model
Functions
e(sample) marks estimation sample
20. Starred results (∗ ) are not always stored.

4 Postestimation commands
Many standard postestimation commands for the regress command can be used after
the ardl command. Importantly, the results obtained with some of them can differ
depending on whether the model is specified in the ARDL level form (2) or one of the EC
forms (3) or (6). For example, the estat ovtest includes higher-order powers of the
dependent variable—which is either yt or ∆yt —as regressors in an auxiliary regression.
This complication does not apply to postestimation commands based on residuals—such
as estat bgodfrey and estat imtest—because the error term ut is unaffected by the
model’s reparameterization.
The Pesaran, Shin, and Smith (2001) bounds test for the existence of a long-run level
relationship with Kripfganz and Schneider (2020) CVs and approximate p-values—as
discussed in section 2.3—is implemented in the postestimation command estat ectest.
It requires the option ec or ec1 to be specified with the ardl command.
4.1 Syntax
estat ectest , siglevels(numlist) asymptotic nocritval norule

nodecision

4.2 Options
siglevels(numlist) shows CVs for levels in the numlist, which must have at least one
element. The default is siglevels(10 5 1). Levels are specified as percentiles
but do allow for two digits after the decimal point. There are 221 different lev-
els among which you can choose, indicated by the Stata numlist 0.01 0.02 0.05
0.10(0.10)0.90 1.00(0.50)98.50 99.00(0.10)99.90 99.95 99.98 99.99.
asymptotic requests that the sample size returned by ardl in e(N) be ignored and
show asymptotic CVs instead.
nocritval suppresses display of the CVs table.
norule suppresses display of the decision rule.
nodecision suppresses display of the decision table.
5 Example
We illustrate the ardl command with an example on cryptocurrencies.21 Specifically,
we investigate whether supply and demand factors have a long-run impact on the price
21. As another illustrative example, we replicate the empirical analysis of Pesaran, Shin, and Smith
(2001), who estimate an earnings equation with macroeconomic data for the United Kingdom. This
exercise can be found in the working paper version of this article (Kripfganz and Schneider 2022).
of Bitcoin (variable bprice, in U.S. dollars [USD]). Few would debate that Bitcoin and
many other cryptocurrencies are highly speculative financial assets in the short run.
Taking a long-run perspective, an often-invoked explanation for the steep rise to almost
USD 69,000 per Bitcoin in November 2021 is that ever-increasing demand meets limited
supply. This cannot explain, however, that the spectacular rise of the Bitcoin price was
followed recently by a stark drop to almost USD 15,000 in late 2022. This raises the
question whether supply and demand forces can indeed tame the Bitcoin price in the
long run or whether its speculative (and therefore unpredictable) nature is prevailing.
In econometric terms, we aim to investigate whether we can find evidence for a long-run
equilibrium relationship between the Bitcoin price and its supply and demand factors.
The motivation for the key variables in our dataset follows Ciaian, Rajcaniova, and
Kancs (2016). If a long-run equilibrium exists, the Bitcoin price is expected to be
inversely proportional to its supply, which can be approximated by the historical number
of mined Bitcoins (variable supply). On the demand side, the equilibrium price can
be expected to grow proportionally to the size of the Bitcoin economy, as measured by
the number of daily Bitcoin transactions (ntrans). It would also be inversely related
to its velocity. Here the (inverse) velocity is proxied by so-called coin days destroyed
(ddestr). Broadly speaking, this is an aggregate measure of how much time has elapsed
between two transactions with the same coin. More transactions with coins that have
been dormant for a longer time indicate an increase in economic activity.
Because Bitcoin is predominantly priced in USD, a depreciation of the dollar makes it
cheaper to carry out Bitcoin transactions for investors in the rest of the world, therefore
increasing demand. For simplicity, following Ciaian, Rajcaniova, and Kancs (2016), we
just include the USD/EUR exchange rate (fxeu_f) in our set of explanatory variables.22
The linearization of the equilibrium relationship requires a log transformation for all the
variables, indicated by the prefix ln_ in the variable names used below.23 In our sample,
we have 3,255 daily observations from January 1, 2014, to November 29, 2022.24 We
do not use pre-2014 data, because Ciaian, Rajcaniova, and Kancs (2016) find evidence
of a structural break in 2013.25
We start with a visual inspection of the key variables.26 In the left panel of figure 1,
the evolution of the log Bitcoin price and its supply are shown. The latter largely
follows a deterministic path, which is prescribed by the underlying Bitcoin protocol.
The Bitcoin price shows all signs of a nonstationary variable. While it shares a similar
22. The _f in the fxeu_f variable name indicates that we imputed the missing weekend values by
carrying forward the Friday value.
23. This is an important deviation from Ciaian, Rajcaniova, and Kancs (2016), who refrain from this
transformation because of zero values for some of the variables in the early days of Bitcoin. Here
we focus on a later sample period, where Bitcoin was well established.
24. The data sources are coinmetrics.io for bprice, supply, and ntrans, https: // blockchair.com for
ddestr, and https: // fred.stlouisfed.org (Federal Reserve Bank of St. Louis) for fxeu (FRED code
DEXUSEU). The data were downloaded in December 2022 and January 2023.
25. Our estimation results are sensitive to the inclusion of earlier data points. It can be argued that
the pre-2014 period may not be representative for the subsequent dynamics, because Bitcoin was
still in a maturation stage.
26. The code for replicating figure 1 can be found in the ancillary file ardl_example.do.
upward trend with the supply, the quasideterministic nature of the mining process
precludes that the two series could be cointegrated. To avoid distorting the bounds
test, it is thus advisable to exclude the supply from the long-run relationship for testing
purposes. It can still enter the regression model as an exogenous price determinant—a
zt variable in terms of the notation from section 2.
11.5 16.9 20.0 0.4
10.5 16.8
18.0 0.3
9.5 16.7
16.0 0.2
8.5 16.6
7.5 16.5
14.0 0.1
6.5 16.4
12.0 0.0
5.5 16.3
4.5 16.2 10.0 -0.1

2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
left axis: ln_bprice left axis: ln_ntrans ln_ddestr

right axis: ln_supply right axis: ln_fxeu_f
Figure 1. Time-series graphs for the main regression variables
The right panel of figure 1 depicts the demand side factors. The USD/EUR exchange
rate is clearly nonstationary, while coin days destroyed look fairly stationary. The pic-
ture is less clear about the daily Bitcoin transactions series, which appears to follow
different time trends at different periods in our sample and therefore is likely nonsta-
tionary. We could verify these assessments with conventional unit-root tests, but this
is not necessary for ARDL estimation and bounds testing. It is one of the latter’s ad-
vantages that it can deal with mixtures of I(0) and I(1) variables. We will confirm our
initial assessment further below in the context of fitting a VEC model, where pretesting
for the order of integration is required.
There is no apparent reason to believe that the observed time trend in the Bitcoin
price is entirely attributable to the underlying time trend in the other variables. This
calls for the inclusion of a restricted time trend—case 4—in the EC model. Under the
null hypothesis of the bounds test, the log Bitcoin price would then follow a random
walk with drift. Alternatively, if α 6= 0, it can be either cointegrated or trend stationary.
Given that we have daily data, we choose a maximum lag order p∗ = 7, such that
the lags can cover up to one week. Thanks to the large number of observations, we do
not have to be conservative with the degrees of freedom and therefore select the optimal
lag combination with the AIC rather than the BIC. This reduces the risk of misspecifying
the model dynamics, which in turn might invalidate the bounds test. We therefore add
the maxlags(7) and aic options to our ardl command line. The quasideterministic
supply variable is specified with the exog() option, which also constrains its lag order to
zero. If we were to add it as a conventional independent variable instead, the command
would error out because of collinearity of the lags.27 To complete our specification, we
add a linear time trend by including the name of the time identifier in the trend()
option. While the optimization over all 3,584 lag combinations finishes in virtually no
time using our fast Mata algorithm, we illustrate how to display a progress bar with
the option dots, which might be useful for larger models:
. use ardl_example
(Bitcoin price related data, -ardl-, SJ (type -notes- for sources))
. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f, exog(ln_supply) trendvar(date)
> aic maxlags(7) dots
Optimal lag selection, % complete:
20% 40% 60% 80% 100%
..................................................
AIC optimized over 3584 lag combinations
ARDL(2,1,3,2) regression
Sample: 2014-01-08 thru 2022-11-29 Number of obs = 3,248
F(13, 3234) = 519948.59
Prob > F = 0.0000
R-squared = 0.9995
Adj R-squared = 0.9995
Log likelihood = 5950.3079 Root MSE = 0.0388
ln_bprice Coefficient Std. err. t P>|t| [95% conf. interval]
ln_bprice
L1. .954804 .0175476 54.41 0.000 .9203984 .9892096
L2. .038809 .0175277 2.21 0.027 .0044424 .0731756
ln_ntrans
--. .0234125 .0062507 3.75 0.000 .0111568 .0356682
L1. -.0097699 .0062652 -1.56 0.119 -.022054 .0025141
ln_ddestr
--. -.0014938 .0010689 -1.40 0.162 -.0035896 .000602
L1. -.0006291 .0010955 -0.57 0.566 -.0027769 .0015188
L2. .0006551 .0010921 0.60 0.549 -.0014863 .0027964
L3. .0027045 .0010154 2.66 0.008 .0007137 .0046953
ln_fxeu_f
--. .4878456 .1658287 2.94 0.003 .1627056 .8129856
L1. -.7029864 .2349611 -2.99 0.003 -1.163674 -.2422987
L2. .2828991 .1662484 1.70 0.089 -.0430638 .608862
ln_supply -.0283604 .0420257 -0.67 0.500 -.1107601 .0540393

date .0000114 4.86e-06 2.34 0.020 1.82e-06 .0000209
_cons .0863434 .5877493 0.15 0.883 -1.066055 1.238742
27. Alternatively, we could specify the supply variable as an independent variable and manually restrict
its lag order to zero with the lags() option. While this yields identical ARDL coefficient estimates,
it would later on include the supply variable in the long-run relationship when we reparameterize
the model in EC form. As argued above, this is problematic with regard to the bounds test.
The optimal model chosen by the AIC is an ARDL(2,1,3,2) model.28 The supply of
Bitcoin has no statistically significant effect, which could justify removing this regressor
at a later stage. In contrast, the linear time trend, represented by the variable date, is
statistically significant at the 5% level, in line with our earlier observations.29 Before
turning our attention to the bounds test, we should inspect the residuals for potential
serial correlation:
. estat bgodfrey, lags(1/3)
Breusch--Godfrey LM test for autocorrelation
lags(p) chi2 df Prob > chi2
1 0.161 1 0.6880
2 0.161 2 0.9224
3 0.768 3 0.8571
H0: no serial correlation
The Lagrange multiplier test does not provide reason for concern about residual
serial correlation. We now refit the model in error-correction form—(6)—using the ec
option. While the AIC would give us the same lag orders again, we can also directly
specify the optimal lag orders with the lags() option. However, we then need to exert
some caution to obtain results for the same estimation sample as above. Allowing for
a maximum of seven lags, we set aside the first seven data points for the optimal lag
determination. The estimation sample was held fixed by the ardl command even for
models with lower lag orders. To base the bounds test again on the same estimation
sample, we restrict it in the next step with the e(sample) function. To obtain the correct
CVs with the bounds test for case 4, we now also need to add the option restricted,
which includes the time trend in the long-run relationship:
28. The BIC would select an overly parsimonious ARDL(1,0,0,0) model. However, there would be
evidence of remaining serial correlation in the residuals, potentially invalidating the bounds test.
29. If we were unsure about including a time trend, we could also refit the model without it and
compare the specifications again with the AIC. Here the AIC is lower—and therefore preferable—
for the model with trend. Note that the Bitcoin supply remains statistically insignificant (at the
5% level) even in the model without linear time trend.
. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f if e(sample), exog(ln_supply)

> trendvar(date) lags(2 1 3 2) ec restricted
ARDL(2,1,3,2) regression
R-squared = 0.0175
Adj R-squared = 0.0135
Log likelihood = 5950.3079 Root MSE = 0.0388
D.ln_bprice Coefficient Std. err. t P>|t| [95% conf. interval]
ADJ
ln_bprice
L1. -.006387 .0018308 -3.49 0.000 -.0099767 -.0027973
LR
ln_ntrans 2.135988 .9257322 2.31 0.021 .3209067 3.951069
ln_ddestr .1936262 .2219016 0.87 0.383 -.2414558 .6287081
ln_fxeu_f 10.60878 2.015474 5.26 0.000 6.657047 14.56052
date .001778 .0006272 2.83 0.005 .0005483 .0030077
SR
ln_bprice
LD. -.038809 .0175277 -2.21 0.027 -.0731756 -.0044424
ln_ntrans
D1. .0097699 .0062652 1.56 0.119 -.0025141 .022054
ln_ddestr
D1. -.0027305 .0014205 -1.92 0.055 -.0055157 .0000547
LD. -.0033596 .0012433 -2.70 0.007 -.0057972 -.0009219
L2D. -.0027045 .0010154 -2.66 0.008 -.0046953 -.0007137
ln_fxeu_f
D1. .4200873 .1659749 2.53 0.011 .0946607 .745514
LD. -.2828991 .1662484 -1.70 0.089 -.608862 .0430638
ln_supply -.0283604 .0420257 -0.67 0.500 -.1107601 .0540393

_cons .0863434 .5877493 0.15 0.883 -1.066055 1.238742
The first coefficient in the ADJ section of the regression output is the negative speed-
of-adjustment coefficient πy = −α. Its magnitude is very small. At this stage, we
should not be fooled by the reported p-value and confidence interval into believing
that it is statistically significant. The t statistic for this coefficient does not have a
standard distribution under the null hypothesis. In fact, this is the test statistic that
we consider under the second step of the bounds test. The long-run coefficients θ in the
LR section all have the expected sign. Coin days destroyed is statistically insignificant,
which is not too surprising given the suspected differences in the integration orders. The
trend coefficient (date) is reported in the LR section because of the option restricted.
Without that option, it would be reported in the SR section together with the other
short-run coefficients ω, ψxi , and γ. This does not affect any of the other coefficients,
but it will matter for the bounds test. Because of first differencing, the lag orders for the
short-run terms are one less than those in the ARDL level representation. Also, note that
the exogenous supply variable is not transformed into first differences and just enters the
model with the same coefficient as in the ARDL specification. Before we start interpreting
the coefficient estimates, we first need to establish with the bounds test whether a long-
run relationship exists. We do this with the estat ectest postestimation command:
. estat ectest
Pesaran, Shin, and Smith (2001) bounds test
H0: no level relationship F = 6.156
Case 4 t = -3.489
Finite sample (3 variables, 3248 observations, 8 short-run coefficients)
Kripfganz and Schneider (2020) critical values and approximate p-values
10% 5% 1% p-value
I(0) I(1) I(0) I(1) I(0) I(1) I(0) I(1)
F 2.957 3.721 3.391 4.209 4.315 5.235 0.000 0.002

t -3.128 -3.819 -3.412 -4.128 -3.964 -4.709 0.041 0.186
do not reject H0 if
either F or t are closer to zero than critical values for I(0) variables
(if either p-value > desired level for I(0) variables)
reject H0 if
both F and t are more extreme than critical values for I(1) variables
(if both p-values < desired level for I(1) variables)
decision: no rejection (.a), inconclusive (.), or rejection (.r) at levels:
10% 5% 1%
decision . . .a
In the top-right corner, the bounds-test output displays the test statistics for the
first two testing steps, as outlined in section 2.3. The command reports the Kripfganz
and Schneider (2020) CVs for finite samples. However, because of the large sample size,
they are virtually identical with the asymptotic CVs.30 First, we consider the F statistic
for the joint null hypothesis πy = 0, πx = 0, and c1 = 0. The last coefficient captures
the restriction on the time trend. The test statistic is larger than the upper-bound
CVs—which would be the exact CVs if all long-run forcing variables were I(1)—for the
conventional significance levels. This is most easily seen by looking at the approximate
p-values, which are computed with the response-surface methodology of Kripfganz and
Schneider (2020). We therefore reject the null hypothesis in this first step. However,
this is not yet sufficient evidence in favor of a long-run relationship, because we need to
rule out the degenerate cases.
30. To use asymptotic CVs, we must specify the option asymptotic with estat ectest.
Second, we need to consider the individual null hypothesis πy = 0. The test statis-
tic is the same as in the ADJ section of the regression output, but only estat ectest
provides the appropriate CVs. Here the conclusion depends on the chosen significance
level. If we take a conservative stance with the 1% level, the t statistic is closer to zero
than the lower-bound CV. We would therefore not reject the null hypothesis. The statis-
tical significance of the long-run coefficients in the EC regression output then becomes
irrelevant, because the equilibrium correction term yt−1 − θxt drops out from (6) with
πy = α = 0 under the null hypothesis. In the long run, the log Bitcoin price can thus
be characterized by a unit-root process that is not driven by the independent variables
in our model.
As we move to a more relaxed stance on the risk of committing a type-I error—
rejecting the null hypothesis when it is actually true—the bounds test becomes incon-
clusive at the 5% significance level. Here the value of the t statistic falls inside the
two bounds, although it exceeds the lower bound only narrowly. Given the presence of
I(1) independent variables, the evidence still points more strongly toward not rejecting
the null hypothesis. When we move further to the 10% level, the test statistic remains
within the two bounds. Because the long-run coefficient of the only I(0) variable, coin
days destroyed, is statistically insignificant, the upper-bound CV carries much more
weight than the lower bound. As a way of resolving this inconclusiveness, we can redo
the bounds test for a model without coin days destroyed:
. ardl ln_bprice ln_ntrans ln_fxeu_f, exog(ln_supply) trendvar(date) aic
> maxlags(7) ec restricted
(output omitted )
. estat ectest, norule
Case 4 t = -3.219
10% 5% 1% p-value
I(0) I(1) I(0) I(1) I(0) I(1) I(0) I(1)
F 3.362 4.016 3.882 4.588 5.002 5.805 0.003 0.010

t -3.128 -3.647 -3.412 -3.952 -3.963 -4.531 0.081 0.222
decision: no rejection (.a), inconclusive (.), or rejection (.r) at levels:
10% 5% 1%
decision . .a .a
Assuming that all integration orders are known to be I(1), the bounds test still fails
to reject the null hypothesis because the t statistic is less negative than the upper-bound
CV at all significance levels. At the 5% level, it now even falls short of the lower bound.
Consequently, the statistical significance of the long-run coefficients is not informative.
Evidence of a cointegrating relationship could not be established.31
We can cross-check results with the Johansen (1995) framework using the vecrank
and vec commands. Here, unlike the ARDL framework, we should specify only nonsta-
tionary variables. To be on the safe side, we might want to initially run the augmented
Dickey and Fuller (1979) unit-root pretest for each of the variables. This can be done
with the dfuller command. Because this unit-root test is a special case of the bounds
test when there is only one variable, we can also use our ardl command for this purpose.
As an example, we show this for the Bitcoin price:32
. ardl ln_bprice, trendvar(date) aic maxlags(7) ec restricted

(output omitted )
. estat ectest, norule nodecision
Case 4 t = -1.663
10% 5% 1% p-value
I(0) I(1) I(0) I(1) I(0) I(1) I(0) I(1)
F 5.364 5.390 6.305 6.349 8.376 8.481 0.907 0.906

t -3.130 -3.132 -3.414 -3.423 -3.964 -4.000 0.769 0.771
. dfuller ln_bprice if e(sample), lags(`=e(lags)[1,1]-1') trend
Augmented Dickey--Fuller test for unit root
Variable: ln_bprice Number of obs = 3,248
Number of lags = 1
H0: Random walk with or without drift
Dickey--Fuller
Test critical value
statistic 1% 5% 10%
Z(t) -1.663 -3.960 -3.410 -3.120
MacKinnon approximate p-value for Z(t) = 0.7669.
31. If we had modeled an unrestricted trend by removing the option restricted, the qualitative con-
clusions would have remained unchanged because the decision primarily rests on the t statistic,
whose CVs are invariant to restricting the time trend.
32. Note that the dfuller command requests the lag order to be specified for the first-differenced
model, which is one less than the optimal lag order obtained with the ardl command for the level
model.
The t statistic from the bounds test equals the Dickey and Fuller (1979) test statistic.
The CVs also virtually coincide.33 The F test reported by estat ectest corresponds to
the Dickey and Fuller (1981) F statistic, which is not implemented elsewhere in Stata.
Both tests confirm our prior assessment that the log Bitcoin price is nonstationary.
While not shown here, similar tests for the other variables also support our initial
classification. We can now proceed with the cointegration rank tests. For the best
comparability with the previous results, we choose a lag order of 3—which was the
maximum lag order for any variable selected by the AIC in the ARDL model—and restrict
the estimation sample again to coincide with the one above:
. vecrank ln_bprice ln_ntrans ln_fxeu_f if date >= td(08jan2014), lags(3)

> trend(rtrend) max levela
Johansen tests for cointegration
Trend: Restricted Number of obs = 3,248
Sample: 2014-01-08 thru 2022-11-29 Number of lags = 3
Maximum Trace Critical value

rank Params LL Eigenvalue statistic 5% 1%
0 21 21528.446 109.1701 42.44 48.45
1 27 21574.8 0.02814 16.4624*1*5 25.32 30.45
2 31 21581.604 0.00418 2.8545 12.25 16.26
3 33 21583.031 0.00088
Maximum Eigenvalue Critical value

rank Params LL Maximum 5% 1%
0 21 21528.446 92.7077 25.54 30.34
1 27 21574.8 0.02814 13.6079 18.96 23.65
2 31 21581.604 0.00418 2.8545 12.52 16.26
3 33 21583.031 0.00088
* selected rank
The Johansen (1995) trace and maximum-eigenvalue tests both indicate a cointe-
gration rank of one. However, this is only a necessary but not sufficient condition for
the presence of an error-correction mechanism in the process of the log Bitcoin price.34
In the next step, we fit the VEC model:
33. Theoretically, the lower-bound and upper-bound CVs reported by estat ectest should coincide for
this special case. However, because they were obtained from separate response-surface regressions,
minor numerical discrepancies occur; see Kripfganz and Schneider (2020) for details. For smaller
sample sizes, these CVs can be more accurate than the original Dickey and Fuller (1979) ones,
especially when the regression is augmented with multiple lags.
34. It is possible that the cointegrating relationship comprises only variables other than the Bitcoin
price. Conversely, while there should be at most one cointegrating relationship involving the Bitcoin
price for the ARDL approach to be applicable, which is satisfied here, it is worth reiterating that a
cointegration rank larger than one for the entire system by itself would not automatically invalidate
the ARDL procedure. This is again due to the possibility of cointegrating relationships among the
other variables themselves; see also the discussion in section 2.2.
. vec ln_bprice ln_ntrans ln_fxeu_f if date >= td(08jan2014), rank(1) lags(3)

> trend(rtrend) alpha noetable
Vector error-correction model
AIC = -13.26835
Log likelihood = 21574.8 HQIC = -13.25023
Det(Sigma_ml) = 3.41e-10 SBIC = -13.21776
Cointegrating equations
Equation Parms chi2 P>chi2
_ce1 2 138.2818 0.0000
Identification: beta is exactly identified

Johansen normalization restriction imposed
beta Coefficient Std. err. z P>|z| [95% conf. interval]
_ce1
ln_bprice 1 . . . . .
ln_ntrans -3.169487 .3007237 -10.54 0.000 -3.758895 -2.580079
ln_fxeu_f -15.18478 1.632316 -9.30 0.000 -18.38406 -11.9855
_trend -.0010287 .0001443 -7.13 0.000 -.0013116 -.0007458
_cons 34.41289 . . . . .
Adjustment parameters
Equation Parms chi2 P>chi2
D_ln_bprice 1 21.04656 0.0000

D_ln_ntrans 1 68.06329 0.0000
D_ln_fxeu_f 1 .0319188 0.8582
alpha Coefficient Std. err. z P>|z| [95% conf. interval]
D_ln_bprice
_ce1
L1. -.0034951 .0007618 -4.59 0.000 -.0049882 -.0020019
D_ln_ntrans
_ce1
L1. .0186699 .002263 8.25 0.000 .0142345 .0231053
D_ln_fxeu_f
_ce1
L1. .0000144 .0000806 0.18 0.858 -.0001435 .0001723
Clearly, according to the bottom table for the speed-of-adjustment coefficients, the
USD/EUR exchange rate is not loading onto the cointegrating relationship. The re-
spective coefficient for the Bitcoin price is also very small. For practical matters, the
Bitcoin price hardly reacts to deviations from the equilibrium relationship. Thus, the
statistical question of whether there exists a long-run relationship should not bear too
much weight in the final assessment, because it would take a very long time for the
Bitcoin price to return to such an equilibrium. Somewhat problematic is the wrong
sign of the statistically significant adjustment coefficient for the number of transactions.
This points toward an instability in the system. By focusing only on the equation for
the Bitcoin price, the ARDL approach avoids this issue. Another disadvantage of the
vecrank command is that it does not allow inclusion of exogenous I(0) variables, unlike
ardl. Furthermore, by estimating a system of equations, the number of coefficients
to be estimated can be substantially larger in the VEC model. Notably, the long-run
coefficients are broadly consistent with the ARDL results, even though their relevance
should not be overstated because of the minuscule or nonexistent error adjustment.
So far, there is no convincing evidence in favor of a long-run relationship of the
Bitcoin price with traditional supply and demand side characteristics. However, the
demand for Bitcoin may generally depend on other or additional factors than those
for well-established currencies and investment assets. For example, it may depend on
how well the cryptocurrency market is understood and trusted by potential investors.
Furthermore, macrofinancial developments can affect the willingness to invest in high-
risk assets. Ciaian, Rajcaniova, and Kancs (2016) therefore include the number of views
of Bitcoin’s Wikipedia page (wikivw) as a measure of investment attractiveness, the Dow
Jones Industrial Average stock market index (djon_f) as a proxy for investor sentiment,
and the Brent crude oil price (oprc_f) as an indicator of macroeconomic risks.35 Given
its statistical insignificance, we remove the Bitcoin supply in the following specifications.
. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f ln_wikivw ln_djon_f ln_oprc_f,

> trend(date) aic maxlags(7) ec restricted maxcombs(2000000) dots
(output omitted )
(output omitted )
To economize on space, we summarize the results for the speed-of-adjustment coeffi-

cient (ADJ) and the long-run coefficients (LR, excluding the linear time trend) in table 1
instead of showing detailed Stata output. For ease of comparability, column 1 repeats
the results from our initial regression further above. Let us first look at column 2. The
stock market index and the oil price index do not appear to be relevant long-run forcing
variables for the Bitcoin price, irrespective of whether any long-run relationship exists
in the first place. In contrast, the long-run coefficient of Wikipedia views is statistically
significant. With its inclusion, the bounds test now also conclusively rejects the null
hypothesis, although only at the 10% level. However, the economic significance of this
result remains limited because of the very slow speed of adjustment. Compared with
the benchmark specification in column 1, the main statistical reason for the reduction
in the bounds-test p-values is the near doubling of the speed-of-adjustment coefficient,
which is immediately reflected in a larger t statistic. However, this cannot mask the
fact that the economic effect size is still negligible.
35. Data sources are https://wikishark.com (Vardi et al. 2021) for wikivw and https://fred.stlouisfed.
org (Federal Reserve Bank of St. Louis) for djon (S&P Dow Jones Indices LLC, FRED code DJIA)
and oprc (FRED code DCOILBRENTEU).
Table 1. ARDL long-run estimation results in EC representation
Column
D.ln_bprice 1 2 3 4 5
date >10jul2016 >10jul2016
<12may2020
ADJ L.ln_bprice −0.006 −0.011 −0.011 −0.010 −0.011
LR ln_ntrans 2.136∗∗ 1.285∗∗∗ 1.256∗∗∗ 1.556∗∗ 1.504∗
ln_ddestr 0.194 0.027 0.025 0.005 0.190
ln_fxeu_f 10.609∗∗∗ 7.862∗∗∗ 7.838∗∗∗ 8.538∗∗∗ 9.054∗∗
ln_wikivw 0.315∗∗∗ 0.320∗∗∗ 0.422∗∗ 0.389∗
ln_djon_f 1.396 1.514 0.453 0.431
ln_oprc_f 0.065 . .
F 6.156∗∗∗
∗∗∗ 5.198∗∗∗
∗∗∗ 5.934∗∗∗
∗∗∗ 3.985∗∗
∗∗∗ 2.872∗∗
t −3.489∗∗ −4.521∗∗∗∗ −4.522∗∗∗∗ −3.355∗ −2.706
note: Stars indicate the significance level (∗ p < 0.1; ∗∗ p < 0.05; ∗∗∗ p < 0.01.) Conven-
tional p-values for the ADJ coefficient are invalid and thus not reported. For the bounds
test, both upper-bound and lower-bound significance levels are indicated. A conclusive
decision requires either significance or insignificance with respect to both the upper and
lower bounds. Significance only with respect to the lower bound indicates inconclusiveness.
In column 3, we exclude the irrelevant oil prices from the model. Despite their in-
significant long-run coefficients, we keep the stock market index and coin days destroyed
because they still have significant short-run effects. The estimates hardly differ from
the previous specification.
. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f ln_wikivw ln_djon_f, trend(date) aic
> maxlags(7) ec restricted maxcombs(300000) dots
(output omitted )
(output omitted )
Over time, Bitcoin (and cryptocurrencies in general) became more and more acces-
sible to a wider audience and also attracted the interest of professional investors. This
may have lead to a gradual change in the fundamental relationship between the Bitcoin
price and its determinants. In econometric terms, we may have to worry about param-
eter instability. Stata offers several diagnostics for structural breaks, which we can use
here because the ardl command supports all standard postestimation commands for
regress. A routinely applied tool is the cumulative sum (CUSUM) test:
. estat sbcusum, ylabel(, angle(horizontal)) ttitle("") name(sb1)

Cumulative sum test for parameter stability
Sample: 08jan2014 thru 29nov2022 Number of obs = 3,248
H0: No structural break
Test Critical value
Type statistic 1% 5% 10%
Recursive 1.0067 1.1430 0.9479 0.8499

. estat sbcusum, ols ylabel(, angle(horizontal)) ttitle("") name(sb2)

Cumulative sum test for parameter stability
Sample: 08jan2014 thru 29nov2022 Number of obs = 3,248
Test Critical value
Type statistic 1% 5% 10%
OLS 0.4834 1.6276 1.3581 1.2238
. graph combine sb1 sb2, ysize(2) xsize(5)
Recursive cusum plot of D.ln_bprice OLS cusum plot of D.ln_bprice

with 95% confidence bands around the null with 95% confidence bands around the null
4 2
2 1
0 0
-2 -1
-4 -2
01jul2014 01jul2016 01jul2018 01jul2020 01jul2022 01jul2014 01jul2016 01jul2018 01jul2020 01jul2022
Figure 2. CUSUM plots
The CUSUM test based on OLS residuals does not trigger a warning sign. In contrast,
the test based on recursive residuals rejects the null hypothesis of parameter stability
at the 5% significance level. However, because the recursive CUSUM process travels
beyond the 95% confidence bounds only very briefly, we may not have to worry too
much. Figure 2 shows that the drift away from zero occurs rather gradually over time.
This does not suggest a specific date for a structural break, other than that it may have
occurred relatively early during our sample period. However, potential break points
can be spotted in figure 1. While the Bitcoin supply did not turn out to be a relevant
predictor in the earlier regressions, the discrete slowdowns in the mining of new Bitcoin
at July 10, 2016, and May 12, 2020—so-called “halving dates”36 —could possibly have
wider repercussions.
36. The Bitcoin mining algorithm ensures that the supply of new Bitcoins will ultimately come to an
end. After a certain number of new Bitcoin blocks have been created, the rewards for the mining
of a new Bitcoin are halved, driving inefficient miners out of the market and therefore reducing the
supply growth.
. estat sbknown, break(td(10jul2016) td(12may2020))

Wald test for a structural break: Known break dates
Sample: 08jan2014 thru 29nov2022
Break dates: 10jul2016 12may2020
Number of obs = 3,248
chi2(42) = 52.2187
Prob > chi2 = 0.0072
(output omitted )
. estat sbknown, break(td(10jul2016) td(12may2020))
> breakvars(L.ln_bprice ln_ntrans ln_ddestr ln_fxeu_f ln_wikivw ln_djon_f)
Wald test for a structural break: Known break dates
Sample: 08jan2014 thru 29nov2022
Break dates: 10jul2016 12may2020
Number of obs = 3,248
chi2(12) = 4.9395
Prob > chi2 = 0.9642
(output omitted )
Indeed, a parameter stability test with these known structural-break dates rejects
the null hypothesis. However, if we restrict the test to the speed-of-adjustment and
long-run coefficients, no instability is found. The latter is reassuring regarding our
earlier results. Accounting for structural breaks in the short-run coefficients would
become a potential issue if we were interested in a more detailed analysis of the short-
run dynamics. Nevertheless, as a robustness check, we refit the model by considering
only the observations after the first halving date. In another specification, we further
curtail the sample with the second halving date.
. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f ln_wikivw ln_djon_f

> if date >= td(10jul2016), trend(date) aic maxlags(7) ec restricted
> maxcombs(300000) dots
(output omitted )
(output omitted )
. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f ln_wikivw ln_djon_f
> if date > td(10jul2016) & date < td(12may2020), trend(date) aic maxlags(7)
> ec restricted maxcombs(300000) dots
(output omitted )
(output omitted )
The main results are shown in columns 4 and 5 of table 1. The effect sizes hardly
changed, especially for the speed of adjustment, which is instrumental for existence
of an equilibrium correction mechanism. Interestingly though, the bounds test now
does conclusively not reject the null hypothesis of no level relationship at least at the
5% significance level. Compared with the specifications in columns 2 and 3, this is
mainly driven by the larger standard error of the speed-of-adjustment coefficient, partly
because of the smaller sample size. Turning the argument around, the large size of the
unrestricted sample—daily observations for almost nine years—previously enabled us
to statistically detect (at the 10% significance level) an economically insignificant effect.
As a word of caution, the reliability of the bounds test could be hampered by the
nonnormality of the regression errors. Heteroskedasticity and normality tests with the
postestimation commands estat hettest and estat imtest tend to point in that di-
rection (not shown here).37 This is not unexpected when working with financial data.
Ideas for further exploration include the incorporation of potential asymmetric effects
and other nonlinearities, a quest for alternative explanatory variables, or a general-
ized autoregressive conditional heteroskedasticity modeling approach. We leave these
avenues to the interested reader.
Overall, based on the results presented here, there do not seem to be strong forces in
place that keep the log Bitcoin price in an equilibrium relationship with the candidate
long-run forcing variables. Even if we accept column 2 or 3 as our preferred specification
and take a liberal stand on the type-I error probability, the economic relevance of the
rejected bounds test remains negligible because of the slow speed of adjustment. It
appears that the price of Bitcoin is hardly driven by the underlying fundamentals but
might be following the path of a predominantly speculative asset. If we accept the
statistical conclusion from one of the other specifications that there is no significant
long-run relationship present, we could proceed by refitting a more parsimonious version
of the model purely in first differences, potentially also using the BIC instead of the AIC
as a lag order selection criterion. This could then be used for forecasting purposes or
further analyses of the dynamic adjustment processes. For the purpose of this article,
however, our curiosity shall end here.38
6 Conclusion
In this article, we have described the ardl command for the estimation of ARDL models
with time-series data. The lag orders can be prespecified or chosen optimally with the
AIC or BIC. For this purpose, the command is able to fit tens of thousands of candidate
models in virtually no time. Two useful reparameterizations of the model in error-
correction form allow for an interpretation of the coefficients as short-run and long-run
effects. The command further enables testing for the existence of a long-run level re-
lationship using the popular bounds test, which is implemented as a postestimation
37. For our application, increasing the ARDL lag orders does not provide an improvement.
38. In the working paper version of this article (Kripfganz and Schneider 2022), we present a forecasting
example with a different dataset.
feature. For nonstationary variables, this amounts to cointegration testing. Yet the
ARDL approach is flexible to allow for both stationary and nonstationary variables. The
package provides the recently improved Kripfganz and Schneider (2020) CVs for the
bounds test, which allow accurate inference for almost all practically relevant combina-
tions of sample size, number of long-run forcing variables, lag orders, and deterministic
model components.
7 Acknowledgments
We thank Michael Binder for his support and guidance during early stages of this
project. Moreover, we are grateful for numerous comments and suggestions from the
Stata community that helped to improve our ardl package. This includes countless
email communications, discussions on the Statalist forum, and exchanges of ideas at
the 2016 Stata Conference in Chicago, the 2017 and 2018 German Stata Users Group
meetings in Berlin and Konstanz, respectively, and the 2018 U.K. Stata Conference in
London.
8 Programs and supplemental material

To install the software files as they exist at the time of publication of this article, type
. net sj 23-4
. net install st0734 (to install program files, if available)
. net get st0734 (to install ancillary files, if available)
9 References
Adeleye, N., E. Osabuohien, E. Bowale, O. Matthew, and E. Oduntan. 2018. Fi-
nancial reforms and credit growth in Nigeria: Empirical insights from ARDL and
ECM techniques. International Review of Applied Economics 32: 807–820. https:
// doi.org / 10.1080 / 02692171.2017.1375466.
Ang, J. B. 2007. CO2 emissions, energy consumption, and output in France. Energy
Policy 35: 4772–4778. https: // doi.org / 10.1016 / j.enpol.2007.03.032.
. 2010. Finance and inequality: The case of India. Southern Economic Journal
76: 738–761. https: // doi.org / 10.4284 / sej.2010.76.3.738.
Aruga, K., M. M. Islam, and A. Jannat. 2020. Effects of COVID-19 on Indian en-
ergy consumption. Southern Economic Journal 12: 5616. https: // doi.org / 10.3390 /
su12145616.
Bahmani-Oskooee, M., and T. J. Brooks. 1999. Bilateral J-curve between U.S. and
her trading partners. Weltwirtschaftliches Archiv 135: 156–165. https://doi.org/10.
1007 / BF02708163.
Bekun, F. V., F. Emir, and S. A. Sarkodie. 2019. Another look at the relationship be-
tween energy consumption, carbon dioxide emissions, and economic growth in South
Africa. Science of the Total Environment 655: 759–765. https: // doi.org / 10.1016 / j.
scitotenv.2018.11.271.
Belloumi, M. 2014. The relationship between trade, FDI and economic growth in Tunisia:
An application of the autoregressive distributed lag model. Economic Systems 38:
269–287. https: // doi.org / 10.1016 / j.ecosys.2013.09.002.
Blackburne, E. F., III, and M. W. Frank. 2007. Estimation of nonstationary
heterogeneous panels. Stata Journal 7: 197–208. https: // doi.org / 10.1177 /
1536867X0700700204.
Büyükşahin, B., and M. A. Robe. 2014. Speculators, commodities and cross-market
linkages. Journal of International Money and Finance 42: 38–70. https: // doi.org /
10.1016 / j.jimonfin.2013.08.004.
Ciaian, P., M. Rajcaniova, and d. Kancs. 2016. The economics of BitCoin price forma-
tion. Applied Economics 48: 1799–1815. https: // doi.org / 10.1080 / 00036846.2015.
1109038.
. 2018. Virtual relationships: Short- and long-run evidence from BitCoin and
altcoin markets. Journal of International Financial Markets, Institutions and Money
52: 173–195. https: // doi.org / 10.1016 / j.intfin.2017.11.001.
Davis, E. P., and H. Zhu. 2011. Bank lending and commercial property cycles: Some
cross-country evidence. Journal of International Money and Finance 30: 1–21. https:
// doi.org / 10.1016 / j.jimonfin.2010.06.005.
De Vita, G., and A. Abbott. 2004. Real exchange rate volatility and U.S. exports: An
ARDL bounds testing approach. Economic Issues 9: 69–78.
Dickey, D. A., and W. A. Fuller. 1979. Distribution of the estimators for autoregressive
time series with a unit root. Journal of the American Statistical Association 74:
427–431. https: // doi.org / 10.2307 / 2286348.
. 1981. Likelihood ratio statistics for autoregressive time series with a unit root.
Econometrica 49: 1057–1072. https: // doi.org / 10.2307 / 1912517.
Ditzen, J. 2018. Estimating dynamic common-correlated effects in Stata. Stata Journal
18: 585–617. https: // doi.org / 10.1177 / 1536867X1801800306.
. 2021. Estimating long-run effects and the exponent of cross-sectional depen-
dence: An update to xtdcce2. Stata Journal 21: 687–707. https: // doi.org / 10.1177 /
1536867X211045560.
Doğan, B., M. B. Jebli, K. Shahzad, T. H. Farooq, and U. Shahzad. 2020. Investigating
the effects of meteorological parameters on COVID-19: Case study of New Jersey,
United States. Environmental Research 191: 110148. https: // doi.org / 10.1016 / j.
envres.2020.110148.
Engle, R. F., and C. W. J. Granger. 1987. Co-integration and error correction: Repre-
sentation, estimation, and testing. Econometrica 55: 251–276. https: // doi.org / 10.
2307 / 1913236.
Enisan, A. A., and A. O. Olufisayo. 2009. Stock market development and economic
growth: Evidence from seven sub-Sahara African countries. Journal of Economics
and Business 61: 162–171. https: // doi.org / 10.1016 / j.jeconbus.2008.05.001.
Esaku, S. 2022. Institutionalized democracy and the shadow economy in the short- and
long-run: Empirical analysis from Uganda. Humanities and Social Sciences Commu-
nications 9: 165. https: // doi.org / 10.1057 / s41599-022-01128-1.
Fatai, K., L. Oxley, and F. G. Scrimgeour. 2004. Modelling the causal relationship be-
tween energy consumption and GDP in New Zealand, Australia, India, Indonesia, The
Philippines and Thailand. Mathematics and Computers in Simulation 64: 431–445.
https: // doi.org / 10.1016 / S0378-4754(03)00109-5.
Fedderke, J. W., P. Perkins, and J. M. Luiz. 2006. Infrastructural investment in long-

run economic growth: South Africa 1875–2001. World Development 34: 1037–1059.
https: // doi.org / 10.1016 / j.worlddev.2005.11.004.
Frankel, J., S. L. Schmukler, and L. Servén. 2004. Global transmission of interest rates:
Monetary independence and currency regime. Journal of International Money and
Finance 23: 701–733. https: // doi.org / 10.1016 / j.jimonfin.2004.03.006.
Halicioglu, F. 2009. An econometric study of CO2 emissions, energy consumption, in-

come and foreign trade in Turkey. Energy Policy 37: 1156–1164. https://doi.org/10.
1016 / j.enpol.2008.11.012.
Hassler, U., and J. Wolters. 2006. Autoregressive distributed lag models and cointegra-
tion. Allgemeines Statistisches Archiv 90: 59–74. https: // doi.org / 10.1007 / s10182-
006-0221-5.
He, F., Z. Hu, W. Zhang, L. Cai, G. Cai, and K. Aoyagi. 2017. Construction and
evaluation of two computational models for predicting the incidence of influenza in
Nagasaki Prefecture, Japan. Scientific Reports 7: 7192. https: // doi.org / 10.1038 /
s41598-017-07475-3.
Jalil, A., and S. F. Mahmud. 2009. Environment Kuznets curve for CO2 emissions: A
cointegration analysis for China. Energy Policy 37: 5167–5172. https: // doi.org / 10.
1016 / j.enpol.2009.07.044.
Johansen, S. 1995. Likelihood-Based Inference in Cointegrated Vector Autoregressive

Models. Oxford: Oxford University Press. https://doi.org/10.1093/0198774508.001.
0001.
Jordan, S., and A. Q. Philips. 2018. Cointegration testing and dynamic simulations of
autoregressive distributed lag models. Stata Journal 18: 902–923. https: // doi.org /
10.1177 / 1536867X1801800409.
Katircioglu, S. T. 2009. Revisiting the tourism-led-growth hypothesis for Turkey using

the bounds test and Johansen approach for cointegration. Tourism Management 30:
17–20. https: // doi.org / 10.1016 / j.tourman.2008.04.004.
Kirikkaleli, D., H. Güngör, and T. S. Adebayo. 2022. Consumption-based carbon emis-

sions, renewable energy consumption, financial development and economic growth in
Chile. Business Strategy and the Environment 31: 1123–1137. https: // doi.org / 10.
1002 / bse.2945.
Kripfganz, S., and V. Sarafidis. 2021. Instrumental-variable estimation of large-T panel-

data models with common factors. Stata Journal 21: 659–686. https: // doi.org / 10.
1177 / 1536867X211045558.
Kripfganz, S., and D. C. Schneider. 2020. Response surface regressions for critical value
bounds and approximate p-values in equilibrium correction models. Oxford Bulletin
of Economics and Statistics 82: 1456–1481. https: // doi.org / 10.1111 / obes.12377.
. 2022. ardl: Estimating autoregressive distributed lag and equilibrium correction

models. Research Center for Policy Design Discussion Paper 2TUPD-2022-006, Tohoku
University. http: // hdl.handle.net / 10097 / 00135205.
Lee, G. H. Y., and S. P. Lee. 2014. Childcare availability, fertility and female labor
force participation in Japan. Journal of the Japanese and International Economies
32: 71–85. https: // doi.org / 10.1016 / j.jjie.2014.01.002.
Morley, B. 2006. Causality between economic growth and immigration: An ARDL

bounds testing approach. Economics Letters 90: 72–76. https: // doi.org / 10.1016 / j.
econlet.2005.07.008.
Murthy, V. N. R., and A. A. Okunade. 2016. Determinants of U.S. health expendi-

ture: Evidence from autoregressive distributed lag (ARDL) approach to cointegration.
Economic Modelling 59: 67–73. https: // doi.org / 10.1016 / j.econmod.2016.07.001.
Narayan, P. K. 2005. The saving and investment nexus for China: Evidence from
cointegration tests. Applied Economics 37: 1979–1990. https: // doi.org / 10.1080 /
00036840500278103.
Narayan, P. K., and R. Smyth. 2005. Electricity consumption, employment and real
income in Australia evidence from multivariate Granger causality tests. Energy Policy
33: 1109–1116. https: // doi.org / 10.1016 / j.enpol.2003.11.010.
Narayan, P. K., R. Smyth, and M. Nandha. 2004. Interdependence and dynamic linkages
between the emerging stock markets of South Asia. Accounting and Finance 44:
419–439. https: // doi.org / 10.1111 / j.1467-629x.2004.00113.x.
Ntanos, S., M. Skordoulis, G. Kyriakopoulos, G. Arabatzis, M. Chalikias, S. Galatsidas,

A. Batzios, and A. Katsarou. 2018. Renewable energy and economic growth: Evi-
dence from European countries. Sustainability 10: 2626. https: // doi.org / 10.3390 /
su10082626.
Obstfeld, M., J. C. Shambaugh, and A. M. Taylor. 2005. The trilemma in history:

Tradeoffs among exchange rates, monetary policies, and capital mobility. Review of
Economics and Statistics 87: 423–438. https: // doi.org / 10.1162 / 0034653054638300.
Oteng-Abayie, E. F., and J. M. Frimpong. 2006. Bounds testing approach to cointe-

gration: An examination of foreign direct investment trade and growth relationships.
American Journal of Applied Sciences 3: 2079–2085.
Persyn, D., and J. Westerlund. 2008. Error-correction–based cointegration tests for

panel data. Stata Journal 8: 232–241. https://doi.org/10.1177/1536867X0800800205.
Pesaran, M. H., and Y. Shin. 1998. An autoregressive distributed-lag modelling ap-

proach to cointegration analysis. In Econometrics and Economic Theory in the Twen-
tieth Century: The Ragner Frisch Centennial Symposium, ed. S. Steiner, 371–413.
Cambridge: Cambridge University Press. https://doi.org/10.1017/CCOL521633230.
011.
Pesaran, M. H., Y. Shin, and R. J. Smith. 2001. Bounds testing approaches to the
analysis of level relationships. Journal of Applied Econometrics 16: 289–326. https:
// doi.org / 10.1002 / jae.616.
Phillips, P. C. B., and B. E. Hansen. 1990. Statistical inference in instrumental variables

regression with I(1) processes. Review of Economic Studies 57: 99–125. https: // doi.
org / 10.2307 / 2297545.
Sari, R., S. Hammoudeh, and U. Soytas. 2010. Dynamics of oil price, precious metal
prices, and exchange rate. Energy Economics 32: 351–362. https://doi.org/10.1016/
j.eneco.2009.08.010.
Schaffer, M. E. 2010. egranger: Stata module to perform Engle–Granger cointegration

tests and 2-step ECM estimation. Statistical Software Components S457210, Depart-
ment of Economics, Boston College. https://ideas.repec.org/c/boc/bocode/s457210.
html.
Shambaugh, J. C. 2004. The effect of fixed exchange rates on monetary policy. Quarterly
Journal of Economics 119: 301–352. https: // doi.org / 10.1162 / 003355304772839605.
Shin, Y., B. Yu, and M. Greenwood-Nimmo. 2014. Modelling asymmetric cointegration

and dynamic multipliers in a nonlinear ARDL framework. In Festschrift in Honor of
Peter Schmidt: Econometric Methods and Applications, ed. W. C. Horrace and R. C.
Sickles, 281–314. New York: Springer. https://doi.org/10.1007/978-1-4899-8008-3_9.
Song, H., S. Lin, S. F. Witt, and X. Zhang. 2011. Impact of financial/economic crisis
on demand for hotel rooms in Hong Kong. Tourism Management 32: 172–186. https:
// doi.org / 10.1016 / j.tourman.2010.05.006.
Stoian, A., and F. Iorgulescu. 2020. Fiscal policy and stock market efficiency: An ARDL
bounds testing approach. Economic Modelling 90: 406–416. https: // doi.org / 10.
1016 / j.econmod.2019.12.023.
Vardi, E., L. Muchnik, A. Conway, and M. Breakstone. 2021. WikiShark: An online tool
for analyzing Wikipedia traffic and trends. In WWW ’21: Companion Proceedings of
the Web Conference 2021, 558–571. New York: Association for Computing Machinery.
https: // doi.org / 10.1145 / 3442442.3452341.
Varona, L., and J. R. Gonzales. 2021. Dynamics of the impact of COVID-19 on the
economic activity of Peru. PLOS ONE 16: e0244920. https: // doi.org / 10.1371 /
journal.pone.0244920.
Wang, Q. 2012. Long-run covariance and its applications in cointegration regression.
Stata Journal 12: 515–542. https: // doi.org / 10.1177 / 1536867X1201200312.
Wang, Y.-S. 2009. The impact of crisis events and macroeconomic activity on Taiwan’s
international inbound tourism demand. Tourism Management 30: 75–82. https: //
doi.org / 10.1016 / j.tourman.2008.04.010.
Wolde-Rufael, Y. 2006. Electricity consumption and economic growth: A time series
experience for 17 African countries. Energy Policy 34: 1106–1114. https: // doi.org /
10.1016 / j.enpol.2004.10.008.
Zhang, H., H. Song, L. Wen, and C. Liu. 2021. Forecasting tourism recovery amid
COVID-19. Annals of Tourism Research 87: 103149. https: // doi.org / 10.1016 / j.
annals.2021.103149.
Zhang, X., E. A. Davidson, D. L. Mauzerall, T. D. Searchinger, P. Dumas, and Y. Shen.

2015. Managing nitrogen for sustainable development. Nature 528: 51–59. https:
// doi.org / 10.1038 / nature15743.
About the authors

Sebastian Kripfganz received his PhD in economics from Goethe University Frankfurt in 2015
and is now a senior lecturer in econometrics at the University of Exeter Business School. He
has a strong research interest in methods for estimation and inference in dynamic econometric
models.
Daniel Schneider received his PhD in economics from Goethe University Frankfurt in 2013
and is now a statistical programmer and research scientist at the Max Planck Institute for
Demographic Research in Rostock. His research interests include time-series analysis and
discrete-time multistate models.

Kripfganz Schneider 2023 Ardl Estimating Autoregressive Distributed Lag and Equilibrium Correction Models

Uploaded by

Copyright:

Available Formats

Kripfganz Schneider 2023 Ardl Estimating Autoregressive Distributed Lag and Equilibrium Correction Models

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kripfganz Schneider 2023 Ardl Estimating Autoregressive Distributed Lag and Equilibrium Correction Models

Uploaded by

Copyright:

Available Formats

The Stata Journal (2023)

23, Number 4, pp. 983–1019 DOI: 10.1177/1536867X231212434

ardl: Estimating autoregressive distributed lag

Abstract. We present a command, ardl, for the estimation of autoregressive

© 2023 StataCorp LLC st0734

1. Occasionally, the abbreviation ADL is used in the literature instead of ARDL.

2 Econometric model and methods

t = 1 + p∗ , . . . , T . Leaving aside the variables zt , this is a general ARDL (p, q, . . . , q)

variables xt are weakly exogenous/long-run forcing—ruling out any contemporaneous

2.2 Error-correction representation

Now recall the hypothesized long-run equilibrium relationship between yt and xt in

of xt determine the integration order of yt . If the variables in xt with nonzero long-run

2.3 Bounds test

1. No deterministic model components are included (c0 = c1 = 0).

1. First, we test the joint null hypothesis

versus the alternative hypothesis

The test statistic is a conventional t statistic for statistical insignificance of the

2.4 Practical guidelines

which is a restricted version of (7) with πy = 0 and πx = 0. In both cases, it

To avoid pretesting problems, keep model simplifications—like those at stage 6—to

3 The ardl command

ardl depvar indepvars if in

noconstant trendvar (trendvarname) restricted regstore(storename)

perfect maxlags(numlist) aic bic maxcombs(combnum)

3.3 Stored results

20. Starred results (∗ ) are not always stored.

estat ectest , siglevels(numlist) asymptotic nocritval norule

11.5 16.9 20.0 0.4

4.5 16.2 10.0 -0.1

left axis: ln_bprice left axis: ln_ntrans ln_ddestr

Figure 1. Time-series graphs for the main regression variables

ln_bprice Coefficient Std. err. t P>|t| [95% conf. interval]

ln_supply -.0283604 .0420257 -0.67 0.500 -.1107601 .0540393

lags(p) chi2 df Prob > chi2

H0: no serial correlation

. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f if e(sample), exog(ln_supply)

D.ln_bprice Coefficient Std. err. t P>|t| [95% conf. interval]

ln_supply -.0283604 .0420257 -0.67 0.500 -.1107601 .0540393

F 2.957 3.721 3.391 4.209 4.315 5.235 0.000 0.002

F 3.362 4.016 3.882 4.588 5.002 5.805 0.003 0.010

. ardl ln_bprice, trendvar(date) aic maxlags(7) ec restricted

F 5.364 5.390 6.305 6.349 8.376 8.481 0.907 0.906

Z(t) -1.663 -3.960 -3.410 -3.120

MacKinnon approximate p-value for Z(t) = 0.7669.

. vecrank ln_bprice ln_ntrans ln_fxeu_f if date >= td(08jan2014), lags(3)

Maximum Trace Critical value

Maximum Eigenvalue Critical value

. vec ln_bprice ln_ntrans ln_fxeu_f if date >= td(08jan2014), rank(1) lags(3)

_ce1 2 138.2818 0.0000

Identification: beta is exactly identified

beta Coefficient Std. err. z P>|z| [95% conf. interval]

D_ln_bprice 1 21.04656 0.0000

alpha Coefficient Std. err. z P>|z| [95% conf. interval]

. ardl ln_bprice ln_ntrans ln_ddestr ln_fxeu_f ln_wikivw ln_djon_f ln_oprc_f,

To economize on space, we summarize the results for the speed-of-adjustment coeffi-

Table 1. ARDL long-run estimation results in EC representation

. estat sbcusum, ylabel(, angle(horizontal)) ttitle("") name(sb1)

Recursive 1.0067 1.1430 0.9479 0.8499

. estat sbcusum, ols ylabel(, angle(horizontal)) ttitle("") name(sb2)

OLS 0.4834 1.6276 1.3581 1.2238