From 63a116fb2840ed8300fd7c88532e8d4fe355bdaa Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?S=C3=A9bastien=20Villemot?= <sebastien@dynare.org>
Date: Tue, 3 May 2022 17:46:59 +0200
Subject: [PATCH] =?UTF-8?q?New=20+=3D=20and=20*=3D=20syntaxes=20in=20?=
=?UTF-8?q?=E2=80=9Cendval(learnt=5Fin=3D=E2=80=A6)=E2=80=9D=20blocks?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
doc/manual/source/the-model-file.rst | 30 +++++++++++++++----
..._foresight_with_expectation_errors_setup.m | 12 +++++++-
preprocessor | 2 +-
.../pfwee_learnt_in.mod | 8 ++---
4 files changed, 40 insertions(+), 12 deletions(-)
diff --git a/doc/manual/source/the-model-file.rst b/doc/manual/source/the-model-file.rst
index 4fe506aeb9..0c41bd2b55 100644
--- a/doc/manual/source/the-model-file.rst
+++ b/doc/manual/source/the-model-file.rst
@@ -3915,6 +3915,7 @@ and ``endval`` blocks which are given a special ``learnt_in`` option.
end;
This syntax means that:
+
- from the perspective of period 1, ``x`` is expected to be equal to 1 in
periods 1 and 2, to 1.2 in periods 3 and 4, and to 1.4 in period 5;
- from the perspective of periods 2 (and 3), ``x`` is expected to be
@@ -3925,23 +3926,40 @@ and ``endval`` blocks which are given a special ``learnt_in`` option.
.. block:: endval(learnt_in=INTEGER) ;
- |br| The ``endval(learnt_in=INTEGER)`` can be used to specify temporary
- shocks that are learnt in a specific period.
+ |br| The ``endval(learnt_in=INTEGER)`` can be used to specify terminal
+ conditions that are learnt in a specific period.
Note that an ``endval(learnt_in=1)`` block is equivalent to a regular
:bck:`endval` block.
+ It is possible to express the terminal condition by specifying the level of
+ the exogenous variable (using an equal symbol, as in a regular
+ :bck:`endval` blocks without the ``learnt_in`` option). But it is also
+ possible to express the terminal condition as an addition to the value
+ expected from the perspective of the previous previous period (using the
+ ``+=`` operator), or as a multiplicative factor over that previously
+ expected value (using the ``*=`` operator).
+
*Example*
::
- endval(learnt_in = 2);
+ endval(learnt_in = 3);
x = 1.1;
+ y += 0.1;
+ z *= 2;
end;
- This syntax means that, in period 2, the agents learn that the terminal
- condition for ``x`` will be 1.1. This value will be the realized one,
- unless there is another ``endval(learnt_in=p)`` block with ``p>2``.
+ This syntax means that, in period 3, the agents learn that:
+
+ - the terminal condition for ``x`` will be 1.1;
+ - the terminal condition for ``y`` will be 0.1 above the terminal
+ condition for ``y`` that was expected from the perspective of period 2;
+ - the terminal condition for ``z`` will be 2 times the terminal condition
+ for ``z`` that was expected from the perspective of period 2.
+
+ Those values will be the realized ones, unless there is another
+ ``endval(learnt_in=p)`` block with ``p>3``.
.. command:: perfect_foresight_with_expectation_errors_setup ;
perfect_foresight_with_expectation_errors_setup (OPTIONS...);
diff --git a/matlab/perfect-foresight-models/perfect_foresight_with_expectation_errors_setup.m b/matlab/perfect-foresight-models/perfect_foresight_with_expectation_errors_setup.m
index 4664d627d1..2e7ad0b599 100644
--- a/matlab/perfect-foresight-models/perfect_foresight_with_expectation_errors_setup.m
+++ b/matlab/perfect-foresight-models/perfect_foresight_with_expectation_errors_setup.m
@@ -90,7 +90,17 @@ else
idx = find([M_.learnt_endval.learnt_in] == p);
for i = 1:length(idx)
j = idx(i);
- oo_.pfwee.terminal_info(M_.learnt_endval(j).exo_id, p) = M_.learnt_endval(j).value;
+ exo_id = M_.learnt_endval(j).exo_id;
+ switch M_.learnt_endval(j).type
+ case 'level'
+ oo_.pfwee.terminal_info(exo_id, p) = M_.learnt_endval(j).value;
+ case 'add'
+ oo_.pfwee.terminal_info(exo_id, p) = oo_.pfwee.terminal_info(exo_id, p) + M_.learnt_endval(j).value;
+ case 'multiply'
+ oo_.pfwee.terminal_info(exo_id, p) = oo_.pfwee.terminal_info(exo_id, p) * M_.learnt_endval(j).value;
+ otherwise
+ error('Unknown type in M_.learnt_endval')
+ end
end
end
oo_.pfwee.shocks_info(:, :, p) = oo_.pfwee.shocks_info(:, :, p-1);
diff --git a/preprocessor b/preprocessor
index 8c07fb5e43..3a820fffa2 160000
--- a/preprocessor
+++ b/preprocessor
@@ -1 +1 @@
-Subproject commit 8c07fb5e4310b2ec2988dc62256e1cf0971c05dd
+Subproject commit 3a820fffa2f7291be5732c33bde552526eab1602
diff --git a/tests/deterministic_simulations/pfwee_learnt_in.mod b/tests/deterministic_simulations/pfwee_learnt_in.mod
index 878beb00b7..5c7e396ddb 100644
--- a/tests/deterministic_simulations/pfwee_learnt_in.mod
+++ b/tests/deterministic_simulations/pfwee_learnt_in.mod
@@ -50,7 +50,7 @@ shocks(learnt_in = 3);
end;
endval(learnt_in = 3);
- x = 1.2;
+ x += 0.1;
end;
// Dummy block, that will be overwritten by the next one
@@ -67,7 +67,7 @@ shocks(learnt_in = 6, overwrite);
end;
endval(learnt_in = 6);
- x = 1.1;
+ x *= 0.75;
end;
perfect_foresight_with_expectation_errors_setup(periods = 7);
@@ -106,7 +106,7 @@ oo_.exo_simul = [ saved_exo; oo_.exo_simul ];
// Information arriving in period 3 (temp shocks + permanent shock in future)
oo_.exo_simul(4,1) = 1.4;
oo_.exo_simul(8,1) = 1.5;
-oo_.exo_steady_state = 1.2;
+oo_.exo_steady_state = 1.1+0.1;
oo_.exo_simul(end, 1) = oo_.exo_steady_state;
oo_.steady_state = evaluate_steady_state(oo_.steady_state, M_, options_, oo_, true);
oo_.endo_simul(:, end) = oo_.steady_state;
@@ -122,7 +122,7 @@ oo_.exo_simul = [ saved_exo; oo_.exo_simul ];
// Information arriving in period 6 (temp shocks + permanent shock)
oo_.exo_simul(7,1) = 1*0.8;
oo_.exo_simul(8,1) = 1.5*0.8;
-oo_.exo_steady_state = 1.1;
+oo_.exo_steady_state = (1.1+0.1)*0.75;
oo_.exo_simul(end, 1) = oo_.exo_steady_state;
oo_.steady_state = evaluate_steady_state(oo_.steady_state, M_, options_, oo_, true);
oo_.endo_simul(:, end) = oo_.steady_state;
--
GitLab