mergeron 2025.739319.1__tar.gz → 2025.739319.3__tar.gz

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: mergeron
-Version: 2025.739319.1
+Version: 2025.739319.3
 Summary: Analyze merger enforcement policy using Python
 License: MIT
 Keywords: merger policy analysis,merger guidelines,merger screening,policy presumptions,concentration standards,upward pricing pressure,GUPPI
@@ -93,10 +93,6 @@ backend to ``None`` to skip fine-tuning plots for PDF generation.
     _ = fig.legend(loc=(0.4, 0.7), frameon=False)
 
 
-
-.. image:: ./docs/readme_content/output_5_0.png
-
-
 *Analyzing FTC Merger Investigations Data*
 
 .. code:: python

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/README.rst

@@ -52,10 +52,6 @@ backend to ``None`` to skip fine-tuning plots for PDF generation.
     _ = fig.legend(loc=(0.4, 0.7), frameon=False)
 
 
-
-.. image:: ./docs/readme_content/output_5_0.png
-
-
 *Analyzing FTC Merger Investigations Data*
 
 .. code:: python

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/pyproject.toml

@@ -13,7 +13,7 @@ keywords = [
     "upward pricing pressure",
     "GUPPI",
 ]
-version = "2025.739319.1"
+version = "2025.739319.3"
 
 # Classifiers list: https://pypi.org/classifiers/
 classifiers = [

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/__init__.py

@@ -12,7 +12,7 @@ from ruamel import yaml
 
 _PKG_NAME: str = Path(__file__).parent.stem
 
-VERSION = "2025.739319.1"
+VERSION = "2025.739319.3"
 
 __version__ = VERSION
 

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/core/empirical_margin_distribution.py

@@ -43,11 +43,12 @@ from types import MappingProxyType
 
 import numpy as np
 import urllib3
+from joblib import Parallel, delayed
 from numpy.random import PCG64DXSM, Generator, SeedSequence
 from scipy import stats  # type: ignore
 from xlrd import open_workbook  # type: ignore
 
-from .. import VERSION, ArrayDouble, this_yaml  # noqa: TID252
+from .. import NTHREADS, VERSION, ArrayDouble, this_yaml  # noqa: TID252
 from .. import WORK_DIR as PKG_WORK_DIR  # noqa: TID252
 from .. import data as mdat  # noqa: TID252
 from . import _mappingproxy_from_mapping
@@ -67,7 +68,7 @@ u3pm = urllib3.PoolManager()
 def margin_data_getter(  # noqa: PLR0912
     _table_name: str = "margin",
     *,
-    data_archive_path: Path | None = None,
+    data_archive_path: Path = MGNDATA_ARCHIVE_PATH,
     data_download_flag: bool = False,
 ) -> DamodaranMarginData:
     if _table_name != "margin":  # Not validated for other tables
@@ -203,11 +204,12 @@ def margin_data_builder(
 
 
 def margin_data_resampler(
-    _dist_parms: tuple[ArrayDouble, ArrayDouble] | None = None,
+    _dist_parms: ArrayDouble,
     /,
     *,
-    sample_size: int | tuple[int, ...] = (10**6, 2),
+    sample_size: int | tuple[int, ...],
     seed_sequence: SeedSequence | None = None,
+    nthreads: int = NTHREADS,
 ) -> ArrayDouble:
     """
     Generate draws from the empirical distribution bassed on Prof. Damodaran's margin data.
@@ -231,6 +233,8 @@ def margin_data_resampler(
         SeedSequence for seeding random-number generator when results
         are to be repeatable
 
+    nthreads
+        Number of threads to use in generating margin data.
     Returns
     -------
         Array of margin values
@@ -241,22 +245,26 @@ def margin_data_resampler(
 
     _seed = seed_sequence or SeedSequence(pool_size=8)
 
-    _x, _w = _dist_parms[:, 0], _dist_parms[:, 1] 
+    _x, _w = _dist_parms[:, 0], _dist_parms[:, 1]
 
     margin_kde = stats.gaussian_kde(_x, weights=_w, bw_method="silverman")
     margin_kde.set_bandwidth(bw_method=margin_kde.factor / 3.0)
 
     if isinstance(sample_size, int):
-        return np.array(
-            margin_kde.resample(sample_size, seed=Generator(PCG64DXSM(_seed)))[0]
-        )
+        return margin_kde.resample(sample_size, seed=Generator(PCG64DXSM(_seed))).T
+
     elif isinstance(sample_size, tuple) and len(sample_size) == 2:
-        _ssz, _ncol = sample_size
         ret_array = np.empty(sample_size, float)
-        for idx, _col_seed in enumerate(_seed.spawn(_ncol)):
-            ret_array[:, idx] = margin_kde.resample(
-                _ssz, seed=Generator(PCG64DXSM(_col_seed))
-            )[0]
+
+        _ssz, _ncol = sample_size
+        dat_list = Parallel(n_jobs=min(nthreads, _ncol), prefer="threads")(
+            delayed(margin_kde.resample)(_ssz, seed=Generator(PCG64DXSM(_col_seed)))
+            for _col_seed in _seed.spawn(_ncol)
+        )
+
+        for _i in range(_ncol):
+            ret_array[:, [_i]] = dat_list[_i].T
+
         return ret_array
     else:
         raise ValueError(f"Invalid sample size: {sample_size!r}")

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/core/guidelines_boundaries.py

@@ -109,7 +109,7 @@ class GuidelinesThresholds:
                 _r := gbfn.round_cust(_fc / (_fc + 1), frac=0.05),
                 _g := guppi_from_delta(dh_s, m_star=1.0, r_bar=_r),
                 _dr := 1 - _r,
-                _cmcr := 0.03,  # Not strictly a Guidelines standard
+                _cmcr := _g,  # Not strictly a Guidelines standard
                 _ipr := _g,  # Not strictly a Guidelines standard
             ),
         )
@@ -132,7 +132,7 @@ class GuidelinesThresholds:
                     _g,
                     (1 - _r_i) / 2,
                     _cmcr,
-                    _ipr := _g,
+                    _ipr,
                 )
                 if self.pub_year == 2010
                 else HMGThresholds(

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/demo/visualize_empirical_margin_distribution.py

@@ -21,11 +21,14 @@ WORK_DIR = globals().get("WORK_DIR", PKG_WORK_DIR)
 
 SAMPLE_SIZE = 10**6
 BIN_COUNT = 25
-margin_data_obs, margin_data_wts, margin_data_stats = emd.margin_data_builder()
+margin_data, margin_data_stats = emd.margin_data_builder()
+
+margin_data_obs, margin_data_wts = margin_data[:, 0], margin_data[:, 1]
+
 print(repr(margin_data_obs))
 print(repr(margin_data_stats))
 
-plt, mgn_fig, mgn_ax, set_axis_def = boundary_plot(mktshares_plot_flag=False)
+plt, mgn_fig, mgn_ax, set_axis_def = boundary_plot(mktshare_plot_flag=False)
 mgn_fig.set_figheight(6.5)
 mgn_fig.set_figwidth(9.0)
 

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/gen/__init__.py

@@ -299,7 +299,7 @@ class PCMDistribution(str, Enameled):
 
 @this_yaml.register_class
 @enum.unique
-class FM2Constraint(str, Enameled):
+class PCMRestriction(str, Enameled):
     """Firm 2 margins - derivation methods."""
 
     IID = "i.i.d"
@@ -307,9 +307,7 @@ class FM2Constraint(str, Enameled):
     SYM = "symmetric"
 
 
-def _pcm_dp_conv(
-    _v: ArrayFloat | Sequence[float] | None, _i: PCMSpec
-) -> ArrayFloat:
+def _pcm_dp_conv(_v: ArrayFloat | Sequence[float] | None, _i: PCMSpec) -> ArrayFloat:
     if _v is None or len(_v) == 0 or np.array_equal(_v, DEFAULT_DIST_PARMS):
         if _i.dist_type == PCMDistribution.EMPR:
             return margin_data_builder()[0]
@@ -320,10 +318,10 @@ def _pcm_dp_conv(
                 return DEFAULT_BETA_BND_DIST_PARMS
             case _:
                 return DEFAULT_DIST_PARMS
-    elif (_i.dist_type == PCMDistribution.EMPR and not isinstance(_v, np.ndarray)):
-            raise ValueError(
-                "Invalid specification; use output of mergeron.core.empriical_margin_distribution.margin_data_builider()."
-            )
+    elif _i.dist_type == PCMDistribution.EMPR and not isinstance(_v, np.ndarray):
+        raise ValueError(
+            "Invalid specification; use ..core.empriical_margin_distribution.margin_data_builider()[0]."
+        )
     elif isinstance(_v, Sequence | np.ndarray):
         return np.asarray(_v, float)
     else:
@@ -332,8 +330,6 @@ def _pcm_dp_conv(
             "sequence of Numpy arrays, or Numpy ndarray."
         )
 
-    return _v
-
 
 @frozen
 class PCMSpec:
@@ -373,9 +369,7 @@ class PCMSpec:
     """
 
     @dist_parms.default
-    def __dpwd(
-        _i: PCMSpec,
-    ) -> ArrayFloat:
+    def __dpwd(_i: PCMSpec) -> ArrayFloat:
         return _pcm_dp_conv(None, _i)
 
     @dist_parms.validator
@@ -403,16 +397,22 @@ class PCMSpec:
                     f'for PCM with distribution, "{_i.dist_type}" is incorrect.'
                 )
 
-        elif (
-            _i.dist_type == PCMDistribution.EMPR
-            and not isinstance(_v, np.ndarray)
-        ):
+        elif _i.dist_type == PCMDistribution.EMPR and not isinstance(_v, np.ndarray):
             raise ValueError(
                 "Empirical distribution requires deserialzed margin data from Prof. Damodaran, NYU"
             )
 
-    firm2_pcm_constraint: FM2Constraint = field(kw_only=True, default=FM2Constraint.IID)
-    """See :class:`FM2Constraint`"""
+    pcm_restriction: PCMRestriction = field(kw_only=True, default=PCMRestriction.IID)
+    """See :class:`PCMRestriction`"""
+
+    @pcm_restriction.validator
+    def __prv(_i: PCMSpec, _a: Attribute[PCMRestriction], _v: PCMRestriction) -> None:
+        if _v == PCMRestriction.MNL and _i.dist_type == PCMDistribution.EMPR:
+            print(
+                "NOTE: Estimated Firm 2 parameters will not be consistent with "
+                "the empirical distribution of margins in the source data. For "
+                "consistency, respecify pcm_spec.pcm_restriction = PCMRestriction.IID."
+            )
 
 
 @this_yaml.register_class

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/gen/data_generation.py

@@ -26,10 +26,10 @@ from .. import (  # noqa: TID252  # noqa
 from ..core import guidelines_boundaries as gbl  # noqa: TID252
 from ..core.guidelines_boundaries import HMGThresholds  # noqa: TID252
 from . import (
-    FM2Constraint,
     INVResolution,  # noqa: F401
     MarketSampleData,
     PCMDistribution,
+    PCMRestriction,
     PCMSpec,
     PriceSpec,
     SeedSequenceData,
@@ -44,7 +44,7 @@ from .data_generation_functions import (
     gen_margin_price_data,
     gen_share_data,
 )
-from .upp_tests import compute_upp_test_counts  # type: ignore # has pytypes marker ...
+from .upp_tests import compute_upp_test_counts
 
 __version__ = VERSION
 
@@ -100,10 +100,10 @@ class MarketSample:
     def _psv(self, _a: Attribute[PCMSpec], _v: PCMSpec, /) -> None:
         if (
             self.share_spec.recapture_form == RECForm.FIXED
-            and _v.firm2_pcm_constraint == FM2Constraint.MNL
+            and _v.pcm_restriction == PCMRestriction.MNL
         ):
             raise ValueError(
-                f'Specification of "PCMSpec.firm2_pcm_constraint", as {FM2Constraint.MNL!r} '
+                f'Specification of "PCMSpec.pcm_restriction", as {PCMRestriction.MNL!r} '
                 f'requires that "ShareSpec.recapture_form" be {RECForm.INOUT!r} '
                 f"or {RECForm.OUTIN!r}, not {RECForm.FIXED!r} as presently specified"
             )
@@ -179,7 +179,7 @@ class MarketSample:
         shr_sample_size = sample_size * self.hsr_filing_test_type
         shr_sample_size *= (
             SSZConstant.MNL_DEP
-            if self.pcm_spec.firm2_pcm_constraint == FM2Constraint.MNL
+            if self.pcm_spec.pcm_restriction == PCMRestriction.MNL
             else 1
         )
         shr_sample_size = int(shr_sample_size)
@@ -379,28 +379,33 @@ class MarketSample:
 
         sim_enf_cnts_kwargs: SamplingFunctionKWArgs = SamplingFunctionKWArgs({
             "sample_size": subsample_sz,
-            "nthreads": self.nthreads,
+            "nthreads": thread_count,
         })
 
-        res_list = Parallel(n_jobs=thread_count, prefer="threads")(
+        res_list = Parallel(n_jobs=min(thread_count, iter_count), prefer="threads")(
             delayed(self.__sim_enf_cnts)(
                 _enf_parm_vec,
                 _sim_test_regime,
                 **sim_enf_cnts_kwargs,
                 seed_data=_rng_seed_data_ch,
             )
-            for _iter_id, _rng_seed_data_ch in enumerate(rng_seed_data)
+            for _rng_seed_data_ch in rng_seed_data
         )
 
         res_list_stacks = UPPTestsCounts(*[
             np.stack([getattr(_j, _k) for _j in res_list])
             for _k in ("by_firm_count", "by_delta", "by_conczone")
         ])
+
         upp_test_results = UPPTestsCounts(*[
-            np.hstack((
-                (_gv := getattr(res_list_stacks, _g.name))[0, :, :_h],
-                np.einsum("ijk->jk", _gv[:, :, _h:], dtype=np.int64),
-            ))
+            (
+                np.array([], int)
+                if not (_gv := getattr(res_list_stacks, _g.name)).any()
+                else np.hstack((
+                    _gv[0, :, :_h],
+                    np.einsum("ijk->jk", _gv[:, :, _h:], dtype=int),
+                ))
+            )
             for _g, _h in zip(res_list_stacks.__attrs_attrs__, [1, 1, 3], strict=True)
         ])
         del res_list, res_list_stacks

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/gen/data_generation_functions.py

@@ -28,9 +28,9 @@ from ..core.pseudorandom_numbers import (  # noqa: TID252
 from . import (
     DEFAULT_BETA_BND_DIST_PARMS,
     DEFAULT_FCOUNT_WTS,
-    FM2Constraint,
     MarginDataSample,
     PCMDistribution,
+    PCMRestriction,
     PCMSpec,
     PriceDataSample,
     PriceSpec,
@@ -549,11 +549,11 @@ def gen_margin_price_data(  # noqa: PLR0914
             # del _price_array_gen
         case PriceSpec.CSY:
             # TODO:
-            # evolve FM2Constraint (save running MNL test twice); evolve copy of _mkt_sample_spec=1q
+            # evolve PCMRestriction (save running MNL test twice); evolve copy of _mkt_sample_spec=1q
             # generate the margin data
             # generate price and margin data
             frmshr_array_plus = np.hstack((_frmshr_array, _nth_firm_share))
-            pcm_spec_here = evolve(_pcm_spec, firm2_pcm_constraint=FM2Constraint.IID)
+            pcm_spec_here = evolve(_pcm_spec, pcm_restriction=PCMRestriction.IID)
             margin_data = _gen_margin_data(
                 frmshr_array_plus,
                 np.ones_like(frmshr_array_plus, np.float64),
@@ -570,7 +570,7 @@ def gen_margin_price_data(  # noqa: PLR0914
             price_array_here = 1 / (1 - pcm_array)
             price_array = price_array_here[:, :2]
             nth_firm_price = price_array_here[:, [-1]]
-            if _pcm_spec.firm2_pcm_constraint == FM2Constraint.MNL:
+            if _pcm_spec.pcm_restriction == PCMRestriction.MNL:
                 # Generate i.i.d. PCMs then take PCM0 and construct PCM1
                 # Regenerate MNL test
                 purchase_prob_array = _aggregate_purchase_prob * _frmshr_array
@@ -665,14 +665,13 @@ def _gen_margin_data(
     _nthreads: int,
     /,
 ) -> MarginDataSample:
-    dist_type_pcm, dist_parms_pcm, dist_firm2_pcm = (
-        getattr(_pcm_spec, _f)
-        for _f in ("dist_type", "dist_parms", "firm2_pcm_constraint")
+    dist_type_pcm, dist_parms_pcm, pcm_restriction_ = (
+        getattr(_pcm_spec, _f) for _f in ("dist_type", "dist_parms", "pcm_restriction")
     )
 
     pcm_array = (
         np.empty_like(_frmshr_array[:, :1])
-        if _pcm_spec.firm2_pcm_constraint == FM2Constraint.SYM
+        if _pcm_spec.pcm_restriction == PCMRestriction.SYM
         else np.empty_like(_frmshr_array)
     )
 
@@ -680,7 +679,10 @@ def _gen_margin_data(
     beta_min, beta_max = [0.0] * 2  # placeholder
     if dist_type_pcm == PCMDistribution.EMPR:
         pcm_array = margin_data_resampler(
-            dist_parms_pcm, sample_size=pcm_array.shape, seed_sequence=_pcm_rng_seed_seq
+            dist_parms_pcm,
+            sample_size=pcm_array.shape,
+            seed_sequence=_pcm_rng_seed_seq,
+            nthreads=_nthreads,
         )
     else:
         dist_type_: Literal["Beta", "Uniform"]
@@ -721,16 +723,10 @@ def _gen_margin_data(
         pcm_array = (beta_max - beta_min) * pcm_array + beta_min
         del beta_min, beta_max
 
-    if dist_firm2_pcm == FM2Constraint.SYM:
+    if pcm_restriction_ == PCMRestriction.SYM:
         pcm_array = np.hstack((pcm_array,) * _frmshr_array.shape[1])
-    if dist_firm2_pcm == FM2Constraint.MNL:
+    if pcm_restriction_ == PCMRestriction.MNL:
         # Impose FOCs from profit-maximization with MNL demand
-        if dist_type_pcm == PCMDistribution.EMPR:
-            print(
-                "NOTE: Estimated Firm 2 parameters will not be consistent with "
-                "the empirical distribution of margins in the source data. For "
-                "consistency, respecify pcm_spec.firm2_pcm_constraint = FM2Constraint.IID."
-            )
         purchase_prob_array = _aggregate_purchase_prob * _frmshr_array
 
         pcm_array[:, [1]] = np.divide(

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/gen/enforcement_stats.py

@@ -335,10 +335,13 @@ def enf_cnts_byconczone(_cnts_array: ArrayBIGINT, /) -> ArrayBIGINT:
         # Logical-and of multiple vectors:
         hhi_zone_test = (
             1
-            * np.stack([
-                cnts_byhhipostanddelta[:, _idx] == _val
-                for _idx, _val in enumerate(zone_val)
-            ], axis=1)
+            * np.stack(
+                [
+                    cnts_byhhipostanddelta[:, _idx] == _val
+                    for _idx, _val in enumerate(zone_val)
+                ],
+                axis=1,
+            )
         ).prod(axis=1) == 1
 
         cnts_byconczone = np.vstack((

{mergeron-2025.739319.1 → mergeron-2025.739319.3}/src/mergeron/gen/upp_tests.py

@@ -142,10 +142,6 @@ def compute_upp_test_counts(  # noqa: PLR0914
             hhi_delta_ranged,
             np.ones_like(hhi_delta_ranged),
             upp_test_arrays,
-            # *[
-            #     1 * getattr(upp_test_arrays, _a.name)
-            #     for _a in upp_test_arrays.__attrs_attrs__
-            # ],
         ),
         dtype=int,
     )
@@ -194,11 +190,23 @@ def _compute_test_array_seq(
                 np.sqrt(np.einsum("ij,ij,ij->i", _wt_array[:, ::-1], _g, _g))[:, None]
                 for _g in _test_measure_seq
             )
+        case UPPAggrSelector.CPG:
+            test_array_seq = (
+                np.expm1(
+                    np.einsum("ij,ij->i", _wt_array[:, ::-1], np.log1p(_g))[:, None]
+                )
+                for _g in _test_measure_seq
+            )
         case UPPAggrSelector.DIS:
             test_array_seq = (
                 np.sqrt(1 / 2 * np.einsum("ij,ij->i", _g, _g))[:, None]
                 for _g in _test_measure_seq
             )
+        case UPPAggrSelector.GMN:
+            test_array_seq = (
+                np.expm1(np.einsum("ij->i", np.log1p(_g))[:, None] / _g.shape[1])
+                for _g in _test_measure_seq
+            )
         case UPPAggrSelector.MAX:
             test_array_seq = (_g.max(axis=1, keepdims=True) for _g in _test_measure_seq)
         case UPPAggrSelector.MIN:
@@ -210,7 +218,12 @@ def _compute_test_array_seq(
             )
         case UPPAggrSelector.OSD:
             test_array_seq = (
-                np.sqrt(np.einsum("ij,ij,ij->i", _wt_array, _g, _g))[:, None]
+                np.sqrt(np.einsum("ij,ij,ij->i", _wt_array, _g, _g)[:, None])
+                for _g in _test_measure_seq
+            )
+        case UPPAggrSelector.OSG:
+            test_array_seq = (
+                np.expm1(np.einsum("ij,ij->i", _wt_array, np.log1p(_g))[:, None])
                 for _g in _test_measure_seq
             )
         case _: