Replication materials for "Monetary Policy and Oil-Shock Pass-Through: Evidence from Morocco, 2020-2025"

This dataset accompanies the article "Monetary Policy and Oil-Shock Pass-Through: Evidence from Morocco, 2020–2025" by Otmane Lahrizi and Meriem Demdoumi (ENCG, Université Ibn Tofaïl, Morocco), submitted to Energy Economics. RESEARCH QUESTION ; The dataset was assembled to test whether Bank Al-Maghrib's 2022–2024 monetary tightening cycle was associated with a smaller pass-through of global oil-price shocks to Moroccan inflation than the prior loose regime. Morocco is an analytically clean setting: it imports the entirety of its refined petroleum since the 2015 SAMIR refinery closure, fuel retail prices follow a bi-monthly indexation formula tied to international quotations, and BAM ran a complete tightening cycle (1.50% to 3.00% to 2.25%) within the sample. WHAT THE DATA CONTAINS ; 66 monthly observations from January 2020 to June 2025: Moroccan year-on-year CPI inflation (peak 10.1% in February 2023, highest in three decades); Brent crude price (USD 18 in April 2020 to USD 123 in June 2022); BAM policy rate; MAD/USD exchange rate; Eurozone HICP as global control; CPI sub-components CP04 (energy and fuels) and CP07 (transport); and a binary tight-regime dummy = 1 for September 2022 to June 2024. SOURCES ; Morocco CPI from Haut-Commissariat au Plan via IMF IFS; Brent from FRED (DCOILBRENTEU); BAM rate from published Monetary Policy Committee communiqués; MAD/USD from FRED (DEXMAUS); Eurozone HICP from Eurostat; energy import bill from Office des Changes (CVS-CJO series). Daily Brent prices were converted to monthly arithmetic averages before log-differencing. NOTABLE FINDINGS ; Using the included Python code, the state-dependent Local Projection interaction coefficient gamma_h is negative at 12 of 13 horizons, reaching −10.2 percentage points at horizon 8 months. A joint test rejects the null of no interaction (sign-test p < 0.001; Wald chi-squared(13) = 25.84, p = 0.018). The finding survives five robustness exercises: continuous-rate specification, wild bootstrap, Brent realized volatility, a 9-cell lag-sensitivity grid, and exclusion of the COVID acute period. HOW TO INTERPRET ; Inflation, BAM rate, and HICP are in percentage points. Δlog Brent and Δlog MAD/USD are monthly log-differences (0.01 = 1% monthly change). The tight dummy is binary. CPI sub-components follow HCP COICOP classification. HOW TO USE ; The script run_lp_estimation.py reproduces every quantitative result in the paper. make_figures.py regenerates the three main figures. Run time under 30 seconds. Requires Python 3.10+, numpy, pandas, openpyxl, scipy, matplotlib. LIMITATIONS ; The tight-regime sub-sample is small (22 months), limiting horizon-by-horizon precision; joint inference is the appropriate reading. The reduced-form Brent shock conflates supply, demand, and uncertainty components. Substantial fiscal compensation (Caisse de Compensation expenditure rising from ~12.5 to 42 billion dirhams in 2021–2022) operated concurrently with the monetary tightening.本数据集为发表于《能源经济学》（Energy Economics）的论文《货币政策与石油冲击传导：2020-2025年摩洛哥证据》（作者：Otmane Lahrizi、Meriem Demdoumi，隶属摩洛哥伊本·托法伊大学国家商业与管理学院（ENCG））的配套数据。 ## 研究问题 本数据集旨在验证摩洛哥中央银行（Bank Al-Maghrib, BAM）2022-2024年的货币政策紧缩周期，是否较此前的宽松政策阶段，更能削弱全球石油价格冲击向摩洛哥通胀的传导效应。摩洛哥是一个分析场景纯净的研究样本：自2015年萨米尔（SAMIR）炼油厂关停以来，该国完全依赖进口成品油；燃油零售价格采用与国际报价挂钩的双月指数化定价公式；且样本期内BAM完整实施了一轮货币政策紧缩周期（政策利率从1.50%上调至3.00%，后下调至2.25%）。 ## 数据集内容 包含2020年1月至2025年6月共66个月度观测值，具体包括：摩洛哥同比居民消费价格指数（CPI）通胀率（2023年2月达峰值10.1%，为三十年来最高值）、布伦特原油（Brent crude）价格（2020年4月为18美元，2022年6月升至123美元）、BAM政策利率、摩洛哥迪拉姆（MAD）兑美元汇率、作为全球控制变量的欧元区调和消费者物价指数（HICP）、CPI细分项CP04（能源与燃料）及CP07（交通运输），以及二元紧缩阶段虚拟变量（2022年9月至2024年6月取值为1）。 ## 数据来源 摩洛哥CPI数据来自摩洛哥计划高级专员公署（Haut-Commissariat au Plan），通过国际货币基金组织国际金融统计数据库（IMF IFS）获取；布伦特原油价格来自圣路易斯联邦储备银行经济数据库（FRED，代码DCOILBRENTEU）；BAM政策利率来自公开的货币政策委员会公报；MAD/USD汇率来自FRED（代码DEXMAUS）；欧元区HICP数据来自欧盟统计局（Eurostat）；能源进口账单数据来自摩洛哥外汇办公室（Office des Changes，序列CVS-CJO）。每日布伦特原油价格在进行对数差分（log-differencing）前，已转换为月度算术平均值。 ## 主要研究发现 借助附带的Python代码，状态依赖型局部投影（Local Projection）交互项系数γ_h在13个预测期中有12个为负值，在第8个月预测期达到-10.2个百分点。联合检验拒绝了“无交互效应”的原假设（符号检验p < 0.001；沃尔德卡方检验χ²(13) = 25.84，p = 0.018）。该研究结论通过五项稳健性检验：连续利率设定、野生Bootstrap（Wild Bootstrap）、布伦特原油已实现波动率、9格滞后项敏感性网格分析，以及剔除新冠疫情高峰期样本。 ## 变量说明 通胀率、BAM政策利率及HICP均以百分点为单位。Δlog布伦特原油价格与Δlog MAD/USD汇率为月度对数差分序列（0.01代表月度变化1%）。紧缩虚拟变量为二元变量。CPI细分项采用协调消费者支出分类（COICOP）标准。 ## 使用方法 脚本run_lp_estimation.py可复现论文中的所有定量结果；make_figures.py可重新生成三张核心图表。运行时长不超过30秒，需依赖Python 3.10及以上版本，以及numpy、pandas、openpyxl、scipy、matplotlib库。 ## 局限性说明 紧缩阶段子样本量较小（仅22个月），限制了单预测期的估计精度，因此联合推断是更合适的解读方式。简约形式的布伦特原油冲击包含了供给、需求与不确定性等多重成分。货币政策紧缩同期，摩洛哥实施了大规模财政补贴（补偿基金（Caisse de Compensation）支出从2021-2022年的约125亿迪拉姆增至420亿迪拉姆）。