Inspecteur des Finances Publiques (DGFiP) common mistakes
Inspecteur des Finances Publiques (DGFiP) · Understand the traps and how to avoid them.
In a dissertation on the LOLF, the candidate opens with three paragraphs of general history of the French budget since 1959 and only reaches the question (e.g. 'la LOLF a-t-elle amélioré la performance de la dépense publique ?') after two pages, leaving no room to develop the actual problematique.
Candidates reproduce the chronological narrative found in handbooks because it is easy to memorize and feels 'safe'; they mistake coverage of context for construction of a problem. The exam rewards a problématique announced early and a demonstration, not a history lesson.
What to do instead
Adopt a fixed skeleton: (1) accroche of 3–5 lines tied to current events, (2) definition of terms, (3) problématique formulated as a tension between two legitimate propositions, (4) announcement of a two- or three-part plan. Cap the introduction at one page and rehearse it on a timer (20 minutes maximum) for ten different subjects.
In the note de synthèse, the candidate paraphrases documents one after another ('le document 1 indique que… le document 2 précise que…') instead of reorganising them around thematic axes, and produces a 6-page note for a dossier requiring 4 pages.
Under time pressure, sequential reading becomes sequential writing; candidates fear losing information and therefore copy structure from the dossier. They also fail to allocate reading time, so the drafting phase is compressed and no reorganisation happens.
What to do instead
Use a three-pass method: 30 min skim to label each document with 3 keywords, 30 min thematic grouping into 2–4 axes with a table of cross-references, 90 min drafting. Never cite a document number in the body; attribute ideas to their author (e.g. 'la direction du budget souligne…'). Practise on a strict page count with a timer and count lines per page in advance.
On a TVA question, the candidate recites the general rule (taxable base, rates, right to deduct) but omits the specific exception asked about - for instance the treatment of a livraison à soi-même, the co-contracting regime for works on immovable property, or the territoriality rule for B2C services - and answers with the general regime.
Revision is organised by headings of the code rather than by operational situation, so candidates know the architecture but not the exceptions. The concours tests the exception because that is what a real dossier requires.
What to do instead
Build fiches by 'cas pratique' rather than by article: for each theme (TVA immobilière, autoliquidation, territorialité, franchises), write a one-page sheet with the qualifying facts, the applicable rule, the exception and a 5-line worked example with figures. Redo past subjects by writing the full calculation, not by reading the corrigé.
On an impôt sur les sociétés or impôt sur le revenu exercise, the candidate produces the correct reasoning but forgets a step: failing to reintegrate a charge non déductible, forgetting the plafonnement of the déficit reportable, omitting the contribution sociale sur l'IS, or applying the reduced rate of 15 % without checking the conditions of chiffre d'affaires and capital detention.
The candidate reasons in prose and keeps figures in the head instead of on paper; the arithmetic is done once and never cross-checked. Reduced rates and ceilings change frequently, and candidates rely on memory of an older version of the text.
What to do instead
Impose a written template for every calculation: résultat comptable, then a two-column table (réintégrations / déductions) with the article reference for each line, then the tax computation with the rate and its conditions explicitly stated. End with a control total and a one-line sanity check ('le taux effectif obtenu est-il plausible ?'). Verify current rates and thresholds against the legifrance text of the year of the exam.
In droit constitutionnel or droit administratif, the candidate answers a question on the hiérarchie des normes or on the recours pour excès de pouvoir by citing a principle without its source and without the jurisprudential evolution - for example invoking the 'bloc de constitutionnalité' without mentioning the Charte de l'environnement, or treating the jurisprudence Danthony as if it had not modified the regime of vice de forme.
Candidates learn conclusions without the date and the factual matrix, so they cannot adapt when the subject shifts the facts slightly. Jurisprudence is memorised as a slogan rather than as a rule with conditions.
What to do instead
For each leading decision, keep a four-line fiche: facts, question of law, solution, scope and limits (and any later reversal). In the copy, always name the norm, its rank, the judge competent to enforce it, and the effect of the remedy. Test yourself by re-stating each decision in a hypothetical variation where the facts change.
In the language paper (English), the candidate translates or writes in a register that is grammatically acceptable but institutionally wrong: rendering 'inspecteur des finances publiques' as 'tax inspector', 'direction départementale' as 'departmental direction', or 'contrôle fiscal' as 'fiscal control', and using French calques such as 'actual' for 'current' or 'eventually' for 'éventuellement'.
Vocabulary is learned from bilingual lists rather than from institutional documents in the target language; candidates underestimate how much the language mark depends on collocations and register rather than on grammar alone.
What to do instead
Build a glossary of 150 institutional terms from authentic English-language sources (OECD, IMF, EU Commission documents on tax administration) and use them in full sentences. For the written paper, practise summarising a French economic article in English in 200 words, then compare your collocations with the source vocabulary list; keep a personal error log of false friends and review it weekly.
On an économie publique question, the candidate describes an institution or a mechanism (e.g. the ONDAM, the règle d'or, the stabilisateurs automatiques) without any analytical apparatus, no elasticity, no multiplier, no incidence reasoning, and no reference to empirical orders of magnitude.
Economics is revised as institutional description because that is how the news covers it; candidates are uncomfortable with the formal apparatus and therefore avoid it, producing a copy that a jury reads as journalistic rather than analytical.
What to do instead
For each policy theme, prepare a two-column fiche: mechanisms and orders of magnitude on one side (elasticity of the tax base, multiplier range, share of GDP, cost per point of ONDAM), analytical tools on the other (incidence, deadweight loss, moral hazard, crowding out). In the copy, always move from mechanism to effect to empirical magnitude to limitation.
In the management public question, the candidate answers with generalities about 'la modernisation de l'action publique' and 'la culture du résultat' without naming a single instrument: no contrat d'objectifs et de moyens, no dialogue de gestion, no indicateur de performance, no distinction between contrôle de gestion et contrôle budgétaire, no reference to the LOLF's three types of performance indicators (efficacité, efficience, qualité de service).
The domain feels like common sense, so candidates prepare it last and rely on general knowledge. They also confuse the vocabulary of private management with the specific instruments of French public management.
What to do instead
Memorise a toolbox of ten named instruments with one concrete example each (e.g. a BOP, a responsable de programme, a compte rendu au Parlement, a revue de dépenses). For each, be able to state who decides, who executes, and how performance is measured. Anchor each answer in a real documented case (a Cour des comptes report, a modernisation programme) rather than an abstract claim.
In culture générale or economic current affairs, the candidate quotes a figure, a date or an institution that was accurate two years earlier - an obsolete deficit ratio, a superseded EU fiscal rule, a reform that was amended, or a minister who has since changed portfolio - and presents it without a date reference.
Revision notes are recycled from previous years and news is absorbed passively; candidates do not systematically refresh figures and do not date their statements, which turns a small inaccuracy into a credibility problem before the jury.
What to do instead
Maintain a one-page 'chiffres et dates' sheet updated monthly from primary sources (INSEE, Banque de France, Cour des comptes, Commission européenne), always written with the date of the figure ('en 2024, le déficit public s'établissait à…'). In the copy, date every quantitative claim and prefer a rounded, sourced figure to a precise, unsourced one.
In procédures fiscales, the candidate describes the contrôle sur pièces and the vérification de comptabilité without the guarantees attached to each: the charte des droits et obligations du contribuable vérifié, the débat oral et contradictoire, the durée maximale sur place, the obligation to notify the nature of the control, and the distinction between the proposition de rectification and the réponse aux observations du contribuable.
The procedure is learned as a list of steps rather than as a balance between the administration's powers and the taxpayer's guarantees; candidates therefore omit exactly what the exam tests, namely the conditions of regularity of the procedure.
What to do instead
Draw a timeline for each procedure with two lanes, 'pouvoirs de l'administration' and 'garanties du contribuable', and place each document at its date with its legal basis. Then practise the reflex question: 'à quel stade une irrégularité rend-elle la procédure irrégulière, et quelle en est la conséquence sur le redressement ?'
Across the whole written paper, the candidate manages time badly: spending 90 minutes on a beautiful introduction, drafting without a plan on the third part, and handing in a copy with an unfinished conclusion, no transition between parts, and illegible corrections.
Perfectionism on the opening combined with no rehearsal under exam conditions; candidates discover their writing speed only on the day. The jury penalises an unfinished copy far more heavily than a plain but complete one.
What to do instead
Simulate the full paper at least four times before the exam with a written time budget: 20 min plan, 15 min introduction, 90 min development, 10 min conclusion, 15 min re-reading. Enforce a hard stop on each section, leave two blank lines per transition to fill at the end, and rehearse a standard conclusion formula so it can be written in five minutes.