a short history of the 2% target.
inflation targeting was invented, almost casually, in new zealand: a 1990 policy framework that gave the central bank one number to hit and independence in hitting it. it worked — inflation fell, expectations anchored — and the idea spread through the decade: canada, the uk, sweden, australia.
the fed came late. for decades its inflation objective was implicit, deliberately vague. only in january 2012 did the committee publish a formal 2% target — against PCE inflation, a detail that matters more than most readers know (we'll cover the gauge itself in two weeks).
why 2 and not 0? two honest reasons: measured inflation overstates true inflation slightly (quality improvements are hard to price), and a small positive buffer keeps rates safely away from the zero bound, preserving room to cut in recessions. zero would be purity at the cost of ammunition.
the number is a design choice, not a law of nature — and it has been argued about, publicly, by serious people, ever since it was written down. worth remembering every time a print is described as "above target." above a choice.
inflation targeting was invented, almost casually, in new zealand: a 1990 policy framework that gave the central bank one number to hit and independence in hitting it. it worked — inflation fell, expectations anchored — and the idea spread through the decade: canada, the uk, sweden, australia.
the fed came late. for decades its inflation objective was implicit, deliberately vague. only in january 2012 did the committee publish a formal 2% target — against PCE inflation, a detail that matters more than most readers know (we'll cover the gauge itself in two weeks).
why 2 and not 0? two honest reasons: measured inflation overstates true inflation slightly (quality improvements are hard to price), and a small positive buffer keeps rates safely away from the zero bound, preserving room to cut in recessions. zero would be purity at the cost of ammunition.
the number is a design choice, not a law of nature — and it has been argued about, publicly, by serious people, ever since it was written down. worth remembering every time a print is described as "above target." above a choice.
the week ahead — august 17 to 21. communication week.
wednesday, 18:00 UTC (21:00 мск): minutes of the july fed meeting. no decision — the record of one already made, and the market reads it for a single thing: the shape of the disagreement inside the committee. that thread has been running since may's four dissents.
friday: flash PMIs, the first survey read on august activity. and late august is, traditionally, podium season — the time of year when central bank speech-making concentrates.
so this week's briefs are about the machinery of communication itself:
monday — how the minutes are actually produced, and why "edited" is precisely what makes them useful.
tuesday — forward guidance: a short history of promises.
wednesday — the dot plot: nineteen opinions wearing one chart.
thursday — blackout rules: why silence is scheduled.
data tells you where the economy was. communication tells you where the committee is. the second one prices currencies.
wednesday, 18:00 UTC (21:00 мск): minutes of the july fed meeting. no decision — the record of one already made, and the market reads it for a single thing: the shape of the disagreement inside the committee. that thread has been running since may's four dissents.
friday: flash PMIs, the first survey read on august activity. and late august is, traditionally, podium season — the time of year when central bank speech-making concentrates.
so this week's briefs are about the machinery of communication itself:
monday — how the minutes are actually produced, and why "edited" is precisely what makes them useful.
tuesday — forward guidance: a short history of promises.
wednesday — the dot plot: nineteen opinions wearing one chart.
thursday — blackout rules: why silence is scheduled.
data tells you where the economy was. communication tells you where the committee is. the second one prices currencies.
how the fed's minutes are made.
they are not a transcript. verbatim transcripts of FOMC meetings exist — and are sealed for five years. the minutes are a different object: drafted by staff in the days after the meeting, circulated to participants, amended, and formally approved before release three weeks later.
that edit is the point. every sentence survived review by the very people it describes — which makes the minutes a deliberate communication instrument, not a leak. when the minutes emphasize a disagreement, the committee chose to show it to you.
the practiced way to read them is to track the counting words. "a couple," "a few," "several," "many," "most" — this is the committee's own vocabulary for the distribution of views, and shifts in it across meetings are measurable signal. "several participants" worrying about something in june becoming "many" in july is a repricing in plain text.
wednesday night, july's minutes land at 18:00 UTC. the question they answer is not what the fed did — you know that. it's how close the vote was to going the other way.
they are not a transcript. verbatim transcripts of FOMC meetings exist — and are sealed for five years. the minutes are a different object: drafted by staff in the days after the meeting, circulated to participants, amended, and formally approved before release three weeks later.
that edit is the point. every sentence survived review by the very people it describes — which makes the minutes a deliberate communication instrument, not a leak. when the minutes emphasize a disagreement, the committee chose to show it to you.
the practiced way to read them is to track the counting words. "a couple," "a few," "several," "many," "most" — this is the committee's own vocabulary for the distribution of views, and shifts in it across meetings are measurable signal. "several participants" worrying about something in june becoming "many" in july is a repricing in plain text.
wednesday night, july's minutes land at 18:00 UTC. the question they answer is not what the fed did — you know that. it's how close the vote was to going the other way.
forward guidance: a short history of promises.
central banks discovered gradually that their most powerful tool is a sentence about the future. the fed's 2003 "considerable period" — a phrase implying rates would stay low for a long time — moved the entire yield curve without moving the policy rate at all.
the zero-bound years industrialized the technique. when rates hit zero in 2008 and couldn't be cut further, promising to keep them at zero became the policy: calendar guidance ("through mid-2013"), then threshold guidance (not until unemployment falls below a level). each formulation was a stronger promise — and a tighter constraint.
that's the tension that defines the tool. guidance only works if the market believes it, and the market only believes committees that keep their promises — so every promise made narrows tomorrow's freedom. break one, and the next decade of your sentences trades at a discount.
fx relevance: guidance moves whole curves, and currencies price curve differentials. one credible sentence about next year outweighs this quarter's data. which is why the market spends more effort parsing intentions than measuring history — and why this week's releases are words, not numbers.
central banks discovered gradually that their most powerful tool is a sentence about the future. the fed's 2003 "considerable period" — a phrase implying rates would stay low for a long time — moved the entire yield curve without moving the policy rate at all.
the zero-bound years industrialized the technique. when rates hit zero in 2008 and couldn't be cut further, promising to keep them at zero became the policy: calendar guidance ("through mid-2013"), then threshold guidance (not until unemployment falls below a level). each formulation was a stronger promise — and a tighter constraint.
that's the tension that defines the tool. guidance only works if the market believes it, and the market only believes committees that keep their promises — so every promise made narrows tomorrow's freedom. break one, and the next decade of your sentences trades at a discount.
fx relevance: guidance moves whole curves, and currencies price curve differentials. one credible sentence about next year outweighs this quarter's data. which is why the market spends more effort parsing intentions than measuring history — and why this week's releases are words, not numbers.
the dot plot, read correctly.
four times a year, each of the nineteen FOMC participants — voters and non-voters alike — marks where they individually think the policy rate should be at year-ends ahead. the dots are anonymous, and the market compresses all nineteen into one number: the median.
three standard misreadings, in ascending order of cost:
treating the median as a commitment. it's a snapshot of opinions that reprice with every data cycle — dots age badly, and the committee itself insists they aren't a plan.
ignoring the distribution. a median with dots tightly clustered around it and the same median with a deep hawk-dove split are different objects. the spread is the information; may's four dissents live in that spread.
forgetting who plots. all nineteen participants plot; only twelve vote. the median can sit meaningfully away from where the actual voting majority stands.
tonight at 18:00 UTC, july's minutes land — no dots attached, but the same skill applies: you're reading the distribution of a committee, not the intention of a person.
four times a year, each of the nineteen FOMC participants — voters and non-voters alike — marks where they individually think the policy rate should be at year-ends ahead. the dots are anonymous, and the market compresses all nineteen into one number: the median.
three standard misreadings, in ascending order of cost:
treating the median as a commitment. it's a snapshot of opinions that reprice with every data cycle — dots age badly, and the committee itself insists they aren't a plan.
ignoring the distribution. a median with dots tightly clustered around it and the same median with a deep hawk-dove split are different objects. the spread is the information; may's four dissents live in that spread.
forgetting who plots. all nineteen participants plot; only twelve vote. the median can sit meaningfully away from where the actual voting majority stands.
tonight at 18:00 UTC, july's minutes land — no dots attached, but the same skill applies: you're reading the distribution of a committee, not the intention of a person.
blackout rules, and the shape of the speaking calendar.
before every FOMC meeting, a communications blackout takes effect — roughly from the second saturday preceding the meeting until the thursday after. no speeches, no interviews, no guidance from any committee participant.
the rule exists for a clean reason: in the final days before a decision, any offhand sentence from a voter would be traded as a leak. the blackout protects the committee from itself.
its side effect structures the whole market calendar. speech-driven volatility can't happen in the quiet window — so it concentrates in the weeks between blackouts, when officials fan out to explain the last decision and shade expectations for the next one. desks literally maintain speaker calendars, weighted by three things: is the speaker a current voter, how far are we from the next blackout, and is the venue one where policy signals are traditionally sent.
which is the note to end communication week on: even central bank speech is scheduled, structured, and ranked. tomorrow — the podium that outranks them all, and the asia desk's practical wrap of the week.
before every FOMC meeting, a communications blackout takes effect — roughly from the second saturday preceding the meeting until the thursday after. no speeches, no interviews, no guidance from any committee participant.
the rule exists for a clean reason: in the final days before a decision, any offhand sentence from a voter would be traded as a leak. the blackout protects the committee from itself.
its side effect structures the whole market calendar. speech-driven volatility can't happen in the quiet window — so it concentrates in the weeks between blackouts, when officials fan out to explain the last decision and shade expectations for the next one. desks literally maintain speaker calendars, weighted by three things: is the speaker a current voter, how far are we from the next blackout, and is the venue one where policy signals are traditionally sent.
which is the note to end communication week on: even central bank speech is scheduled, structured, and ranked. tomorrow — the podium that outranks them all, and the asia desk's practical wrap of the week.
from the asia desk — communication week edition.
everything we covered this week has a practitioner's translation on Mike's channel, and this week the two ran unusually close together.
his minutes-morning filter is the operational version of monday's brief: one question — did rate pricing actually move? — and only if yes, find the sentence that moved it and decide whether the reaction survives the day. minutes that fade by lunch were confirmation, not news; that's the base case three times out of four.
and his piece on why a speech can outweigh a statistic is the desk version of what we called forward guidance: data is the past, policy is the future, and markets price the future. his filter for whose words matter — check the speaker's vote before you check their quote — is worth pinning somewhere visible.
thin-august rules still apply to everything, his side and ours: distrust drama until september's liquidity confirms it.
@equilon_mike.
everything we covered this week has a practitioner's translation on Mike's channel, and this week the two ran unusually close together.
his minutes-morning filter is the operational version of monday's brief: one question — did rate pricing actually move? — and only if yes, find the sentence that moved it and decide whether the reaction survives the day. minutes that fade by lunch were confirmation, not news; that's the base case three times out of four.
and his piece on why a speech can outweigh a statistic is the desk version of what we called forward guidance: data is the past, policy is the future, and markets price the future. his filter for whose words matter — check the speaker's vote before you check their quote — is worth pinning somewhere visible.
thin-august rules still apply to everything, his side and ours: distrust drama until september's liquidity confirms it.
@equilon_mike.
who actually provides fx liquidity — and where it goes in august.
every price you see is someone's standing offer to trade: bank dealing desks and, increasingly, non-bank electronic market makers, quoting both sides of every major pair continuously and earning the spread for bearing the risk of being there.
the size behind a quote is the variable nobody watches. a market maker showing a price for one unit in march may show the same price for half a unit in august — the quote looks identical on your screen; the depth behind it is seasonal. when a real order lands, it eats through more price levels to fill. that's the whole mechanism of thin-market violence: not more sellers, fewer catchers.
it also reprices risk transfer itself: spreads widen at the margins, fills degrade around events, and the same stop-loss that filled cleanly in spring fills worse now — through no fault of the trade.
nothing about august tape is mysterious. it's a staffing chart, expressed as volatility. plan position sizes — and conclusions — accordingly.
every price you see is someone's standing offer to trade: bank dealing desks and, increasingly, non-bank electronic market makers, quoting both sides of every major pair continuously and earning the spread for bearing the risk of being there.
the size behind a quote is the variable nobody watches. a market maker showing a price for one unit in march may show the same price for half a unit in august — the quote looks identical on your screen; the depth behind it is seasonal. when a real order lands, it eats through more price levels to fill. that's the whole mechanism of thin-market violence: not more sellers, fewer catchers.
it also reprices risk transfer itself: spreads widen at the margins, fills degrade around events, and the same stop-loss that filled cleanly in spring fills worse now — through no fault of the trade.
nothing about august tape is mysterious. it's a staffing chart, expressed as volatility. plan position sizes — and conclusions — accordingly.
the week ahead — august 24 to 28. the machinery of month-end.
friday, 12:30 UTC (15:30 мск): PCE — the inflation gauge the fed's 2% target is actually written against. the market watches CPI because it prints first; the committee steers by this one. most months they agree; the interesting nights are when they don't.
around it, the last week of august runs on mechanics, and this week's briefs cover exactly the mechanics in play:
monday — PCE versus CPI: two gauges, one target, and why the fed chose the quieter one.
tuesday — the 4pm london fix: the five-minute window that prices trillions.
wednesday — month-end rebalancing: the biggest trader nobody watches.
thursday — the data supply chain: the path from a statistical agency to your screen, and where retail actually sits in that queue.
september is loading behind all of it: NFP on the 4th, the ECB on the 10th, the fed on the 15th–16th, the bank of england and bank of japan on the 17th. quiet week, loud month ahead.
friday, 12:30 UTC (15:30 мск): PCE — the inflation gauge the fed's 2% target is actually written against. the market watches CPI because it prints first; the committee steers by this one. most months they agree; the interesting nights are when they don't.
around it, the last week of august runs on mechanics, and this week's briefs cover exactly the mechanics in play:
monday — PCE versus CPI: two gauges, one target, and why the fed chose the quieter one.
tuesday — the 4pm london fix: the five-minute window that prices trillions.
wednesday — month-end rebalancing: the biggest trader nobody watches.
thursday — the data supply chain: the path from a statistical agency to your screen, and where retail actually sits in that queue.
september is loading behind all of it: NFP on the 4th, the ECB on the 10th, the fed on the 15th–16th, the bank of england and bank of japan on the 17th. quiet week, loud month ahead.
Q&A: "why does the fed target PCE if everyone watches CPI?"
the two gauges differ by construction, not by accident.
scope: CPI counts what households pay out of pocket. PCE counts what is consumed on households' behalf — including, most importantly, employer- and government-paid healthcare. that alone shifts the weights materially: shelter is roughly a third of CPI and far less of PCE.
substitution: CPI prices a fixed basket, updated with a lag. PCE is chain-weighted — when beef gets expensive and people buy chicken, PCE follows the behavior. result: PCE typically runs a few tenths cooler.
the institutional history: the fed formally shifted its preferred measure to PCE in 2000, and wrote the 2012 target against it — precisely because it wanted the broader, substitution-aware gauge.
market consequence: CPI moves markets because it prints first and feeds the repricing; PCE usually confirms quietly two weeks later. the loud nights are the exceptions — when friday's PCE argues with mid-month's CPI, the rate market has to re-decide which inflation it believes. this friday is only interesting if it argues.
the two gauges differ by construction, not by accident.
scope: CPI counts what households pay out of pocket. PCE counts what is consumed on households' behalf — including, most importantly, employer- and government-paid healthcare. that alone shifts the weights materially: shelter is roughly a third of CPI and far less of PCE.
substitution: CPI prices a fixed basket, updated with a lag. PCE is chain-weighted — when beef gets expensive and people buy chicken, PCE follows the behavior. result: PCE typically runs a few tenths cooler.
the institutional history: the fed formally shifted its preferred measure to PCE in 2000, and wrote the 2012 target against it — precisely because it wanted the broader, substitution-aware gauge.
market consequence: CPI moves markets because it prints first and feeds the repricing; PCE usually confirms quietly two weeks later. the loud nights are the exceptions — when friday's PCE argues with mid-month's CPI, the rate market has to re-decide which inflation it believes. this friday is only interesting if it argues.
Q&A: "what actually happens at 4pm london — and why does price go strange there?"
at 16:00 london time every business day, the WM/Reuters fix is calculated from actual trades and quotes inside a five-minute window. the resulting rate becomes the day's official benchmark — the price index funds are valued at, the rate corporate treasuries book conversions at, the number custodians settle against.
that agreement is the point: a benchmark is a coordination device. by all using one rate, institutions eliminate a thousand small disputes — and concentrate their orders into one window, which is why the minutes around 16:00 london often move violently and then snap back.
the fix also carries a scar. in 2013–14, regulators found dealers at major banks had shared client order information in chat rooms and traded ahead of the window; the fines ran into billions, several traders were dismissed or prosecuted, and the calculation window was widened from one minute to five to make the benchmark harder to push.
month-end fixes are the loudest of all — rebalancing flows concentrate there. wednesday's brief covers who those flows belong to.
at 16:00 london time every business day, the WM/Reuters fix is calculated from actual trades and quotes inside a five-minute window. the resulting rate becomes the day's official benchmark — the price index funds are valued at, the rate corporate treasuries book conversions at, the number custodians settle against.
that agreement is the point: a benchmark is a coordination device. by all using one rate, institutions eliminate a thousand small disputes — and concentrate their orders into one window, which is why the minutes around 16:00 london often move violently and then snap back.
the fix also carries a scar. in 2013–14, regulators found dealers at major banks had shared client order information in chat rooms and traded ahead of the window; the fines ran into billions, several traders were dismissed or prosecuted, and the calculation window was widened from one minute to five to make the benchmark harder to push.
month-end fixes are the loudest of all — rebalancing flows concentrate there. wednesday's brief covers who those flows belong to.